• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    Procedure, system, computer system and computer program to generate data of a prediction of stockpiling of an agricultural product that has a life cycle. The procedure comprises: receiving historical data corresponding to parameters of the life cycle and to a stock that was produced under said parameters; generate learning data from historical data; perform a machine learning procedure from the learning data to adjust a "predictive" software; receive current data corresponding to life cycle parameters; and provide the current data to the predictive software to generate the corresponding forecast of collection. Historical data includes data from sensors of a farm that produced the product. The current data includes sensor data from a farm that is producing the product. The historical/current data may also correspond to parameters of the productive process, the producer, climatological, economic, phytopathological, logistic, market, etc. (Machine-translation by Google Translate, not legally binding)
    公开号:ES2681125A1
    申请号:ES201730306
    申请日:2017-03-08
    公开日:2018-09-11
    发明作者:Ricardo Arjona Antolín；Pedro CARRILLO DONAIRE；Gualberto ASENCIO CORTÉS；Miguel Ángel MOLINA CABANILLAS
    申请人:Easytosee Agtech Ltda Soc；Easytosee Agtech Sl；
    IPC主号:
    专利说明:

    DESCRIPTION
    Procedure, system, computer system and computer program to generate data from a collection forecast of an agricultural product
    The present invention relates to a method implemented in a computer system 5 for generating data of a prediction of collection of an agricultural product for a receiving entity of said product.
    The present invention also relates to a system, computer system, and computer program suitable for performing said collection prediction procedure.
    10
    STATE OF THE PREVIOUS TECHNIQUE
    There are various types of entities that receive significant quantities of one or more agricultural products as part of the life cycle of said products for the purpose of, for example, their distribution. Within these entities, for example, marketers, cooperatives, fish markets, meatballs, etc. can be distinguished.
    These entities normally receive a significant percentage of the overall volume of agricultural products, which is often unknown (by the entity's staff) until 20 shortly before receipt. These entities, therefore, manage large volumes of product whose storage, maintenance and, ultimately, distribution must be planned several weeks in advance.
    The reception of unexpected amounts can generate significant problems in terms of the provision of technical means necessary to deal with such collection. For example, in the face of imminent reception of agricultural products, it is appropriate to have properly provided adequate storage space, means of transport and necessary personnel, tools / products for the conservation of products in warehouses, etc. Some time in advance. Especially important in this scenario acquires an adequate conservation of agricultural products to avoid or minimize their deterioration.
    The stockpiling that a marketing company receives, for example, depends on the agricultural production of the producers that are associated or usually commercialize through said entity, whether for reasons of proximity, logistics, quality of service, etc. A reliable prediction of
    gathering can clearly imply the provision of a higher quality service, with more adequate logistics and, ultimately, with lower (and more competitive) costs.
    5 EXPLANATION OF THE INVENTION
    Therefore, there is a need for new procedures, systems, computer systems and computer program products to estimate a stockpile of agricultural products that allows to determine, in advance, an efficient configuration of technical resources to deal with the final stockpile.
    10
    In a first aspect, there is provided a procedure implemented in a computer system to generate data of a collection prediction of an agricultural product (which has a life cycle) for a receiving entity of said product.
    15 The procedure includes receiving historical data corresponding to parameters of the life cycle of the agricultural product and to a stockpile that occurred under said (historical) parameters. Historical data includes at least data from sensors installed in a farm where the agricultural product was produced.
    20 The procedure further comprises generating a set of learning data from the historical data received, and performing a machine learning procedure from the set of learning data in order to produce or adjust a computer module that implements a predictive model of agricultural product collection.
    25
    The procedure also includes receiving current data corresponding to parameters of the life cycle of the agricultural product, and providing the current data received to the computer module (or software) to generate the collection prediction according to the predictive model implemented by the software. The current data includes at least 30 data from sensors installed in an agricultural farm where the agricultural product is being produced.
    The exploitation that has originated the historical data and the exploitation from which the current data come can be the same farm or different farms with
    similar characteristics. Likewise, the sensors installed in one and another operation can be of similar types and can be arranged in equivalent locations.
    The proposed procedure is based on generating or adjusting software based on known (historical) data and results (actual collection) in order to generalize behaviors based on said information provided in the form of examples. The resulting software implements a predictive model based on, for example, a function that defines a correspondence or correlation between desired inputs (historical data) and outputs (actual collection) of the software.
    10
    The adjusted software (or trained with historical data and actual collections) is subsequently used to generate new results (estimated collection) from current data (compatible with the data used in learning) that have, a priori, an unknown influence on the Outcome. However, several experiments have revealed that said "predictive" software, trained with adequate volumes of historical data and corresponding actual collections, is capable of generating reasonably accurate collection estimates.
    In particular, the proposed procedure has shown acceptable reliability with data from sensors installed in situ, that is, in the field of cultivation or agricultural exploitation. More specifically, the procedure has been shown to be effective with measurement data from temperature, humidity, pressure, irrigation, fertilization, rain, wind, sunlight, etc. sensors. installed on site. That is, these environmental factors have revealed a good correlation or correspondence with agricultural production or final collection 25 and, consequently, their application to machine learning techniques (in the proposed procedure) has certainly proved effective.
