JP2022136057A - Method of generating prediction model for predicting crop production performance, generation apparatus, and generation program - Google Patents

Abstract

To implement a technology for predicting production performance of crops.SOLUTION: An apparatus (20) for generating a prediction model for predicting production performance of crops includes a prediction model generation unit (23) which generates a prediction model (24) which predicts production performance of the year from at least a part of actual measurement values of production performance of previous years, the prediction model (24) referring to actual measurement value data of past annual production performance and representing interannual change of production performance.SELECTED DRAWING: Figure 1

Description

There is an application for exception to loss of novelty

The present invention relates to a prediction model generation method, generation device, and generation program for predicting the production performance of agricultural products.

At the cultivation sites of various crops, attempts have been made to predict the ingredient content of crops during the growing and harvesting periods using weather data for use in cultivation management, shipping plans, sales negotiations, and the like.

Patent Literature 1 describes a technique for predicting solar radiation information and temperature information for setting the sugar content of crops to a target sugar content, and controlling the environment in a greenhouse based on the predicted solar radiation information and temperature information. In the technique described in Patent Document 1, the difference between the standard sugar content specific to crops and the target sugar content is used for sugar content prediction in association with the difference between the average values of solar radiation and temperature for four weeks before harvest and the standard values. ing.

Non-Patent Document 1 describes that a trained model was constructed using time-series data of climate change and time-series data of mandarin orange quality over the past several years. In the technique described in Non-Patent Document 1, the constructed model is used to predict the quality of mandarin oranges at the harvest time of the current year from time-series data of the weather and the quality of mandarin oranges until the harvest time of the current year.

JP 2020-48551 A

Morimoto et al., Plant Environmental Engineering 17(2):90-98, 2005

There are three problems in predicting the component content of agricultural products. The first point is that the number of data measurement points is small. It is rare to frequently measure the ingredient content of agricultural products during the growth process and accumulate data, and in many cases, the ingredient content is measured only at the time of harvest or shipping. The second point is that the component content is also affected by factors other than weather. For example, in perennial crops, fluctuations over time may affect the component content, and in single-year crops, the aging trend of cultivation management that reflects consumer preferences may affect the component content. There is The third point is that in many cases the type and timing of weather data variables that affect component content are unknown.

Therefore, if a prediction model that reflects changes in the crop production environment over time can be realized using past measured values of component contents and meteorological data, it will greatly contribute to cultivation sites.

It is difficult to apply the technique described in Patent Document 1 to sugar content prediction when the standard sugar content specific to crops is unknown. In addition, with the technique described in Patent Document 1, when the harvest date is undecided, the starting point of the solar radiation and temperature data used for prediction is not determined, so it is difficult to apply it to sugar content prediction.

In the technique described in Non-Patent Document 1, it is necessary to measure the component content multiple times during the growth process in order to create time-series data on the quality of mandarin oranges. In addition, the model described in Non-Patent Document 1 can be used for quality prediction for the next season on the cultivation calendar, but it is difficult to predict quality earlier.

One aspect of the present invention has been made to solve the above-described problems, and an object thereof is to realize a technique for predicting the production results of agricultural crops.

A method of generating a prediction model according to one aspect of the present invention is a method of generating a prediction model for predicting the production performance of agricultural products, wherein the actual measurement data of the production performance for each past year is referenced, and the production performance and generating a prediction model that predicts the production performance of the current year from at least a portion of the measured production performance of previous years.

A prediction model generation device according to an aspect of the present invention is a prediction model generation device for predicting the production results of agricultural products, wherein the actual measurement data of the production results for each past year is referred to, and The forecasting model representing the secular change of is provided with a generation unit that generates a forecasting model for predicting the production performance of the current year from at least a part of the measured values of the production performance up to the previous year.

The predictive model generation device according to each aspect of the present invention may be implemented by a computer. A control program for a generating device realized by a computer and a computer-readable recording medium recording it are also included in the scope of the present invention.

ADVANTAGE OF THE INVENTION According to one aspect of the present invention, it is possible to realize a technique for predicting the production results of agricultural products.

It is a block diagram showing an example of the important section composition of the prediction system concerning one mode of the present invention. FIG. 5 is a diagram showing an example of measured value data of production results for each year up to the previous year, which is used by the prediction device according to one aspect of the present invention; It is a figure explaining the secular change of the production result of agricultural products, and the influence which a weather has on a production result. FIG. 5 is a diagram for explaining the effects of weather data variables on production results, which are used in the weather data prediction model generation device according to one aspect of the present invention. FIG. 2 is a diagram illustrating the concept of predicting the production results of agricultural products using a prediction model, which is executed by the prediction system according to one aspect of the present invention; 4 is a flowchart showing an example of prediction processing executed by a prediction device according to one aspect of the present invention; 4 is a flowchart showing an example of a prediction model generation process executed by a prediction model generation device according to an aspect of the present invention; 4 is a flowchart showing an example of a prediction model generation process executed by a weather data prediction model generation device according to an aspect of the present invention. It is a figure which shows the verification result of the sugar content prediction accuracy of an Example. It is a figure which shows the verification result of the sugar content prediction accuracy of an Example. It is a figure which shows the verification result of the acidity prediction accuracy of an Example. It is a figure which shows the verification result of the acidity prediction accuracy of an Example. It is a figure which shows the verification result of the sugar content prediction accuracy of an Example. It is a figure which shows the verification result of the acidity prediction accuracy of an Example. FIG. 4 is a diagram showing the effects of rainfall, average temperature, and hours of sunshine on sugar content in Examples. It is a figure explaining the influence which a soil water content value has on production results. FIG. 11 is a block diagram showing an example of a main configuration of a prediction system according to another aspect of the present invention;

One aspect of the present invention provides a technique for predicting the production performance of agricultural crops. The production results of agricultural crops predicted by one embodiment of the present invention include yield of agricultural crops; sugar content, acidity, protein content, content of components such as molecular components and pigment amounts measured by mass spectrometry, etc.; color, shape, Morphology such as size and length; lodging rate, leaf/flower/stem/fruit growth index (stem length, leaf opening, flower and fruit coloring), number of flowers, number of fruits, leaves including but not limited to the number of That is, according to one aspect of the present invention, it is possible to predict at least one evaluation item for evaluating produced crops. Agricultural crops whose production results are predicted according to one embodiment of the present invention are not particularly limited, and perennial plants such as mandarin oranges, apples, peaches, grapes, loquats, plums, coffee, green tea, beets, angelica, licorice, peony, and rhubarb; Examples include annual plants such as barley, soybean, peanut, millet, tomato, spinach, eggplant, pumpkin, strawberry, perilla, cnidium, rhizome, potato, sweet potato, and ornamental flowers.

[Prediction system 100]
Based on FIG. 1, a prediction system 100 used for predicting the production performance of agricultural products will be described. The prediction system 100 includes a prediction device 10 , a prediction model generation device 20 , and a weather data prediction model generation device 30 . Moreover, the prediction system 100 further includes an input device 40 , a storage device 50 and an output device 60 . The prediction system 100 may include the prediction device 10, the prediction model generation device 20, and the prediction model generation device 30 as independent devices, or may be integrated in one device.

