JP2020149652A - Learning farming system for providing optimal irrigation and fertilizer for farming - Google Patents

Abstract

To provide a learning farming system capable of objective evaluation not only relying on the rules of thumb of a farmer in the cultivation of agricultural products.SOLUTION: In order to perform optimal irrigation and fertilization for changing the state of the field environment, the learning farming system combines and structures measured values of the field environment and empirical rules in a form of a proposition, and incorporates the same into a system in a form that can be handled by a computer. When there is no rule of thumb, a feedback system is built for learning the results of cultivation based on similar crops and similar environments.SELECTED DRAWING: Figure 1

Description

In the present invention, farming is expressed by numerical values that can be handled by a computer, and the empirical rules of farmers are structured and quantified, and these and data based on measurement by an apparatus are combined and water application or fertilization is performed by a calculation procedure or the like. Learning-type feedback system construction method, system and product for farming to make decisions.

When looking at an agricultural field, agriculture is carried out so that the farmers select and grow their crops in order to maximize their profits. For example, in the selection of agricultural products, the economic value of the agricultural products is mainly emphasized, and the varieties are selected in consideration of the adjustment of the shipping time. On the other hand, the environment of the field has a great influence on the cultivation of agricultural products. For example, there are sunshine duration such as sunlight, air temperature, humidity and carbon dioxide concentration, soil temperature, humidity and fertilizer concentration. Therefore, depending on the environmental conditions, there are some cultivation environments and crops that are difficult to grow.

In house cultivation and plant factories that can control most of the cultivation environment, if the environment that is effective for growing the crop can be investigated in advance, controlling the cultivation environment will reduce and facilitate farming work. On the other hand, the most widely practiced cultivation of agricultural products in open fields is based on the premise that natural crops are used, and the cultivation environment that can be intentionally controlled is mainly water, fertilizer, weeding, and humidity in the soil due to disinfection. Only control fertilizer concentration and hygiene.

Even when growing the same crop, the environment changes drastically when the field changes, and the characteristics of the soil and the response to fluctuating weather are judged by the farmer's empirical rules, which is an objective and standard for the crop. There is no general guideline for training. If so, it is a method based on specific environmental conditions.

Although these have produced some results, in many cases, there is no objective evaluation as to whether or not the farming is optimal. As a result, improvements in current farming also depend on rules of thumb.

Even if farming based on empirical rules is optimal in the current farming environment, it is not possible to explain the effects and meanings obtained by empirical rules because objective analysis is not possible. Therefore, the inheritance of the rule of thumb based on the objective explanation is also hindered.

Values for farmers are not uniform. For example, some farmers want to make a big profit by making a large amount, and some farmers want to make a big profit by improving the quality even in a small amount. However, in many systems, one value is often used for optimization. Therefore, the conventional system cannot respect the diverse values of farmers.

This will be described with reference to FIG. When the farmer's rule of thumb is related to the measurement items and water application and fertilization, and can be structured as "A is B" and "the relationship between X and Y can be expressed in this way". , Defined as a proposition, and used as input information 3 of a structured empirical rule by a farmer.

Known coefficiented research results 1 for each crop such as plant characteristics, variety information, growth time, etc. regarding the crops cultivated by the farmer, and the planting time, watering, etc. that the farmer considers optimal based on the farmer's crop cultivation experience rules The input information 2 of the fertilizer application amount and the input information 3 of the structured empirical rule regarding the costs such as utility costs and labor costs required for farming are used as the input information to the system after the evaluation 11 of the input information by the system. Then, mainly from the input information 2 of the water application and fertilizer application amount based on the empirical rule by the farmer, the daily plan is quantified as the water application amount and fertilizer application amount for one year and further every hour. .. In addition, one year's worth of daily plans will be used as the planning table. Next, from the input information 3 of the structured rule of thumb by the farmer, the work schedule related to farming and the structured rule of thumb information are created.

These created information is input to the structured database (DB) on the system as the managed schedule attached information 5. As a result, a plan for the crop and its environment can be prepared. These are used as reference plan tables.

Based on the crops and environment specified by the farmer, the reference plan is selected from the similar environments of the similar crops that are the basis of the reference plan. Then, by using the selected reference plan table as the plan table, it is set as the default value when there is no input to the DB based on the empirical rule.