    The resulting stockpiling estimates can be used to define, with sufficient time in advance, an efficient configuration of technical resources necessary to cope with the final collection. Said efficient configuration of technical resources may comprise, for example, adequate storage space, sufficient transportation within or between warehouses, necessary personnel, tools / products to handle / preserve the stored product or products, etc.
    5
    10
    fifteen
    twenty
    25
    30
    35
    The actual preparations used for the learning process may include data relating to, for example, the quantity of product collected, optimum time of collection based on, for example, an adequate degree of maturation, etc. The estimated collection can also therefore include data corresponding to said conditions, depending on the training received by the "predictive" software.
    The optimal time of collection can have an important impact on the proper preservation of the product after its collection and, in particular, during storage. An excessively mature agricultural product may require more complex care, while an optimum ripening point may facilitate product preservation. An accurate estimate of collection data that includes the optimal collection time may allow a better estimate of technical resources for product conservation and, therefore, their subsequent distribution under optimal conditions.
    In some implementations, the procedure may also include the determination of said estimate of technical resources to deal with the final collection. This resource estimate can be made through tables (for example: lookup tables) that relate total and / or partial historical collections with technical resources necessary to deal with such total and / or partial preparations. These tables may have been fed from data processed in previous executions of the procedure, together with the definition by an operator of the system of appropriate resource configurations for the resulting collections (total and / or partial). These predefined resource configurations may comprise resources to conserve the product, which may depend on at least one optimal estimated collection time included in the estimated collection data.
    In some examples, the procedure may further comprise selecting the machine learning procedure (or machine learning) from a set of predefined machine learning procedures. Such selection may depend on a predefined indicator representative of a degree of influence of the parameters / data received in the collection prediction. This predefined indicator can be obtained from a table (for example: lookup table) that defines relationships between, for example, the volume and type of data received with the technique or combination of machine learning techniques that have offered greater reliability or precision in previous executions of the procedure. This
    5
    10
    fifteen
    twenty
    25
    30
    35
    precision can be defined, for example, based on a deviation between the actual stockpile and the estimated stockpile produced in previous executions of the procedure.
    According to examples of the procedure, the set of machine learning procedures comprises procedures based on at least one of the following approaches: artificial neural networks, decision trees, association rules, genetic algorithms, support vector machines, clustering algorithms , Bayesian networks, or in a combination of all or some of these approaches (or techniques).
    In some implementations, the generation of the learning data set may comprise calculating, for each life cycle parameter, a representative statistical or average value of the received data corresponding to said parameter, and determining those received data that have a deviation from the calculated average / statistical value that exceeds a predefined deviation threshold. Once detected these outliers (or outliers) can be discarded or corrected in order to avoid unwanted distortions in the results. This principle can be applied to both historical and current data.
    According to examples of the procedure, the generation of the learning data set may comprise determining, for each life cycle parameter, data corresponding to said parameter that have not been received based on a predefined relationship with data that has been received. . For example, if temperature data is received under a ratio of 100 measurements generated every hour and, on the other hand, for a given time a number less than 100 measurements is received, it can be concluded that temperature data is missing. In this case, for example, a request can be made to the provider system of such data to return them in full. This criterion can be applied to both historical and current data.
    In some examples, the generation of the learning data may comprise determining derived data from received data, including said average derived data, variances, restrictions or indices calculated by means of predefined mathematical relationships or statistics or a combination of both. That is, different levels of aggregation, accumulation or summary of data received can be generated in order to streamline and ultimately improve the learning process. These different levels of aggregation, accumulation or summary can be multidimensional, in which each
    5
    10
    fifteen
    twenty
    25
    30
    35
    dimension can correspond to a certain typology (or parameter) of the processed data. The generation of the learning data set may comprise selecting derived data by means of heuristic or meta-heuristic search procedures or a combination of both.
    In some implementations, the historical / current data received may correspond to one or more of the following parameters of the agricultural product life cycle: parameters of an agricultural product production process, climatological parameters of an area where the farm is located agricultural, phytopathological parameters, parameters of the receiving entity of the agricultural product, parameters of a sector of the agricultural product, parameters of a producer or operator of the agricultural exploitation, parameters of suppliers of the agricultural sector, market parameters, etc. All these parameters have revealed a certain correlation or correspondence with the estimated / actual final collection.
    The data corresponding to the parameters of the agricultural product production process may comprise one or more of the following types of data: on type of seeds, fertilizers, anti-pest treatments, frequency of irrigation, phytosanitary treatments, planting date. It has been proven that these data also tend to have a notable influence on the final collection and, therefore, on the resulting collection. For example, more resistant seeds, more effective fertilizers, stricter anti-pest measures, etc. They can lead to more abundant crops.
    The data corresponding to the climatic parameters of the area in which the farm is located may comprise one or more of the following types of data: on precipitation occurred / expected, amount of precipitated water measured / expected, wind measured / expected, temperature measured / expected, measured / expected humidity. These data can come from a weather station in the area or even from a company dedicated to the provision of historical and / or current weather measurements and / or forecasts. The climatic parameters of the area have also been clearly influential in the estimated / final collection throughout the life cycle of the proposed procedure and software.