The input device 40 receives an input operation to the prediction system 100 by the user. As an example, the input device 40 receives input of data used by the prediction device 10 to predict the production results of agricultural products. The input device 40 also receives input of data used for generating a prediction model in the prediction model generation device 20 . Furthermore, the input device 40 accepts input of data used by the prediction model generation device 30 to generate a prediction model of weather data. The input device 40 may be, for example, a keyboard, mouse, touch sensor, or the like.

Storage device 50 stores programs and data used in prediction system 100 . The storage device 50 stores, for example, various data input via the input device 40 . The storage device 50 also stores, as an example, a prediction model, input information, and output information used by the prediction device 10 to predict the production results of agricultural products. Further, the storage device 50 stores, as an example, the learning data used to generate the prediction model and the generated prediction model in the prediction model generation device 20 . In addition, the storage device 50 stores, as an example, the learning data used to generate the prediction model and the generated prediction model in the prediction model generation device 30 . The storage device 50 may have a database on a cloud or a server that stores various data.

The output device 60 outputs the result predicted by the prediction device 10 . Moreover, the output device 60 may output cultivation management guidance information based on the result predicted by the prediction device 10 . The cultivation management guidance information includes, for example, mulching implementation time, information on the maintenance of drainage channels, information on the number and time of fruit thinning, information on fertilization, information on implementation of pruning, and the like.

The mode of output by the output device 60 is not particularly limited. The output device 60 may be, for example, a display device that displays the information as an image, a printing device that prints the information, or an alarm device that outputs the information as sound. Also, the output device 60 may be a display of a mobile device such as a smart phone that displays the results predicted by the prediction device 10, cultivation management guidance information based on the results, and the like.

(Prediction device 10)
The prediction device 10 is a prediction device that predicts the production results of agricultural products. The prediction device 10 uses a prediction model representing changes in production performance over time, which is generated with reference to actual measurement data of production performance for each past year. Predict production performance for the current year.

The prediction device 10 includes a control section 11 . The control unit 11 comprehensively controls each unit of the prediction device 10, and is realized by, for example, a processor and a memory. In this example, a processor accesses storage (not shown), loads a program (not shown) stored in the storage into memory, and executes a sequence of instructions contained in the program. Each unit of the control unit 11 is thus configured. The control unit 11 includes an input data acquisition unit 12 and a prediction unit 13 as the respective units.

The input data acquisition unit 12 acquires input data, which is input information for a prediction model for predicting production results. The input data acquiring unit 12 reads from the storage device 50 the measured value data including at least a part of the measured values of the production results up to the previous year based on the input signal indicating the prediction start instruction from the input device 40 . Also, the input data acquisition unit 12 may acquire measured value data input via the input device 40 . The input data acquisition unit 12 outputs the acquired input data to the prediction unit 13 .

Here, the measured values of the production results up to the previous year are intended to mean the measured values of the production results measured before the current year, which is the target year for predicting the production results. Actual production results may be measured at least once each year during the growing season, harvesting season, or shipment. The measured value data may include at least the measured values of the production results of the previous year only, but may also include the measured values of the production results of several years up to the previous year.

The prediction unit 13 uses a prediction model 24 representing changes in production results over time, which is generated with reference to actual measurement data of production results for each past year, to obtain at least part of the actual measurement values of production results up to the previous year. Predict the production performance for the current year from The prediction unit 13 reads the prediction model 24 from the storage device 50, inputs the measured value data sent from the input data acquisition unit 12 to the prediction model 24, and acquires the current year's production results output from the prediction model 24. . The prediction unit 13 outputs the acquired data representing the production results for the current year to the output device 60 as a prediction result.

The prediction model 24 used by the prediction unit 13 to predict production results may be a prediction model 24 generated by a prediction model generation device 20 described later. As an example, the prediction model 24 is generated by performing machine learning using measured value data as learning data, at least a part of the measured values of the production results up to the previous year is input information, and the production results of the current year are output. It is a predictive model that is information. That is, by using the predictive model 24, it is possible to predict the production results by taking into consideration the annual fluctuations in the production results of agricultural products (fluctuations in the production results) as changes over time.

As an example, it is known that perennial plants such as mandarin oranges tend to have a lower production performance in the year following a year of good production performance due to a decrease in tree vigor, such as a decrease in the stored nutrients of the tree. Perennial plants alternate between front years (years with good production results) and back years (years with poor production results). In this way, since the production results of agricultural products fluctuate on an annual basis, annual fluctuations are taken into consideration by using the prediction model 24 that predicts the production results of the current year from the measured values of the production results up to the previous year. Prediction of production performance is possible.

As another example, in the case of monoennial plants, the method of cultivation management may vary from year to year, and this variation in the method of cultivation management may lead to fluctuations in production results from year to year. Therefore, even for single-year plants, by using the prediction model 24 that predicts the production results of the current year from the measured values of the production results up to the previous year, it is possible to predict the production results that take into account the annual fluctuations caused by changes in cultivation management methods. Predictable.

Here, the measured value data used for predicting production results will be described with reference to FIG. FIG. 2 is a diagram showing an example of actual measurement value data of production results for each year up to the previous year, which is used by the prediction device 10. As shown in FIG. In the graph on the left side of FIG. 2, the X axis indicates the calendar, the Y axis indicates the sugar content, and the Z axis indicates the year. By extracting data on annual changes in sugar content on specific days from such data on annual changes in sugar content in each year, we can obtain yearly changes in sugar content on specific days, as shown in the graph on the right side of Fig. 2. A graph representing The graph on the left side of FIG. 2 was obtained from the results of measuring the sugar content multiple times in each year. FIG. 2 is shown for explaining the concept of the present invention and does not limit the present invention to the embodiment shown in FIG.

As an example, in FIG. 2, a graph (left graph) representing annual changes in sugar content for the past 15 years and a graph (right graph) representing changes in sugar content for the past 15 years on October 30 are shown. showing. In this way, the prediction device 10 can predict the production results in consideration of annual fluctuations by using the data representing the annual sugar content change on a specific day.

When actually measuring annual changes in production results, the measurement is repeated multiple times a year using a reference plant individual (one predetermined reference tree or the like). In other words, for sampling, leaves and fruits are thinned out several times a year from a reference plant individual. Loss of leaves and fruits due to sampling can change the translocation of photosynthetic products in individual plants, and injury stress caused by sampling can change the state of individual plants. Therefore, the production results are measured in a state different from normal growth, and the data may not represent the actual growth process. If the production results of agricultural products are predicted based on the data obtained in this way, there is a risk that the prediction accuracy will deteriorate. The prediction device 10 predicts the production result by taking into consideration the yearly change in the production result on a specific day rather than the yearly change in the production result, so it is possible to make a highly accurate prediction.