The system shall be able to measure the environment as much as possible, and the measured items are stored in the DB as cultivation environment measurement information and record / evaluation information 8 of growth results.

For example, when the measurement can be performed as in the environment measurement value example 6 by the system, the application example 13 of the environment measurement information to the Penman method can be used. In that case, the amount of transpiration can be actually measured or estimated by numerical analysis as in Example 7 of the numerical analysis result by the system. In the Penman method, the possible evapotranspiration of plants is calculated from the cultivation environment measurement information using only general meteorological information, and the amount is estimated.

The information existing in the plan table and the attached information 5 managed on the DB and the numerical analysis result example 7 by the system are corrected by the correction 12 based on the measured value of the plan. In practice, it focuses on structured heuristics and annual schedules, and makes corrections to the day's plan based on the water and fertilizer consumed the day before.

As a result of the correction 12 based on the measured values of the plan, the system creates an execution table for the day. In addition, the execution table is composed of command creation and execution 21 for automatic control of irrigation (water application) and execution 22 for fertilization recommendation. Then, according to the execution table for the day, water application and fertilization are automatically performed from the system side or instructed to a person.

Before the execution table of the day is created, the price, crop trading market index, shipping amount, etc. of the plan table to be shipped or sold as agricultural products are corrected by the correction information 4 of the plan table by the farmer or an external system. Sometimes.

As the work of the day, the correction contents of the plan and the results for farming are automatically recorded in the system by the environmental measurement information and the record / evaluation information 8 of the results. In addition, a quantified evaluation of farming will be added at a later date, and the results will be recorded in the system.

The environmental measurement information and the record / evaluation information 8 of the actual results are reflected in the plan table and the attached information 5 managed on the DB through the analysis and learning 14 based on the actual results as the substance of the function.
For example, corrections to the annual plan, achievements and evaluations are reflected in the next year's annual plan by using numerical analysis and learning-type artificial intelligence.

It is a farming work in which farmers can be actively involved in crops, and the optimum amount and timing of irrigation and fertilization, which are necessary and indispensable for the growth of crops, can be treated as the focus of decision-making.

When growing crops that are fields owned by a farmer, the farmer has a method of watering or fertilizing that can generate a certain amount of profit. Therefore, as shown in the plan table managed on the DB and the attached information 5, by inputting the daily plan for one year by the farmer, the command creation and execution 21 for automatic control of irrigation (water application) is shown. System automation can be performed for part of farming.

By automating the farming system, all or part of the farmer's rules of thumb have been structured and inherited by the system. In other words, as long as the system is used, the farmer does not have to be a skilled farmer with experience, and even beginners can perform optimal farming for market strategies such as growing crops and evaluating the economy. It becomes.

In addition, if you are a beginner in a situation where you cannot expect input to the system by a skilled farmer, you can expect to achieve certain results by using the reference plan.

The plan table once quantified by the plan table managed on the DB and the attached information 5 can be used as a related element for objective evaluation.

As shown in the environmental measurement information and the record / evaluation information 8 of the actual results, it can be used as an objective evaluation element by giving a numerical evaluation to the actual results.

Analysis by achievements and learning As shown in 14, by using numerical related elements and evaluation elements, analysis using learning-type artificial intelligence including numerical analysis becomes possible. By using the result, it can be used for evaluation and improvement of the plan table managed on the DB and the plan table shown in the attached information 5.

This is shown in analysis and learning 14 based on actual results, even if the schedule and attached information 5 managed on the DB initially depend on the empirical rule of input to the system by a skilled farmer. , An objective evaluation of the rule of thumb will be made by going through an objective method. The ability to repeat objective evaluations of empirical rules and review by learning shows that this system is a feedback system by learning. In addition, the learning effect can be expected to improve by using the system for a long period of time.

And it can be expected that the rule of thumb can be improved by being able to objectively evaluate the rule of thumb.

As a result, this feedback system can be used to know the optimal timing and amount of watering and fertilization under the crops to be cultivated and the farming environment at that time.

There may be a plurality of evaluation elements shown in the environmental measurement information and the record / evaluation information 8 of the results. Analysis based on actual results and learning In the analysis shown in 14, evaluation factors such as shipping time, shipping amount, and quality of agricultural products can be specified as one of the economic-related items. For example, by accelerating the shipping time, the commercial value of agricultural products can be increased.