    The data corresponding to the phytopathological parameters may comprise one or more of the following types of data: on pests occurred / expected, involvement
    occurred / planned for the production of the agricultural product. These data can come from the producers themselves, agricultural laboratories or any entity that produces data on pests occurred / planned. It has also been corroborated that pest data can be very useful when forecasting crops and associated stockpiles. The existence or absence of a pest with negative effects in a crop field can especially condition the final collection and, therefore, the collection of agricultural products that are derived.
    The data corresponding to the parameters of the receiving entity of the agricultural product 10 (for example: trading company) may comprise one or more of the following types of
    data: upon receipt, prices, stock, qualities, costs. All these factors have
    revealed as habitually influential in the gathering that a certain entity may experience with that task. For example, if a marketer has more or less storage capacity, more or less associated costs, etc. the final collection is
    15 will be clearly affected by these circumstances. Various experiments with the
    procedure / software have shown that this data shows a clear correlation or correspondence with the resulting collection.
    The data corresponding to the parameters of a producer or operator of the agricultural holding 20 may comprise one or more of the following types of data: on specific production; past, present, expected returns; pruning; planting care, surface.
    The data corresponding to the parameters of agricultural sector suppliers may comprise one or more of the following types of data: on type, cost, quantity, effectiveness of fertilizers, pesticides, seeds. The data corresponding to the market parameters may comprise one or more of the following types of data: on historical / current product prices, product price forecasts.
    30 The data corresponding to the parameters of the sector related to the agricultural product may comprise one or more of the following types of data: on production in the area, country, continent; export; logistic costs; production of other areas, countries, continents; production of agricultural product substitutes.
    5
    10
    fifteen
    twenty
    25
    30
    35
    In a second aspect, a system is provided to generate data of a collection prediction of an agricultural product (which has a life cycle) for a receiving entity of said product.
    The system comprises means to receive historical data corresponding to parameters of the life cycle of the agricultural product and to a collection that occurred under said parameters, including said historical data at least data from sensors installed in an agricultural farm in which the product was produced. agricultural.
    The system also includes means for generating a set of learning data from the historical data received, and means for performing a machine learning procedure from the set of learning data in order to produce or adjust a computer module (or software) that implements a predictive model of agricultural product collection.
    The system also comprises means to receive current data corresponding to parameters of the life cycle of the agricultural product, including said current data at least data from sensors installed in an agricultural farm in which the agricultural product is being produced, and means to provide the current data received to the "predictive" software to generate the collection prediction according to the predictive model implemented by the "predictive" software.
    In a third aspect, a computer system is provided comprising a memory and a processor, in which the memory stores computer program instructions executable by the processor, these instructions comprising functionalities for executing any one of the procedures for predicting the collection of a agricultural product described above.
    In a fourth aspect, the invention provides a computer program comprising program instructions for causing a (computer) system to perform any one of the above procedures for predicting the collection of an agricultural product.
    Said computer program may be stored in physical storage media, such as recording media, a computer memory, or a read-only memory, or it may be carried by a carrier wave, such as electrical or optical.
    5
    10
    fifteen
    twenty
    25
    30
    35
    BRIEF DESCRIPTION OF THE DRAWINGS
    Particular examples of the present invention will now be described by way of non-limiting example, with reference to the accompanying drawings, in which:
    Figure 1 shows a schematic representation of a system for generating a collection prediction in a given application context, in accordance with examples of the invention;
    Figure 2 shows a flow chart of a method of generating a collection prediction, according to an example of the invention;
    Figure 3 shows a flow chart of a method of generating a collection prediction, according to another example of the invention; Y
    Figure 4 shows a flow chart of a block generating learning data that can be part of any of the procedures illustrated by Figures 2 and 3 or the like.
    DETAILED DESCRIPTION OF THE INVENTION
    In the following, numerous concrete details of the invention will be described in order to provide a complete understanding thereof. However, one skilled in the art should understand that the present invention can be practiced without some or all of these specific details. On the other hand, certain well known elements have not been described in detail so as not to unnecessarily complicate the description of the present invention.
    Figure 1 shows a schematic representation of a system for generating a collection prediction in a given application context, according to examples of the invention. As shown in the figure, a main computer system 100 for generating a collection prediction of an agricultural product can reside in the cloud 102 or can service through the cloud 102. This main computer system 100 can comprise a memory 101 and a processor (not shown), in which memory 101 stores computer program instructions executable by the processor,
    5
    10
    fifteen
    twenty
    25
    30
    35
    These instructions include functionalities for executing a collection prediction procedure such as, for example, those described with reference to other figures.
    Figure 1 also shows an agricultural holding 103 that can have various sensors 110-112 installed in corresponding crop field (s). These sensors 110-112 can be connected through suitable connections 113-115 with the main computer system 100. These connections 113-115 can be wireless or wired or of any kind that allows the transfer of measurements taken or data obtained by the sensors 110-112 to the main computer system 100. The aforementioned connections 113-115 may comprise a connection through a communications network such as, for example, the Internet. The sensors 110 - 112 can be, for example, temperature and / or humidity and / or pressure and / or irrigation and / or fertilization and / or rain and / or wind and / or sunlight (duration, intensity, etc.) sensors. ). Various experiments have revealed that these environmental parameters can have a great influence on the volume of agricultural product produced / collected and, consequently, on the stockpiling that a certain receiving entity of the product will have to face (trading, cooperative, fish market, alhondiga, etc. .).