In addition, the prediction unit 13 is generated by referring to measurement data including measured value data of production results for each past year and observation value data of meteorological data for those years, and predicts changes in production results over time and meteorological data. A predictive model 24 representing the impact can be used to predict the current year's production performance. The prediction unit 13 reads out the prediction model 24 from the storage device 50, inputs the actually measured value data sent from the input data acquisition unit 12 and the forecast value of the weather data for the current year into the prediction model 24, and outputs from the prediction model 24. Get the production results for the current year. Forecast values of weather data can be obtained, for example, from a database published by an organization that provides weather information, such as the Meteorological Agency. The prediction device 10 may store the forecast value of the obtained weather data in the storage device 50 and read it from the storage device 50 when the prediction unit 13 predicts the production results.

The predictive model 24 is generated by performing machine learning using measured data including actual measured value data of production results for each past year and observed value data of weather data for those years as learning data. Such a prediction model 24 is a prediction model whose input information is at least a part of the measured values of the production results up to the previous year and the forecast values of the current year's weather data, and whose output information is the production results of the current year. . By using such a prediction model 24, it is possible to predict the production results in consideration of the fluctuations in the production results of agricultural products from year to year and the weather data that affects the production results. By additionally considering meteorological data closely related to agriculture, it is possible to make more accurate predictions than when predicting production results by considering only fluctuations in production results from year to year.

The effects of aging and weather on production performance are described with reference to FIG. FIG. 3 is a diagram for explaining changes over time in the production results of agricultural products and the effects of weather on the production results. In the graph of FIG. 3, the X-axis indicates the year and the Y-axis indicates the sugar content, and the graph is a graph showing changes in the sugar content for each year on a specific day. It is thought that not only changes in production results over time but also weather influences the change in sugar content from year to year. Therefore, as shown in the graph on the right side of Fig. 3, the change in sugar content (solid line data) for each year is expressed by the sugar content change (dashed line data) due to aging and the difference between the broken line data and the solid line data. It decomposes into sugar content changes (range of arrows) due to the influence of weather.

The sugar content indicated by the dashed line in FIG. 3, which represents the effect of aging, is a sugar content that is predicted in consideration of the above-mentioned fluctuations in production results from year to year, and is defined as a standard sugar content (standard production performance). Since the standard sugar content is predicted by considering only annual fluctuations, it may differ from the actual sugar content represented by the solid line in FIG. The difference between the actual sugar content and the reference sugar content predicted by considering annual fluctuations is considered to be the influence of the weather on the sugar content.

Since the prediction model 24 is a prediction model that predicts the current year's production results from the measured data including the measured value data of the production results for each past year and the observed value data of the meteorological data of those years, the annual fluctuation In addition, it is possible to predict production results taking into account the influence of weather. Therefore, in the graph on the right side of FIG. 3, the reference production performance represented by the dashed line is corrected in the range of the arrows representing the influence of weather, resulting in a more accurate production performance closer to the actual production performance represented by the solid line. Prediction of production performance is possible.

In addition, the forecast model 24 is a forecast model generated for each forecast target point, and includes at least a part of the actual measurement value of the production performance up to the previous year at the forecast target point and the forecast value of the current year's weather data at the forecast target point. Therefore, it can be a model that predicts the production performance of the current year at the prediction target location.

The effects of aging and weather on production performance can vary from region to region where crops are grown. Therefore, by predicting the production results using the prediction model 24 created for each prediction target point, which is the region where the crops whose production results are to be predicted, are grown, it is possible to predict the production results with higher accuracy.

According to the prediction device 10, it is possible to predict the production results of the current year based on at least part of the measured values of the production results up to the previous year. It is possible to predict the production performance of As a result, the year's production results can be predicted at the beginning of the cultivation period, and cultivation management such as fertilization, irrigation, and fruit thinning can be changed based on the prediction results, thereby improving the year's production results. In addition, it is possible to determine the optimum working time and harvesting time based on the predicted production results.

(Prediction model generation device 20)
The predictive model generating device 20 is a predictive model generating device that predicts the production results of agricultural products. The predictive model generation device 20 refers to actual measured value data of production results for each past year, and is a predictive model representing changes in production results over time. Generate a predictive model that predicts the production performance of

The predictive model generation device 20 includes a control section 21 . The control unit 21 controls each unit of the prediction model generation device 20 in an integrated manner, and is realized by, for example, a processor and a memory. In this example, a processor accesses storage (not shown), loads a program (not shown) stored in the storage into memory, and executes a sequence of instructions contained in the program. Thereby, each part of the control part 21 is comprised. The control unit 21 includes a learning data acquisition unit 22 and a prediction model generation unit (generation unit) 23 as the respective units.

The learning data acquisition unit 22 acquires learning data for learning the prediction model 24 . The learning data acquisition unit 22 reads out learning data from the storage device 50 based on an input signal representing an instruction to start learning from the input device 40 . Also, the learning data acquisition unit 22 may acquire learning data input via the input device 40 . The learning data acquisition unit 22 outputs the acquired learning data to the prediction model generation unit 23 .

The prediction model generation unit 23 refers to the measured value data of the production results for each past year to create a prediction model representing changes in the production results over time. generate a predictive model 24 that predicts the production performance of

The prediction model generation unit 23 generates the prediction model 24 by performing machine learning using the learning data acquired from the learning data acquisition unit 22 . Here, the learning data are measured value data. The prediction model generation unit 23 generates the prediction model 24 using, for example, known machine learning methods such as neural networks, decision trees, random forests, and support vector machines. The prediction model 24 generated by the prediction model generation device 20 is a prediction model that uses as input information at least a part of the measured values of the production results up to the previous year and outputs the current year's production results.

In addition, the predictive model generation unit 23 may generate, as an example, a statistical model representing the correlation between the production results for each year up to the previous year and the production results for that year. That is, the predictive models 24 may be statistical models such as linear regression models, sparse modeling, generalized linear models, state space models, hierarchical Bayesian models, time series models, clustering, and the like.

Since the prediction model 24 is a model that represents changes in production results over time by referring to actual measurement data of production results for each year in the past, production It can be used to predict grades. The prediction model 24 is a prediction model used for predicting production results in the prediction device 10 described above. The prediction model generator 23 may store the generated prediction model 24 in the storage device 50 .

In addition, the prediction model generating unit 23 refers to measurement data including measured value data of production results for each past year and observation value data of weather data for those years, and determines changes in production results over time and effects of weather. can generate a predictive model 24 that represents Such a prediction model 24 is a prediction model that predicts the current year's production performance based on at least part of the actual production performance values up to the previous year and the current year's weather data forecast values. Observed value data of meteorological data can be obtained, for example, from a database published by an organization that provides meteorological information, such as the Meteorological Agency. The predictive model generation device 20 may store the obtained observation values of the weather data in the storage device 50 and read them from the storage device 50 when the prediction model generation unit 23 generates the prediction model 24 .

By performing machine learning using measured data as learning data, the prediction model generation unit 23 uses as input information at least a part of the actual measurement values of the production results up to the previous year and the forecast values of the weather data for the current year. Generate a predictive model whose output is grades. In addition, the prediction model generation unit 23 generates, as an example, a statistical model representing the correlation between the production results for each year up to the previous year and the observation values of the weather data for that year, and the production results for that year. may be generated as

The prediction model 24 can be used for forecasting production results, taking into consideration weather data that affects production results, as well as annual fluctuations in the production results of crops. By considering meteorological data closely related to agriculture, it can be used for more accurate prediction than when predicting production results by considering only fluctuations in production results from year to year.