Further, when there are a plurality of evaluation elements in the analysis, it is generally possible to give priority (weight) to each evaluation element and perform the analysis. For example, try an analysis in line with the purpose of "suppressing costs first and maximizing yield".

There are multiple methods for prioritizing evaluation items and selecting evaluation items. Therefore, it can be expected that the selection can be made according to the values of the farmer.

The system aims to be constructed only for water application and fertilization, but it can be applied even when the number of controllable environmental elements increases, such as in house cultivation and plant factories. In other words, regarding water application and fertilization, by using the plan table and attached information 5 managed on the learned DB, for example, by limiting other controllable environmental elements to one, the environmental elements can be reduced. You can expect to learn the impact. Of course, even if you select multiple controllable environmental elements, you can expect to learn the effects of those environmental elements.

In other words, it can be expected that the values of environmental factors required for house cultivation and plant factories will be clarified by advancing the learning of the plan table and attached information 5 managed on the DB over time. Then, it can be expected to control the environmental elements that can be controlled so as to approach the value.

This is the relationship between the system by this method and external information.

The content shown in [Fig. 1] is a system that can be implemented on a computer.

1 Known coefficientized research results for each crop 2 Input information of water application and fertilizer application amount based on empirical rules by farmers 3 Input information of structured empirical rules by farmers 4 Correction of planning table by farmers and external systems Information 5 Planning table and attached information managed on the DB 6 Example of environmental measurement value by the system 7 Example of numerical analysis result by the system 8 Environmental measurement information and record / evaluation information of actual results 11 Evaluation of input information by the system 12 Measurement value of the plan Correction by 13 Example of application of environmental measurement information to Penman method 14 Analysis and learning based on actual results 21 Creation and execution of commands for automatic control of irrigation (watering) 22 Execution of fertilization recommendation

Claims (10)

In agriculture, the rules of thumb for farming experience, including matters related to the cultivation and growth of agricultural products, as well as economical production, sales, and selling prices, are described as "A is B" and "The relationship between X and Y is. A method of replacing with a structured numerical value (data) or a set of numerical values that can be handled by a computer by making a proposition such as "can be expressed in this way" The method according to claim 1, in which numerical values that can be measured and evaluated by an apparatus and information representing human measurement and evaluation are integrated as a set of numerical values that can be handled by a computer, and then captured as input by the computer. Using part or all of the quantified evaluation elements and part or all of the quantified evaluation results output by the computer, a function that derives the results from the elements is obtained based on correlations and relationships. Furthermore, by using the obtained function, a method of finding the element that derives the best evaluation result while repeating the cultivation and growth of agricultural products, and constructing a feedback system that automatically adjusts the cause depending on the result. The method according to claim 2, which is a farming work in which the farmer can actively participate in the crop, and the optimum amount and timing of irrigation and fertilization, which are necessary and indispensable for the growth of the crop, are intended. How to focus your decisions The method according to claim 2, wherein the empirical rule is expressed by a structure that can be handled by a computer, and by using the structured empirical rule, it does not depend on people for the cultivation of agricultural products and economic market strategy. How to enable inheritance of rules of thumb By using the method according to claim 2, which integrates the measurement or measurement result by the device and the measurement or measurement result by a human, either the measurement or the measurement result by the device or the measurement or the measurement result by a human. How to build a feedback system that can operate even if it is missing A feedback system according to claim 2, which uses the operational information for a long period of time to improve the accuracy of a function that derives a result from an element, and seeks an element that derives the best evaluation result. How to build The method according to claim 2, which uses a standard method such as the Penman method and a known coefficientized measurement result for each crop as a numerical analysis method to obtain a function for deriving a result from an element and provide feedback. How to build a system The method according to claim 2, wherein a feedback system is constructed by obtaining a function for deriving a result from an element by using one or more methods of numerical analysis or analysis by learning type artificial intelligence. The method according to claim 2, wherein the optimum timing and amount of water application and fertilization are determined by using a feedback system that seeks elements for deriving the best evaluation result. The method according to claim 2, wherein the optimized conditions of evaluation factors other than water application and fertilization are obtained by using the constructed feedback system.

Images