    Alternatively, sensors 110-112 can provide the measurement data to an intermediate computer system (not shown) simply dedicated to, for example, the reception, storage and (pre) processing of data from the sensors. This intermediate computer system may be associated exclusively with the farm 103 or with various farms in, for example, the same area. The intermediate computer system can, therefore, offer data processing services and their subsequent transmission to the main system 100, either as historical data (for learning) or current data (for collection estimation).
    Figure 1 also shows that the main computer system 100 may be connected with respective systems 106, 108 associated with respective agricultural product receiving entities 104, 105, through respective connections 107, 109. These connections may be wireless or wired or of any type that allows an adequate exchange of data between the main system 100 and the systems 106, 108 of the receiving entities 104, 105. These connections 107, 109 may include a connection through a communications network such as, for example, Internet.
    5
    10
    fifteen
    twenty
    25
    30
    35
    According to the configuration shown in Figure 1, the main system 100 can also receive collection data from, for example, the systems 106, 108 of the receiving entities 104, 105 for later crossing with the historical data from the sensors This collection data can therefore be historical data of the collection that occurred under certain parameters of the life cycle of the agricultural product corresponding to at least the data coming from the sensors.
    The main system 100 can also be connected to other systems (not shown) dedicated to the production / transmission of data corresponding to other parameters of the life cycle of the agricultural product or products. Examples of such other parameters may be climatological, phytopathological parameters, of a production process, of an area in which the agricultural exploitation is located, of the receiving entity, of a sector of the product, of a producer or operator of the agricultural exploitation , of suppliers of the agricultural sector, market, etc. Each of these parameters may include various types of data with a certain influence on the final result, as discussed elsewhere in the description.
    In accordance with that described in relation to Figure 1, a network of computer systems that transmit historical and current data to the main system 100 can be provided, so that the latter performs prediction procedures for gathering (s) of agricultural product (s) (s) such as, for example, those proposed in other parts of the description with reference to other figures.
    Figure 2 shows a flow chart of a method of generating a collection prediction, according to an example of the invention. The procedure can be implemented, for example, in a computer system and in a technological context equal to or similar to that illustrated by the previous figure. The procedure can generate data from a collection prediction of an agricultural product (which has a life cycle) for a receiving entity of that product. Said collection prediction data may comprise an estimated quantity of product, an optimal estimated collection time, etc.
    In block 200, the procedure can be initiated as a result of the detection of a certain starting condition. This starting condition may comprise, for example, the receipt of a request for processing historical data (and associated collection) together with the data to be processed. These requests and associated data can be
    5
    10
    fifteen
    twenty
    25
    30
    35
    processed instantly or delayed. In the second case, the request and associated data can be stored in, for example, a waiting queue until the launch of the process in charge of its processing occurs as described below. This release can occur, for example, at a predefined start time of a massive processing of all received data, when a previously initiated ongoing execution has finished, etc.
    In block 201, the historical data corresponding to parameters of the life cycle of the agricultural product and to a collection (quantity, optimal time of collection, etc.) that occurred under said parameters can be received. These data may come from a transmitter system thereof, from the input queue mentioned above, etc. Said historical data may include data from sensors installed in an agricultural farm where the agricultural product was produced and, optionally, other data such as those indicated elsewhere in the description.
    In block 202, a set of learning data can be generated from the historical data received. This generation of learning data may comprise different processing as indicated in other parts of the description (see, for example, figure 4 and description thereof).
    In block 203, a machine learning procedure can be carried out from the set of learning data in order to produce or adjust a computer module (or software) that implements a predictive model of agricultural product collection. This predictive model may have evolved throughout various phases of learning. Specifically, blocks 201-203 may constitute a training thread that can be executed as many times as deemed necessary, based on different historical data sets and associated collection. Therefore, the "predictive" software can be subjected to in-depth training by repeating said thread as many times as deemed appropriate, in order to maximize the reliability of the procedure. The machine learning procedure may be based on one or more known approaches such as, for example, artificial neural networks, decision trees, association rules, genetic algorithms, support vector machines, clustering algorithms, Bayesian networks, or on a combination of all or some of these approaches (or techniques).
    5
    10
    fifteen
    twenty
    25
    30
    35
    In block 204, current data corresponding to parameters of the life cycle of the agricultural product can be received. Such current data may include data from sensors installed in an agricultural farm where the agricultural product is being produced and, optionally, other data such as those indicated elsewhere in the description. This exploitation and the one that originated the historical data (received in block 201) may or may not be the same agricultural exploitation. In the second case, farms may have similar characteristics, and the sensors may be equivalent and be similarly arranged on both farms. The historical and current data received may be of the same type and referred to the same concepts in order to maximize the consistency between the knowledge acquired by the predictive software and the current data processed by said knowledge to determine the corresponding collection estimate. For example, if current data is expected to include data on seeds used, it is desirable that the “predictive” software be trained with learning data that includes data on seeds used equivalent to the data used in training.
    In block 205, the current data received can be provided to the computer module (predictive software) to generate the data of the collection prediction according to the predictive model implemented by the computer module. Metrics can be defined by any known technique that allows validating the reliability of the results produced by the procedure. These metrics can be determined from data generated in previous executions of the procedure.