The prediction model generator 23 preferably generates the prediction model 24 for each prediction target point. The prediction model generation unit 23 refers to the measurement data including the actual measurement values of the production results of each past year at the prediction target point and the observed values of the weather data of those years at the prediction target point, and the prediction target point Generate a prediction model 24 for each.

The effects of aging and weather on production performance can vary from region to region where crops are grown. Therefore, by generating the prediction model 24 for each prediction target point, which is an area where the agricultural products whose production performance is to be predicted, is grown, it can be used for more accurate production performance prediction.

The prediction model generator 23 may regenerate the prediction model 24 based on the prediction result in order to improve the prediction accuracy. If the prediction model 24 is a learning model generated by machine learning, the prediction model 24 may be re-learned. As an example, the prediction model generation unit 23 makes a prediction so that the difference between the measured value of the production performance of a certain past year and the predicted value of the production performance of that year predicted using the prediction model 24 is minimized. Regenerate model 24 .

As an example, the prediction model generation unit 23 calculates, for each year, the difference between the measured production results for each year up to the previous year and the predicted production results for those years predicted using the prediction model 24. , regenerate the forecast model 24 such that the difference is minimized in all years. In addition, the prediction model generation unit 23 generates a value that minimizes the sum of the squares of the differences between the measured production results of a certain past year and the predicted production results of that year predicted using the prediction model 24. Predictive model 24 may be regenerated.

Furthermore, when regenerating the prediction model 24, the prediction model generation unit 23 calculates the difference between the measured value of the production performance of a certain past year and the predicted value of the production performance of that year predicted by the prediction model 24. , the forecast model may be regenerated using regression analysis that recurs on observations of weather data for the year.

As described with respect to the prediction device 10, with reference to FIG. 3, it is possible to decompose changes in production results from year to year into changes in production results due to aging and changes in production results due to weather effects. can. The predictive model generating unit 23 considers the balance between changes in production results due to the effects of aging and changes in production results due to the effects of weather, and calculates changes in production results from year to year as changes in production results due to the effects of changes over time. and changes in production performance due to weather effects.

As an example, the predictive model generation unit 23 uses regularization to calculate a standard production result, which is a predicted value of production results in consideration of aging. First, the standard production performance is calculated for each regularization coefficient by varying the strength of regularization. Then, the difference between the measured value of the production performance and the standard production performance for each regularization coefficient is obtained, and this difference is calculated as the influence of the weather on the production performance. Then, the meteorological influence is regressed using the observed meteorological data for that year. The prediction model generation unit 23 repeats such processing to determine the regularization coefficient that minimizes the difference between the predicted value of the production performance and the measured value, the standard production performance, and the regression coefficient of the weather data. to regenerate.

An example of regression analysis of the influence of weather on production performance using weather data, which is performed when the prediction model generation unit 23 regenerates the prediction model 24, is generated by the prediction model generation device 30 described later. It can be a regression analysis performed by the part 33 .

By regenerating the prediction model 24 in this way, it is possible to generate the prediction model 24 that is capable of predicting production results with higher accuracy.

According to the prediction model generating device 20, it is possible to generate a prediction model for predicting the production results of the current year based on at least part of the measured values of the production results up to the previous year. Alternatively, it can be used to predict the production performance on the shipping date. As a result, it is possible to predict the production results of the year at the beginning of the cultivation period, and to improve the production results of the year by changing cultivation management such as fertilization, irrigation, and fruit thinning based on the prediction results. be able to. It can also be used to determine the optimum working and harvesting times based on predicted production results.

(Prediction model generation device 30)
The prediction model generation device 30 for weather data is a prediction model generation device for predicting the production results of agricultural products. The prediction model generation device 30 is a prediction model representing the production performance and the influence of weather by referring to the observation value data of the weather data before the forecast target date, and is based on the forecast values of the weather data for the forecast target date. generates a prediction model 34 that predicts The predictive model generation device 30 adjusts variables of the weather data so that the difference between the measured value of the production performance of a certain past year and the predicted value of the production performance of that year predicted using the forecast model 34 is minimized. A combination is selected to generate a predictive model 34 .

The predictive model generation device 30 includes a control section 31 . The control unit 31 controls each unit of the prediction model generation device 30 in an integrated manner, and is realized by, for example, a processor and a memory. In this example, a processor accesses storage (not shown), loads a program (not shown) stored in the storage into memory, and executes a sequence of instructions contained in the program. Each unit of the control unit 31 is thus configured. The control unit 31 includes a learning data acquisition unit 32 and a prediction model generation unit 33 as the respective units.

The learning data acquisition unit 32 acquires learning data for learning the prediction model 34 . The learning data acquisition unit 32 reads out learning data from the storage device 50 based on an input signal representing a learning start instruction from the input device 40 . Also, the learning data acquisition unit 32 may acquire learning data input via the input device 40 . The learning data acquisition unit 32 outputs the acquired learning data to the prediction model generation unit 33 .

The prediction model generation unit 33 is a prediction model representing the production results and the effects of weather by referring to the observed values of the weather data before the prediction target day, and calculates the production results from the forecast values of the weather data for the prediction target day. Generate a prediction model 34 to predict.

The prediction model generating unit 33 generates a prediction model 34 that represents the production results and the effects of weather by referring to the learning data including the daily observed weather data and the actual measurement values of the production results for those days. The prediction model generation unit 33 generates the prediction model 34 by performing machine learning using the learning data acquired from the learning data acquisition unit 22 . The prediction model generation unit 33 generates the prediction model 34 using, for example, known machine learning methods such as neural networks, decision trees, random forests, and support vector machines.

In addition, the predictive model generation unit 33 may generate, as an example, a statistical model representing the correlation between the daily weather data and the production results for those days. That is, the predictive model 34 may be a statistical model such as a linear regression model, sparse modeling, generalized linear model, state-space model, hierarchical Bayesian model, time series model, clustering, or the like.

The predictive model generation device 30 adjusts variables of the weather data so that the difference between the measured value of the production performance of a certain past year and the predicted value of the production performance of that year predicted using the forecast model 34 is minimized. A combination is selected to generate a predictive model 34 . Meteorological data expected to affect crop production performance includes many variables such as precipitation, average temperature, maximum temperature, minimum temperature, global solar radiation, and sunshine hours. Variables in meteorological data include interdependent variables. For example, there are combinations of highly interdependent variables, such as lower temperatures on rainy days (relationship between precipitation and temperature). do. Therefore, by selecting the optimal combination of weather data variables for predicting production results, it is possible to generate a prediction model that can accurately predict production results.

The effects of combinations of weather data variables on production performance will now be described with reference to FIG. FIG. 4 is a diagram for explaining the effects of weather data variables on production results. In Fig. 4, the upper left graph shows the effect of precipitation on sugar content, the upper middle graph shows the effect of average temperature on sugar content, and the upper right graph shows the effect of sunshine hours on sugar content. represents. In these graphs, the X axis indicates dates, the Y axis indicates increases and decreases in sugar content, and the Z axis indicates points to be measured. That is, these graphs show the effects of weather data variables on production performance for each measurement point.

In FIG. 4, the lower graph is a graph obtained by superimposing only the graphs of a specific measurement target point in the upper three graphs. That is, this graph collectively shows the influence of each variable of the weather data on the sugar content. In the lower graph, the solid line indicates the effect of precipitation, the dotted line indicates the effect of average temperature, and the circle plot indicates the effect of sunshine hours.

As shown in the upper graph of FIG. 4, the degree to which the production results are affected by variables varies from place to place and day to day. As shown in the lower graph of FIG. 4, by superimposing the graphs of each variable, it is clear that the types and degrees of affected variables are different even at the same place and on the same day.

In this way, by selecting the optimal combination of variables for predicting production results based on the degree of influence of weather data variables on production results, it is possible to generate a prediction model that can predict production results with higher accuracy. can. In addition, by selecting the optimum combination of variables for each day or each place, it is possible to generate a prediction model that can predict production results with higher accuracy.

The prediction model generation unit 33 selects a combination of weather data variables by regression analysis based on each integrated value or average value for each variable of the weather data before the prediction target date. The prediction model generation unit 33 calculates an integrated value obtained by integrating each variable for a predetermined period or an average value within a predetermined period, performs regression analysis based on the calculated integrated value or average value, and performs production Select the combination of variables that minimizes the difference between the actual and predicted performance. As an example, the prediction model generation unit 33 selects an optimum combination of variables by performing regression analysis by machine learning that can analyze highly interdependent variables.

The prediction model generation unit 33 selects a combination of weather data variables by regression analysis based on the start date and end date for calculating the integrated value or average value for each variable, and the number of calculation days. The effects of weather on production performance can vary not only over years, but over short periods such as flowering and fruiting. Therefore, by performing a regression analysis based on the start date and end date of calculation for each variable, or the start date and end date of calculation for each variable, , the optimal combination of variables can be selected for each period. Similarly, by performing regression analysis based on the number of days to be integrated or the number of days to average for each variable (calculation number of days), it is possible to select an optimal combination of variables for a shorter period.

The prediction model generation unit 33 selects a combination of the weather data variables by regression analysis based on the integrated value or average value of the calculated days for each day from the start date to the end date for each variable of the weather data. In this way, by calculating the integrated value or average value of each variable for the number of calculation days on each day in the calculation period from the start date to the end date, the window for the number of calculation days is shifted by one day, and the calculation period A vector can be created representing the impact of each daily variable in the production performance. Then, by performing regression analysis using the created daily vector and selecting the combination of variables that minimizes the difference between the actual measurement value and the predicted value of production performance, the optimal combination of variables for forecasting production performance is determined. can be selected. In addition, by performing regression analysis using daily vectors, it is possible to select a combination of variables by considering the effects of the continuity of daily variables on production performance, such as the effects of intense heat that continues for consecutive days. can.

The prediction model generation unit 33 generates changes in production performance over time and weather forecasts by referring to measurement data including measured value data of production performance for each past year and observation value data of weather data for those years. A prediction model 34 for predicting production results may be generated from at least a part of actual measurement values of production results up to the previous year and forecast values of weather data for the day to be predicted. In addition, such a prediction model 34 is generated by performing machine learning using measurement data as learning data, and as a result, at least a part of the actual measurement value of the production performance up to the previous year and the weather data of the prediction target day. A prediction model 34 may be generated in which forecast values are input information and the production results for the current year are output information. That is, the predictive model generation unit 33 determines the weather conditions when the yearly changes in production performance are decomposed into the effects of aging and the effects of weather in the prediction model generation unit 23 of the prediction model generation device 20 described above. can be regressed to generate a predictive model 34 that represents changes in production performance over time and the effects of weather.

Moreover, it is preferable that the prediction model generation unit 33 generates the prediction model 34 for each prediction target point. The prediction model generation unit 33 refers to the measurement data including the actual measured values of the production results for each past year at the prediction target point and the observed values of the weather data for those years at the prediction target point, and the prediction target point Generate a prediction model 24 for each. By generating the prediction model 34 for each prediction target point, which is an area where the crops whose production performance is to be predicted, is grown, it can be used for more accurate prediction of production performance.

According to the prediction model generation device 30, it is possible to predict the production results of the current year from the forecast values of the weather data for the target day of prediction. It can also be used for prediction. As a result, it is possible to predict the production results of the year at the beginning of the cultivation period, and to improve the production results of the year by changing cultivation management such as fertilization, irrigation, and fruit thinning based on the prediction results. be able to. It can also be used to determine the optimum working and harvesting times based on predicted production results.

(Prediction of production results in prediction system 100)
The concept of the flow of production result prediction processing using the prediction model in the prediction system 100 will be described with reference to FIG. FIG. 5 is a diagram illustrating the concept of predicting the production performance of agricultural products using a prediction model, which is executed by the prediction system according to one aspect of the present invention. Note that the weather data variables shown in FIG. 5 are merely an example, and the weather data variables are not limited to these.

As shown in FIG. 5, a prediction model that has been trained using measured data including actual measured value data of production results for each past year and observed value data of weather data for those years is used. Into the prediction model, actual measurement values of (past) production results up to the previous year and forecast values of weather data for the current year are input. The forecast model divides the predicted value of production performance into the current year's baseline production performance and the current year's weather impact. The prediction model regresses the influence of the current year's weather on the combination of weather data variables and selects the optimum combination of variables. The forecast model regresses the current year's weather impact on the selected combination of variables. The prediction model outputs a predicted value of the current year's production performance based on the current year's standard production performance and the influence of the current year's weather.

(Flow of prediction processing)
A flow of prediction processing (prediction method) by the prediction device 10 will be described with reference to FIG. FIG. 6 is a flowchart showing an example of prediction processing executed by the prediction device according to one aspect of the present invention. As shown in FIG. 6, first, the input data acquiring unit 12 acquires the measured values of the (past) production results up to the previous year (step S1). Next, the input data acquisition unit 12 acquires forecast values of weather data for the current year (step S2). Then, the prediction unit 13 inputs the acquired measured values of the production results up to the previous year and the forecast values of the weather data for the current year to the prediction model 24, and acquires the output predicted values of the production results for the current year (step S3 , the process of predicting). The prediction unit outputs the obtained prediction value to the output device 60 as a prediction result (step S4), and ends the process.