    In block 206, the procedure can be completed by providing the results of the process to a predefined recipient. The results (estimated collection for a given entity) can be provided, for example, by displaying them on the screen, saving them in a repository for later viewing in response to a user request, transmitting them to a computer system associated with the entity receiving the agricultural product, etc.
    Figure 2 shows a concrete example of the procedure for estimating the collection of a particular product for a particular entity. However, the expert will understand that said procedure can process data from various agricultural products to generate estimates of collection for several entities, in accordance with the principles set forth in reference to said figure.
    5
    10
    fifteen
    twenty
    25
    30
    35
    Figure 3 shows a flow chart of a method of generating a collection prediction, according to another example of the invention. The procedure illustrated by this figure may be significantly similar to that of the previous figure. Specifically, blocks 300-302 and 303-306 of this figure may be the same or similar to blocks 200-202 and 203-206 of Figure 2, respectively. One difference is that the procedure of Figure 3 comprises an additional block 307 for selecting the machine learning procedure according to, for example, the criteria described below.
    Such selection may depend on a representative indicator of the degree of influence on the final result of the historical and collection data received and processed by a particular learning algorithm. This (sub) selection process can be based on a data structure (for example, a search table or lookup table) that relates different combinations of received data and learning procedure used with specific values of said influence indicator. This data structure may have been constructed from used data and results obtained in previous executions of the procedure in which a particular learning procedure has been used. Each value attributed to the indicator may have been determined, for example, based on a deviation between the estimated collection and the actual collection corresponding to each combination of data received and learning algorithm used.
    Figure 4 shows a flow chart of a learning data generation block that can be part of any of the procedures illustrated by Figure 2 (block 202) and Figure 3 (block 302), or the like. Said generation of learning data may have as input historical data and associated collection 400 from a block similar to, for example, block 201 of Figure 2 or block 301 of Figure 3. The resulting learning data 405 may be provided to a block similar to, for example, block 203 of figure 2 or block 307 of figure 3.
    In block 401, it is possible to calculate, for each parameter of the life cycle received, an average or statistical value representative of the data received corresponding to said parameter, and determine those data received that have a deviation from the calculated average / statistical value that exceeds a predefined deviation threshold. A data that presents an excessive deviation can be considered as an outlier (or outlier) value.
    5
    10
    fifteen
    twenty
    25
    30
    35
    This identification of outliers can be carried out by applying any known method or technique of detecting outliers.
    In block 402, the atypical values detected can be processed in order to minimize their eventual negative influence on the results of estimation of collection (s). For example, all or some of the atypical values detected can be discarded in order to avoid distortions in the results, and / or they can be returned to their origin for correction and subsequent return to the system that performs the procedure of estimation of collection ( s).
    In block 403, for each life cycle parameter, data corresponding to said parameter that have not been received can be determined based on a predefined relationship with data that has been received. For example, if a certain measurement received follows a temporary pattern of generation by a sensor (for example: generation of 200 measurements every hour), and an irregularity is detected in the measurements from the sensor (for example: they have only been generated 50 measures at a specific time), it can be concluded that measures are missing. In this case, the absence of such measures can be ignored if it is considered that their influence on the result may be insignificant, or the data received may be sent to its source for correction and subsequent incorporation into the procedure, etc. This logic can be implemented, for example, through knowledge rules using any known artificial intelligence approach.
    In block 404, derived data can be determined from received data, including said average derived data, variances, restrictions or indices calculated by means of predefined mathematical relationships and / or statistics. This determination of derived data may include, for example, the performance of heuristic and / or meta-heuristic search procedures. Derived data may comprise, for example, different aggregations of data at different levels in order to streamline the learning process and maximize its effectiveness. Other derived data may be, for example, average values of temperature, humidity, wind, etc. measurements. if it is considered that some aspect of learning may not require concrete measures and the use of average values (or summary) may accelerate such learning.
    Figure 4 shows the different blocks 401 - 404 that make up the data generation
    5
    10
    fifteen
    twenty
    25
    30
    35
    of learning in a certain order. However, said blocks can be executed in a different order than the one shown in the figure. For example, in some implementations, the set of blocks 401 and 402 (outliers), block 403 (data not received) and block 404 (derived data) could be executed in parallel. Or, in other examples, block 403 (data not received) could be executed before block set 401 and 402 (outliers). Etc.
    Any of the collection estimation procedures (s) described may include a final block (not shown) dedicated to determining an estimate of technical resources to cope with the prediction of collection (s) generated. This estimate of technical resources may include the estimation of a storage space and / or means of transport and / or quantity of product for an adequate conservation of the agricultural product, etc. The estimation of resources can be carried out through a process of analysis of a multidimensional database, in which each dimension can correspond to a specific agricultural product, a value of collection of the product, a storage space, a quantification of the necessary transport, etc. . Each different combination of values corresponding to the dimensions considered may comprise, in the BD, a resource estimate value that may be a specific value, a range of values, a set of discrete values, etc. The multidimensional BD analysis can produce an estimate of necessary resources based on, for example, a correspondence between the collection estimate produced and a multidimensional combination existing in the BD together with the corresponding estimate value determined by said multidimensional combination.