(Flow of prediction model generation processing)
The flow of the prediction model generation process (prediction model generation method) by the prediction model generation device 20 will be described with reference to FIG. 7 . FIG. 7 is a flowchart showing an example of a prediction model generation process executed by the prediction model generation device according to one aspect of the present invention. As shown in FIG. 7, first, the learning data acquisition unit 22 acquires actual measurement values of production results for each past year (step S11). Next, the learning data acquisition unit 22 acquires observation values of weather data for those years (step S12). Then, the learning data acquisition unit 22 generates learning data (measurement data) by associating actual measurement values with past yearly production results and observation values of weather data for those years (step S13). Using the generated learning data, the prediction model generation unit 23 uses the actual measurement values of production results up to the previous year (past) and forecast values of weather data for the current year as input information, and the production results for the current year as output information. A certain prediction model 24 is generated (step S14, step of generating). The prediction model generator 23 stores the generated prediction model 24 in the storage device 50 .

(Flow of weather data prediction model generation processing)
The flow of the prediction model 34 generation process (prediction model generation method) by the prediction model generation device 30 will be described with reference to FIG. 8 . FIG. 8 is a flowchart showing an example of a prediction model generation process executed by the weather data prediction model generation device according to one aspect of the present invention. As shown in FIG. 8 , first, the prediction model generation unit 33 sets the starting date and ending date for calculating the integrated value or average value of each variable of the weather data, and the number of days to be integrated or averaged (calculated number of days). Set (step S21). Next, the predictive model generation unit 33 calculates the integrated value or average value of the number of calculated days in each day from the start date to the end date for each variable (step S22). Then, the prediction model generator 33 regresses the prediction model 34 using the calculated values, and obtains the error between the predicted value and the measured value (step S23). The prediction model generation unit 33 obtains the combination of variables that minimizes the error, the start date and end date of the integrated value or the average value, and the number of calculation days (step S24). The prediction model generation unit 33 generates the prediction model 34 based on the obtained combination of variables, the start date and end date of the integrated value or the average value, and the number of calculation days (step S25, step of generating). The prediction model generator 33 stores the generated prediction model 34 in the storage device 50 (step S26).

[Example of realization by software]
The functions of the prediction device 10, the prediction model generation device 20, and the prediction model generation device 30 (hereinafter referred to as "devices") are programs for causing a computer to function as these devices, and each control of these devices It can be realized by a program for causing a computer to function as a block (especially each unit included in the control unit 11, the control unit 21, and the control unit 31).

In this case, these devices comprise a computer having at least one control device (eg processor) and at least one storage device (eg memory) as hardware for executing the program. Each function described in the above embodiment is realized by executing the above program using the control device and the storage device.

The program may be recorded on one or more computer-readable recording media, not temporary. This recording medium may or may not be included in these devices. In the latter case, the program may be supplied to these devices via any transmission medium, wired or wireless.

Also, part or all of the functions of the above control blocks can be realized by logic circuits. For example, integrated circuits in which logic circuits functioning as the control blocks described above are formed are also included in the scope of the present invention. In addition, it is also possible to implement the functions of the control blocks described above by, for example, a quantum computer.

Further, each process described in the above embodiment may be executed by AI (Artificial Intelligence). In this case, the AI may operate on the control device, or may operate on another device (for example, an edge computer or a cloud server).

〔Example〕
For mandarin oranges, we predicted sugar content and acidity using a prediction model generated using past sugar content and acidity data at multiple prediction target points, and verified the accuracy. The accuracy of sugar content prediction and acidity prediction was verified by comparing the measured values in 2020 and the predicted values using the prediction model. The weather observation values in 2018 and 2019, and the weather forecast values as of March 22 in 2020 were used.

The RMSE (root mean square error) between the predicted values and the measured values was obtained at all the prediction target points and all the prediction target days for two years. When meteorological observations were used in sugar content prediction, the RMSE was 0.6561 (Fig. 9). The RMSE was 1.5267 when the weather forecast value was used in the sugar content prediction (Fig. 10). Next, in the acidity prediction, the RMSE was 0.2517 when meteorological observations were used (Fig. 11). In the acidity prediction, the RMSE was 0.2636 when the weather forecast values were used (Fig. 12).

In addition, based on the weather forecast values as of March 22, 2020, the sugar content and acidity in 2020 were predicted, and the accuracy was verified using the measured values up to November 9, 2020. . Weather forecast values were used to verify accuracy up to November 9, 2020. The RMSE in sugar content prediction was 0.7355 (Fig. 13). The RMSE for acidity prediction was 0.1615 (Figure 14).

Thus, it was shown that by using the prediction model, it is possible to make predictions with sufficient accuracy as of March to grasp trends in production results for the next fiscal year.

FIG. 15 shows the seasonal effects of the three weather variables in 2020 high sugar content system fruit sorting prediction at a certain prediction target point. In FIG. 15, the solid line indicates the effect of precipitation, the dotted line indicates the effect of average temperature, and the circle plot indicates the effect of sunshine hours. In addition, since the weather forecast value as of March 22, 2020 is used, it differs from the actual weather effects in 2020. When the weather is similar to that of a normal year after late April, which is the normal value, rainfall has the greatest effect on sugar content among rainfall, average temperature, and hours of sunshine. As shown in FIG. 15, the amount of rainfall until the end of June 2020 had the effect of adjusting the sugar content at the time of fruit sorting upward by up to 0.01 on a daily basis. On the other hand, the rainfall in July and August 2020 had the effect of lowering the sugar content at the time of fruit sorting from 0.01 to a maximum of 0.025 on a daily basis. Multi-sheet covering was applied at this time, but it was suggested that the influence of precipitation in July and August was not zero.

[Other embodiments]
(Aspect using the average value of production results)
The prediction device 10 uses a prediction model representing the average value of the production results generated with reference to the actual measurement data of the production results for each past year, based on at least a part of the actual measurement values of the production results up to the previous year. The production performance for the current year may be predicted. As an example, if the number of years of accumulated production results for each past year is as small as 2 to 5 years, a prediction model that expresses the average value of production results is used instead of a prediction model that expresses changes in production results over time. good too. Here, the average value of production results may be a value obtained by averaging production results for each year in a predetermined number of years.
In addition, the prediction model 24 generated by the prediction model generation device 20 and the prediction model 34 generated by the prediction model generation device 30 are based on actual measured value data of production results for each past year and observed value data of weather data for those years. It may be a prediction model representing the average value of production performance and the influence of weather, generated with reference to measurement data including: Such prediction model 24 and prediction model 34 are capable of predicting production results by taking into account annual variations in crops as average values.

That is, in each of the above-described embodiments, the technical scope of the present invention also includes an embodiment in which all the explanations regarding the aging of the production results are replaced with average values of the production results.

(Aspect using measured value data of soil components as measured data)
The prediction device 10 refers to measured data including actual measured value data of production results for each past year and at least one of observed value data of meteorological data of those years and measured value data of soil components of those years. A forecasting model of production performance over time or averages and the effects of weather may be used to forecast production performance for the current year. Here, the measured value data of soil components are, for example, soil component values (soil moisture value, soil pH value, etc.) measured by a soil sensor such as a soil moisture meter provided in a field where crops are grown. could be.

The prediction model 24 generated by the prediction model generation device 20 is measured data including measured value data of production results for each past year, and at least one of observed value data of meteorological data and measured data of soil components for those years. may be a predictive model representing changes in production performance over time or averages and effects of weather and/or soil composition, generated with reference to. Such a prediction model 24 is generated by performing machine learning using measurement data as learning data, and includes at least a part of the actual measurement values of the production results up to the previous year, the forecast values of the current year's weather data, and the current year's soil At least one of the component measurement values is the input information, and the production result for the current year is the output value. According to the prediction model 24, it is possible to predict the production performance in consideration of soil components.

The forecast model 24 is a forecast model generated for each forecast target point, and includes at least a part of the actual measurement values of the production performance up to the previous year at the forecast target point, the forecast value of the current year's weather data at the forecast target point, and the current year At least one of the measured values of the soil components in the prediction target location may be used to predict the production results for the current year.
The prediction model 24 calculates the difference between the actual measured value of the production performance of a certain past year and the predicted value of the production performance of that year predicted by the prediction model 24 by calculating the difference between the observed value of the weather data of that year and the soil of that year. It may be regenerated using a regression analysis that recurs on at least one of the component measurements.

The prediction model 34 generated by the prediction model generation device 30 is generated by referring to at least one of the observed value data of the weather data before the target date of prediction and the measured value data of the soil components, and the production results and the weather and soil components. The prediction model may be a prediction model that expresses at least one of the effects and predicts the production performance from at least one of the predicted value of the weather data for the prediction target day and the measured value of the soil component.

Such a prediction model 34 uses variables of meteorological data such that the difference between the actual production performance of a certain past year and the predicted production performance of that year predicted using the prediction model 34 is minimized. and a combination of soil component variables. A combination of these variables may be selected by regression analysis based on the integrated value or average value of each variable for each variable of weather data and soil components before the prediction target date. Also, the combination of these variables may be selected by regression analysis based on the start date and end date for calculating the integrated value or average value, and the number of days for calculation. Furthermore, the combination of these variables may be selected by regression analysis based on the integrated values or average values for the number of days calculated for each day from the start date to the end date for each variable of weather data and soil components.

In addition, the prediction model 34 stores measurement data including at least one of actual measurement data of production results for each past year, observation data of meteorological data of those years, and measurement data of soil components of those years. A prediction model generated by reference, representing changes in production results over time or average values and the influence of at least one of weather and soil components, at least part of the actual production results up to the previous year and the target date for prediction A prediction model for predicting production results from at least one of forecast values of weather data and measured values of soil components. Such a prediction model 34 can be generated by performing machine learning using measured data as learning data. Such a prediction model 34 uses as input information at least a part of the actual measured values of the production results up to the previous year and at least one of the predicted values of the weather data for the prediction target day and the measured values of the soil components, and the production results of the current year. is the prediction model whose output information is

The effects of soil components on production performance will be described with reference to FIG. FIG. 16 is a diagram for explaining the influence of soil moisture value on production results. In the graph shown in FIG. 16, the X-axis indicates the date, and the Y-axis indicates the effect value of the soil water content converted into sugar content. The graph shown in FIG. 16 shows the influence of the soil moisture value on the sugar content at harvest.

As shown in FIG. 16, the extent to which the soil moisture value affects the sugar content at harvest differs from day to day. As an example, according to FIG. 16, it can be seen that the soil moisture value from July to September has a great effect on the sugar content at harvest. This finding agrees with the existing knowledge in the field of fruit trees that the sugar content increases due to drought stress in July and August, and the effect saturates thereafter.

In this way, by predicting the production results in consideration of the soil components, it is possible to predict the production results with higher accuracy. In addition, by predicting production generators in consideration of both weather and soil components, further improvement in prediction accuracy can be expected.

That is, in each embodiment described in this specification, all the explanations about the weather data are read as (i) the measured value data of the soil components, or (ii) the measured value data of the weather data and the soil components. The form is also included in the technical scope of the present invention.

(Mode for selecting variables based on information content criteria)
The combination of weather data variables and the combination of soil component variables in the prediction model 34 generated by the prediction model generation device 30 may be selected based on information amount criteria. That is, in the prediction model generation method, a combination of weather data variables and a combination of soil component variables are selected based on the information amount criteria to generate the prediction model 34, which is also within the technical scope of the present invention. included. Here, as the information criterion, a conventionally known information criterion can be used, and examples thereof include the Kullback-Leibler information criterion, the Akaike information criterion, the Bayesian information criterion, and the like.

As an example, if the model selection criteria are not included in the estimation algorithm of each prediction model generated in combination of weather data variables and soil component variables, meteorological data variables and soil components based on information criteria A combination of variables may be selected. As a result, the operator's knowledge, skill level, etc. do not affect the variable selection process, and the variable selection process can be uniformly executed.

(Aspect of Predicting Growth Stage)
The prediction device 10 may predict the growth stage of the crop by comparing the predicted production performance with a reference value representing the performance of each growth stage of the crop. Here, the growing stage of crops means the blooming period, the harvesting period, the transplanting period, and the like of the crops. Prediction of the growth stage in prediction device 10 can be performed in prediction section 13 . As an example, the prediction unit 13 of the prediction device 10 predicts the harvest when the predicted value of the production performance such as the sugar content and acidity obtained by the prediction model 24 exceeds a preset reference value (threshold value). Assuming that the criteria are met, the time when the relevant production performance is predicted is predicted as the harvest time.

The prediction device 10 may also calculate the influence amount of at least one of weather and soil components in the prediction model, and refer to the calculated influence amount to predict the growth stage of the crop.

Agricultural crops include crops that are grown indoors and then transplanted to the outdoors, crops that are stored in the shade with a roof after harvesting, and the like. With respect to such crops, the predictive model generation unit 23 and the predictive model generation unit 33 calculate the amount of influence (influence) of weather or soil components on production performance, and detect the date when the amount of influence discretely changes. be able to. For example, for crops that are stored in covered shade after harvest, the date on which the amount of sunshine duration abruptly drops to near zero corresponds to the date of transfer to the covered storage environment after harvest. Therefore, the prediction device 10 can predict the growth stage of crops by referring to such an influence amount.

(Prediction device for predicting growth stage)
In addition, data representing the transition of the actual growth stage of agricultural products for each past year, and measurement data including at least one of observation value data of weather data of those years and measured value data of soil components of those years At least part of the actual growth stage transition up to the previous year and the current year's weather data The technical scope of the present invention also includes a growth stage prediction device that predicts the transition of the growth stage of the current year from at least one of the predicted values and the measured values of the soil components of the current year. Here, the growth stage transition is intended to be, for example, information regarding the date of switching of each growth stage of crops, such as flowering date, harvest date, transplant date, and the like.

That is, the prediction model used by such a growth stage prediction device is a growth stage prediction model that expresses changes over time or the average value of changes in the growth stages of agricultural crops and the influence of at least one of weather and soil components. The growth stage prediction device inputs at least a part of the actual growth stage transition up to the previous year and at least one of the current year's weather data forecast values and current year's measured soil component values into the growth stage prediction model. to acquire the transition of the output growth stage. As an example, the growth stage prediction device uses a growth stage prediction model to obtain prediction results such as the date of flowering, the number of days from a predetermined date to the flowering date, and the number of days from the sowing date to the flowering date. can do.