    The different examples described herein refer to computer procedures, systems and programs based on artificial intelligence techniques for the prediction of stockpiles and, therefore, efficient commercial, purchasing, service, production and stock management . These procedures, systems and computer programs can, therefore, be very useful in providing services (in these areas) to entities receiving agricultural products such as, for example, meatballs, fish markets, cooperatives and agricultural traders.
    Although only a few particular examples of the invention have been described herein, the person skilled in the art will understand that other alternative examples and / or uses of the invention are possible, as well as obvious modifications and equivalent elements. Besides, the
    5
    10
    fifteen
    twenty
    25
    30
    35
    The present invention encompasses all possible combinations of the concrete examples that have been described. The numerical signs relating to the drawings and placed in parentheses in a claim are only intended to increase the understanding of the claim, and should not be construed as limiting the scope of the claim's protection. The scope of the present invention should not be limited to specific examples, but should be determined only by an appropriate reading of the appended claims.
    Although also the described examples of the invention with reference to the drawings comprise computer systems and procedures performed in computer systems, the invention also extends to computer programs, more particularly to computer programs in or on carrier media, adapted to put In practice the invention. The computer program may be in the form of source code, object code or an intermediate code between source code and object code, such as partially compiled form, or in any other form suitable for use in the implementation of the processes in accordance with the invention. The carrier medium can be any entity or device capable of carrying the program.
    For example, the carrier medium may comprise a storage medium, such as a ROM, for example a CD ROM or a semiconductor ROM, or a magnetic recording medium, for example a floppy disc or a hard disk. In addition, the carrier means may be a transmissible carrier medium such as an electrical or optical signal that can be transmitted via electrical or optical cable or by radio or other means.
    When the computer program is contained in a signal that can be transmitted directly by means of a cable or other device or medium, the carrier medium may be constituted by said cable or other device or medium.
    Alternatively, the carrier means may be an integrated circuit in which the computer program is encapsulated (embedded), said integrated circuit being adapted to perform or for use in performing the relevant procedures.
    On the other hand, the invention can also be implemented by computer systems, such as personal computers, servers, a computer network of computers, laptops, tablets or any other programmable device or
    computer processor In addition or alternatively, programmable electronic devices can also be used, such as programmable logic controllers (ASICs, FPGAs, programmable controllers, etc.).
    5 Accordingly, the invention can be implemented in both hardware and software or firmware, or any combination thereof.
    权利要求:
    Claims (31)
    [1]
    5
    10
    fifteen
    twenty
    25
    30
    35
    1. Procedure implemented in a computer system to generate data of a prediction of collection of an agricultural product for a receiving entity of said product, the agricultural product having a life cycle, and comprising the procedure:
    receive historical data corresponding to parameters of the life cycle of the agricultural product and to a collection that occurred under said parameters, including said historical data at least data from sensors installed in an agricultural farm in which the agricultural product was produced;
    generate a set of learning data from the historical data received; carry out a machine learning procedure based on the learning data set in order to produce or adjust a computer module that implements a predictive model of agricultural product collection;
    receive current data corresponding to parameters of the life cycle of the agricultural product, including said current data at least data from sensors installed in an agricultural holding in which the agricultural product is being produced;
    provide the current data received to the computer module to generate the data of the collection prediction according to the predictive model implemented by the computer module.
    [2]
    2. The method according to claim 1, which comprises selecting the automatic learning procedure from a set of predefined automatic learning procedures, said selection depending on a predefined indicator representative of a degree of influence of the parameters received in the prediction of the collection.
    [3]
    3. The method according to claim 2, wherein the set of machine learning procedures comprises procedures based on at least one of the following approaches: artificial neural networks, decision trees, association rules, genetic algorithms, support vector machines , clustering algorithms, Bayesian networks, or in a combination of all or some of them.
    [4]
    4. The method according to any one of claims 1 to 3, wherein generating the learning data set comprises calculating, for each life cycle parameter, a representative statistical or average value of the received data corresponding to
    twenty
    5
    10
    fifteen
    twenty
    25
    30
    35
    said parameter, and determine those data received that have a deviation from the calculated average / statistical value that exceeds a predefined deviation threshold.
    [5]
    5. The method according to claim 4, wherein generating the learning data set comprises discarding the received data that exceeds the predefined deviation threshold.
    [6]
    6. A method according to any one of claims 1 to 5, wherein generating the learning data set comprises determining, for each life cycle parameter, data corresponding to said parameter that have not been received based on a predefined relationship with data that has been received.
    [7]
    7. A method according to any one of claims 1 to 6, wherein generating the learning data set comprises determining derived data from received data, including said average derived data, variances, restrictions or indices calculated by means of mathematical relationships or predefined statistics, or a combination of both.
    [8]
    A method according to claim 7, wherein generating the learning data set comprises selecting derived data by means of heuristic or meta-heuristic search procedures, or a combination of both.
    [9]
    9. Method according to any one of claims 1 to 8, further comprising determining an estimate of technical resources to cope with the prediction of stockpiling generated.
    [10]
    A method according to claim 9, wherein determining the estimate of technical resources comprises estimating a storage space or means of transport or quantity of product for conservation of the agricultural product, or a combination of all or some of said resources.