In addition, in the embodiment described in this specification, the technical scope of the present invention also includes an embodiment in which all the explanations related to the prediction of the production performance are replaced with the prediction of the transition of the growth stage.

(Method of calculating potential value of production results)
Another form of prediction device will be described with reference to FIG. For convenience of explanation, members having the same functions as the members explained in the prediction system 100 and the prediction device 10 described above are denoted by the same reference numerals, and the explanation thereof will not be repeated. FIG. 17 is a block diagram showing the main configuration of a prediction system 200 according to another embodiment of the invention. The prediction system 200 differs from the prediction system 100 shown in FIG. 1 in that the control unit 211 of the prediction device 210 includes an input value generation unit 212, an output value acquisition unit 213, a setting unit 214, and a calculation unit 215. .

The input value generation unit 212 generates a simulation input value for at least one of the observed meteorological data and the measured value of the soil component based on at least one of the observed value of the past meteorological data and the measured value of the soil component (input value step of generating). The input value generation unit 212 generates a simulation input value by replacing data in at least one of the observed values of the weather data and the measured values of the soil components acquired by the input data acquisition unit 12 on a daily basis. The replacement of the data is performed so as to replace the data on the same day of another past year in the same location. In other words, the simulation input values generated by the input value generation unit 212 can be at least one of observed values of weather data and measured values of soil components obtained by patching data for a plurality of years on a daily basis. The input value generation unit 212 outputs the generated simulation input values to the output value acquisition unit 213 .

The output value acquiring unit 213 acquires the simulation output value of the production performance predicted from the simulation input value using the prediction model 24 or the prediction model 34 (step of acquiring the output value). The output value acquiring unit 213 inputs a simulation input value to the prediction model 24 or the prediction model 34, and acquires the output predicted value of production performance as a simulation output value. The output value acquisition unit 213 outputs the acquired simulation output value to the setting unit 214 .

The setting unit 214 sets the simulation input value when the obtained simulation output value is the highest as the potential value (setting step). The setting unit 214 compares the simulation output values obtained for various simulation input values, and sets the simulation input value when the simulation output value is the highest as the potential value. That is, the setting unit 214 sets the potential value by comparing production performance predicted using a plurality of pieces of meteorological data or soil component data with different patches. The setting unit 214 outputs the set potential value to the calculation unit 215 .

The calculation unit 215 calculates the simulation input value when the potential value is obtained as the optimum environment value (step of calculating). That is, the calculation unit 215 calculates the weather data or the soil component data when the best production results are obtained, and calculates at least one of the weather conditions and the soil component conditions for obtaining the best production results as the optimum environmental value. do.

In the prediction device 210, for at least one of the observed meteorological data and the measured value of the soil component, the simulation input value that is replaced with data on the same day in another past year at the same place is set as the virtual environment condition. The potential value is the production performance predicted by Such simulation input values are, for example, replacing "maximum temperature, sunshine hours and precipitation on March 3, 2010" with "maximum temperature, sunshine hours and precipitation on March 3, 2011". Generated by running the treatment on all dates of the growing season. In this way, by comparing the production results for each virtual environment condition, it is possible to predict the most ideal environment condition.

Note that the input value generating unit 212 determines which weather conditions or soil component conditions at what time based on the information on the influence of the weather or soil components on the production performance obtained from the prediction model generating unit 23 and the predictive model generating unit 33. You can decide to replace it. The input value generation unit 212 may also generate simulation input values using an algorithm that determines data to be replaced based on information about the influence of weather or soil components on production performance.

The present invention is not limited to the above-described embodiments, but can be modified in various ways within the scope of the claims, and can be obtained by appropriately combining technical means disclosed in different embodiments. is also included in the technical scope of the present invention.

10, 210 prediction device 13 prediction unit 20 prediction model generation device 23 prediction model generation unit (generation unit)
30 prediction model generation device 33 prediction model generation unit 100, 200 prediction system

Claims (10)

A method for generating a prediction model for predicting the production performance of agricultural products, comprising:
A prediction model that represents changes over time or an average value of the production performance by referring to the actual measurement data of the production performance for each past year, wherein at least a part of the production performance of the previous year is converted to the current year A method of generating, comprising generating a predictive model that predicts the production performance. In the generating step, by performing machine learning using the measured value data as learning data, at least part of the measured values of the production results up to the previous year are input information, and the production results of the current year are output information. 2. The generating method of claim 1, wherein a predictive model is generated. In the generating step, measurement data including at least one of actual measurement data of production results for each past year, observation data of weather data for those years, and measurement data of soil components for those years. Reference, a prediction model representing changes over time or an average value of the production performance and the influence of at least one of weather and soil components, wherein at least a portion of the production performance measured up to the previous year and weather data for the current year 2. The generation method according to claim 1, wherein the prediction model for predicting the production performance for the current year is generated from at least one of the predicted value of , and the measured value of the soil component for the current year. In the generating step, by performing machine learning using the measured data as learning data, at least a part of the measured production results up to the previous year, forecast values of weather data for the current year, and measured values of soil components for the current year 4. The generating method according to claim 3, wherein at least one of is input information, and the production result for the current year is output information. In the generating step, actual measurement data of the production results for each past year at the target point for prediction, observed value data of meteorological data for those years at the target point for prediction, and measured values of soil components for those years 5. The generation method according to claim 3 or 4, wherein a prediction model for each prediction target point is generated with reference to measurement data including at least one of a step of regenerating the prediction model so that the difference between the measured value of the production performance for a certain past year and the predicted value of the production performance for that year predicted using the prediction model is minimized; 6. The production method of any one of claims 3-5, further comprising. In the regenerating step, the difference between the actual production performance of a certain past year and the predicted production performance of that year predicted by the prediction model is calculated as the observed meteorological data of that year. 7. The generation method of claim 6, wherein the predictive model is regenerated using a regression analysis that regresses based on at least one of measurements of soil constituents for that year. A prediction model generation device for predicting the production performance of agricultural products,
A prediction model that represents changes over time or an average value of the production performance by referring to the actual measurement data of the production performance for each past year, wherein at least a part of the production performance of the previous year is converted to the current year A generation device comprising a generation unit that generates a prediction model for predicting the production performance. A prediction model generation program for predicting the production performance of agricultural products,
A prediction model that represents changes over time or an average value of the production performance by referring to the actual measurement data of the production performance for each past year, wherein at least a part of the production performance of the previous year is converted to the current year A generating program comprising generating a predictive model that predicts the production performance. A method for generating a prediction model for predicting the growth stage of agricultural crops, comprising:
Refers to measurement data that includes data representing changes in the actual growth stage of crops for each year in the past, and at least one of observed meteorological data for those years and measured value data for soil components for those years. A prediction model representing the transition of the growth stage and the influence of at least one of weather and soil components, wherein at least a part of the actual transition of the growth stage up to the previous year, the forecast value of the current year's weather data, and the current year generating a predictive model that predicts the transition of said growth stage for the current year from at least one of the measured values of the soil components of .

Images