    [11]
    11. The method according to any one of claims 1 to 10, wherein the data from sensors includes one or more of the following types of data: temperature, humidity, pressure, irrigation, fertilization, rain, wind, sunlight data .
    5
    10
    fifteen
    twenty
    25
    30
    35
    [12]
    12. A method according to any one of claims 1 to 11, wherein the historical / current data received comprises data corresponding to parameters of an agricultural product production process.
    [13]
    13. The method according to claim 12, wherein the data corresponding to parameters of the agricultural product production process comprise one or more of the following types of data: on type of seeds, fertilizers, anti-pest treatments, frequency of irrigation, treatments Phytosanitary, planting date.
    [14]
    14. The method according to any one of claims 1 to 13, wherein the data
    historical / current received include data corresponding to parameters
    of an area where the farm is located.
    [15]
    15. The method according to claim 14, wherein the data corresponding to weather parameters comprise one or more of the following types of data: on precipitation occurred / expected, amount of precipitated water measured / expected, wind measured / expected, temperature measured / expected, measured / expected humidity.
    [16]
    16. The method according to any one of claims 1 to 15, wherein the data
    historical / current received include data corresponding to parameters
    Phytopathological
    [17]
    17. The method according to claim 16, wherein the data corresponding to phytopathological parameters comprise one or more of the following types of data: on pests occurred / expected, damage occurred / expected to the production of the agricultural product.
    [18]
    18. A method according to any one of claims 1 to 17, wherein the historical / current data received comprises data corresponding to parameters of the recipient entity of the agricultural product.
    [19]
    19. The method according to claim 18, wherein the data corresponding to parameters of the receiving entity of the agricultural product comprise one or more of the following types of data: on receipt, prices, stock, qualities, costs.
    5
    10
    fifteen
    twenty
    25
    30
    35
    [20]
    20. A method according to any one of claims 1 to 19, wherein the historical / current data received comprises data corresponding to parameters of a sector related to the agricultural product.
    [21]
    21. The method according to claim 20, wherein the data corresponding to parameters of the sector comprise one or more of the following types of data: on production of the area, country, continent; export; logistic costs; production of other areas, countries, continents; production of agricultural product substitutes.
    [22]
    22. Method according to any one of claims 1 to 21, wherein the historical / current data received comprises data corresponding to parameters of a producer or operator of the farm.
    [23]
    23. The method according to claim 22, wherein the data corresponding to parameters of the producer or operator of the agricultural holding comprise one or more of the following types of data: on specific production; past, present, expected returns; pruning; planting care, surface.
    [24]
    24. A method according to any one of claims 1 to 23, wherein the historical / current data received comprises data corresponding to parameters of agricultural sector suppliers.
    [25]
    25. The method according to claim 24, wherein the data corresponding to parameters of agricultural sector suppliers comprise one or more of the following types of data: on type, cost, quantity, effectiveness of fertilizers, pesticides, seeds.
    [26]
    26. A method according to any one of claims 1 to 25, wherein the historical / current data received comprises data corresponding to market parameters.
    [27]
    27. The method of claim 26, wherein the data corresponding to market parameters comprise one or more of the following types of data: on historical / current product prices, product price forecasts.
    [28]
    28. A computer program comprising program instructions for having a system execute a method according to any one of claims 1 to 27.
    [29]
    29. A computer program according to claim 28, which is stored on recording media.
    30. A computer program according to claim 28, which is carried by a signal
    carrier
    [31]
    31. System for generating data of a prediction of collection of an agricultural product for a receiving entity of said product, the agricultural product having a life cycle, and
    I understand the system:
    means for receiving historical data corresponding to parameters of the life cycle of the agricultural product and to a stockpile that occurred under said parameters, including said historical data at least data from sensors installed in an agricultural farm in which the agricultural product was produced;
    15 means to generate a set of learning data from the data
    historical received;
    means for performing a machine learning procedure from the set of learning data in order to produce or adjust a computer module that implements a predictive model of agricultural product collection;
    2nd means to receive current data corresponding to the life cycle parameters of the
    agricultural product, including said current data at least data from sensors installed in an agricultural holding where the agricultural product is being produced;
    means to provide the current data received to the computer module so that
    25 generate the collection prediction data according to the predictive model implemented by the computer module.
    [32]
    32. A computer system comprising a memory and a processor, in which the memory stores computer program instructions executable by the processor,
    3o comprising these instructions functionalities for executing a method according to any one of claims 1 to 27.
    类似技术:
    公开号 | 公开日 | 专利标题
    Merlos et al.2015|Potential for crop production increase in Argentina through closure of existing yield gaps
    Lee et al.2013|Agricultural production system based on IoT
    AU2020289735A1|2021-01-21|Generating an agriculture prescription
    O’Shaughnessy et al.2015|Site-specific variable rate irrigation as a means to enhance water use efficiency
    US20180027725A1|2018-02-01|Precision agriculture system
    US10772269B2|2020-09-15|Predicting crop productivity via intervention planning on small-scale farms
    US10185790B2|2019-01-22|Modeling of crop growth for desired moisture content of targeted livestock feedstuff for determination of harvest windows using field-level diagnosis and forecasting of weather conditions and observations and user input of harvest condition states
    CN208207247U|2018-12-07|A kind of agrometeorological hazard early warning system
    BR112020004630A2|2020-09-24|generation of a yield map for an agricultural field using regression and classification methods
    US20160217231A1|2016-07-28|Modeling of crop growth for desired moisture content of bovine feedstuff and determination of harvest windows for high-moisture corn using field-level diagnosis and forecasting of weather conditions and observations and user input of harvest condition states
    WO2016118686A1|2016-07-28|Modeling of crop growth for desired moisture content of targeted livestock feedstuff for determination of harvest windows using field-level diagnosis and forecasting of weather conditions and observations and user input of harvest condition states
    ES2681125B1|2019-09-09|PROCEDURE, SYSTEM, INFORMATIC SYSTEM AND COMPUTER PROGRAM TO GENERATE DATA OF A PREPARATION FOR THE COUPLE OF AN AGRICULTURAL PRODUCT
    US10176280B2|2019-01-08|Modeling of crop growth for desired moisture content of bovine feedstuff and determination of harvest windows for corn silage using field-level diagnosis and forecasting of weather conditions and field observations
    Dahane et al.2020|An IoT Based Smart Farming System Using Machine Learning
    Stevens et al.2012|Understanding the role of spring grazing of lucerne in the Central Otago environment
    Bochtis et al.2010|Feasibility of a modelling suite for the optimised biomass harvest scheduling
    Jiang et al.2014|Estimating the local effect of weather on field crop production with unobserved producer behavior: a bioeconomic modeling framework
    Mani Sai Jyothi et al.2020|Utilization of the internet of things in agriculture: possibilities and challenges
    US20210209705A1|2021-07-08|System and Method for Managing and Operating an Agricultural-Origin-Product Manufacturing Supply Chain
    CN104778630A|2015-07-15|Image processing-based agricultural intelligent platform
    CN106355509A|2017-01-25|Agricultural intelligent management system
    TWI695339B|2020-06-01|System for analyzing and managing the yield and production period of the crop
    Bahrami et al.2020|Determination of Effective Weather Parameters on Rainfed Wheat Yield Using Backward Multiple Linear Regressions Based on Relative Importance Metrics
    Manfrini et al.2016|Development of precision apple fruit growing techniques: monitoring strategies for yield and high-quality fruit production
    Mushtay et al.2017|Determination Of Potential Opportunities For The Food Provision With Grain Production: A Ukrainian And European Viewpoint
    同族专利:
    公开号 | 公开日
    ES2681125B1|2019-09-09|
    WO2018162778A1|2018-09-13|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US5884225A|1997-02-06|1999-03-16|Cargill Incorporated|Predicting optimum harvest times of standing crops|
    WO2011123653A1|2010-03-31|2011-10-06|Earthtec Solutions Llc|Environmental monitoring|
    JP2013042668A|2011-08-22|2013-03-04|Fujitsu Ltd|Information processing device, harvest time-predicting program and harvest time-predicting method|
    US9031884B1|2015-01-23|2015-05-12|Iteris, Inc.|Modeling of plant wetness and seed moisture for determination of desiccant application to effect a desired harvest window using field-level diagnosis and forecasting of weather conditions and observations and user input of harvest condition states|
    WO2016118686A1|2015-01-23|2016-07-28|Iteris, Inc.|Modeling of crop growth for desired moisture content of targeted livestock feedstuff for determination of harvest windows using field-level diagnosis and forecasting of weather conditions and observations and user input of harvest condition states|
    WO2016118684A1|2015-01-23|2016-07-28|Iteris, Inc.|Harvest advisory modeling using field-level analysis of weather conditions and observations and user input of harvest condition states and tool for supporting management of farm operations in precision agriculture|
    US20170038749A1|2015-08-05|2017-02-09|Iteris, Inc.|Customized land surface modeling for irrigation decision support in a crop and agronomic advisory service in precision agriculture|
    US10515715B1|2019-06-25|2019-12-24|Colgate-Palmolive Company|Systems and methods for evaluating compositions|
    法律状态:
    2018-09-11| BA2A| Patent application published|Ref document number: 2681125 Country of ref document: ES Kind code of ref document: A1 Effective date: 20180911 |
    2019-09-09| FG2A| Definitive protection|Ref document number: 2681125 Country of ref document: ES Kind code of ref document: B1 Effective date: 20190909 |
    优先权:
    申请号 | 申请日 | 专利标题
    ES201730306A|ES2681125B1|2017-03-08|2017-03-08|PROCEDURE, SYSTEM, INFORMATIC SYSTEM AND COMPUTER PROGRAM TO GENERATE DATA OF A PREPARATION FOR THE COUPLE OF AN AGRICULTURAL PRODUCT|ES201730306A| ES2681125B1|2017-03-08|2017-03-08|PROCEDURE, SYSTEM, INFORMATIC SYSTEM AND COMPUTER PROGRAM TO GENERATE DATA OF A PREPARATION FOR THE COUPLE OF AN AGRICULTURAL PRODUCT|
    PCT/ES2018/070167| WO2018162778A1|2017-03-08|2018-03-07|Method, system, computer system and computer program for generating data regarding a prediction of the stock of an agricultural product|
    [返回顶部]