Growth Estimation Device, Growth Estimation Method, and Growth Estimation Program

Abstract

To provide a growth estimation device or the like that calculates, by calculation, parameters that have a great influence on plant growth with a small amount of performance data **Solution means**: A growth estimation device 1 that estimates elements having a great influence on plant growth, including: a first calculation unit 24 that calculates a first predicted value P1 regarding the growth state of a plant based on calculation variables representing a plurality of types of elements; a second calculation unit 25 that calculates a second predicted value P2 regarding the growth state of the plant based on the same calculation variables; a coefficient calculation unit 26 that calculates a plurality of first coefficients αi corresponding to each of the plurality of calculation variables in the first model and a plurality of second coefficients βi corresponding to each of the plurality of calculation variables in the second model so that the first predicted value P1 and the second predicted value P2 are the same or approximate within a predetermined range; and an important element extraction unit 27 that extracts important elements having a great influence on plant growth according to the change amounts of the predicted values P1 and P2 with respect to the change amounts of the first coefficient (αi) and / or the second coefficient (βi) in the coefficient calculation unit 26.

Description

Technical Field

[0001] The present invention relates to a growth estimation device that estimates parameters having a great influence on plant growth, etc.

Background Art

[0002] Various parameters are involved in plant growth, and by controlling them, it becomes possible to predict the flowering time or adjust the harvesting time. As a technique for predicting the production results of agricultural crops, for example, the technique disclosed in Patent Document 1 is disclosed.

[0003] The technique shown in Patent Document 1 is a prediction model that refers to the observed value data of meteorological data before the prediction target date and represents the influence of production results and weather, and predicts the production results from the forecast value of the meteorological data on the prediction target date. It includes a prediction model generation unit that generates a prediction model, and the prediction model generation unit selects a combination of variables of meteorological data so that the difference between the actual measurement value of the production results in a certain past year and the predicted value of the production results in that year predicted using the prediction model is minimized, and generates a prediction model.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, with the technique shown in Patent Document 1, it is not possible to know how what parameters are involved in the growth of agricultural crops. In addition, since it generates a model that only approaches past data, it is necessary to collect a large amount of performance data, and there is a problem that it takes a long time to improve the accuracy.

[0006] The present invention has been made to solve the above problems, and an object thereof is to provide a growth estimation device, a growth estimation method, and a growth estimation program that calculate, by calculation, with a small amount of performance data, parameters that have a great influence on plant growth.

Means for Solving the Problems

[0007] The growth estimation device according to the present invention is a growth estimation device that estimates elements having a great influence in plant growth, and based on calculation variables for each of the plurality of types of elements representing the elements, a first calculation unit that calculates a first predicted value related to the growth state of the plant using a first model, a second calculation unit that calculates a second predicted value different from the first model related to the growth state of the plant using a second model based on the same calculation variables as the calculation variables, a coefficient calculation unit that calculates a plurality of first coefficients corresponding to each of the plurality of calculation variables in the first model and a plurality of second coefficients corresponding to each of the plurality of calculation variables in the second model so that the first predicted value and the second predicted value are the same or approximate within a predetermined range, and in the coefficient calculation unit, according to the change amount of the predicted value with respect to the change amount of the first coefficient and / or the second coefficient, an important element extraction unit that extracts important elements having a great influence in plant growth from the elements represented by the calculation variables corresponding to each of the first coefficient and / or the second coefficient.

[0008] Thus, in the growth estimation device according to the present invention, a first predicted value regarding the growth state of a plant is calculated using a first model that calculates the first predicted value based on calculation variables for each of the plurality of types of the above-described elements, and a second predicted value different from the first model regarding the growth state of the plant is calculated using a second model that calculates the second predicted value based on the same calculation variables, and a plurality of first coefficients corresponding to each of the plurality of calculation variables in the first model and a plurality of second coefficients corresponding to each of the plurality of calculation variables in the second model are calculated such that the first predicted value and the second predicted value are the same or approximate within a predetermined range, and important elements having a large influence degree in the growth of the plant are extracted according to the change amount of the predicted value with respect to the change amount of the first coefficient and / or the second coefficient. Therefore, elements having a significant influence on the growth of the plant can be mathematically determined using two calculation models, and an effect that it can greatly contribute to the growth work of the plant can be achieved.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Modes for Carrying Out the Invention

[0010] (The First Embodiment of the Present Invention) The growth estimation device according to this embodiment will be described with reference to FIGS. 1 to 6. The growth estimation device according to this embodiment estimates elements with a large degree of influence from two different calculation models for a plurality of types of elements that affect the growth of plants, such as temperature, humidity, light irradiation, nutrients, moisture, ethylene concentration, and drying period. In the following description, the calculation variable representing elements related to temperature, such as daily average temperature and nighttime temperature, is T, the calculation variable representing elements related to humidity, such as relative humidity, is H, the calculation variable representing elements related to light irradiation, such as sunshine duration (photoperiod) and light intensity, is L, the calculation variable representing elements related to nutrients, such as the application amounts of nitrogen, phosphoric acid, and potassium fertilizers, is N, the calculation variable representing elements related to moisture, such as the moisture content of the soil, is W, the calculation variable representing elements related to ethylene concentration, such as the concentration of plant hormones, is E, and the calculation variable representing elements related to the drying period, such as the duration of the drying period, is D.

[0011] Note that the various elements related to plant growth are not limited to these. For example, environmental variables considering regionality (e.g., soil conditions, previous growth state, growth state of surrounding fields, disaster frequency, etc.) or variables according to the state of seeds and seedlings may be added. Also, estimation may be performed considering all of the above elements, or estimation may be performed considering only some of them.

[0012] FIG. 1 is a block diagram showing the hardware configuration of the growth estimation device according to the present embodiment. The growth estimation device 1 can be obtained by installing a growth estimation program on a computer such as a personal computer or a tablet, and having the computer on which the growth estimation program is installed execute the growth estimation method. In FIG. 1, the growth estimation device 1 includes a CPU 11, a RAM 12, a ROM 13, a hard disk (referred to as HD) 14, an input / output I / F 15, and a communication I / F 16. The operating system and various programs are stored in the ROM 13 and the HD 14, and are read into the RAM 12 as needed, and each program is executed by the CPU 11. The communication I / F 16 is an interface for performing communication between devices. The input / output I / F 15 receives input from input devices such as a keyboard and a touch panel, and displays calculation results and the like on output devices such as a display. Note that the above configuration is merely an example and can be changed as needed.

[0013] FIG. 2 is a functional block diagram showing the configuration of the growth estimation device according to the present embodiment. The growth estimation device 1 acquires, as input information 21, data for each of various elements measured during a predetermined period (for example, one year or one growing season) when a plant is grown, and registers it in the growth information storage unit 23. An input unit 22, a first calculation unit 24 that calculates a first predicted value using a linear regression model based on the data stored in the growth information storage unit 23, and a second calculation unit 25 that calculates a second predicted value using a non-linear regression model based on the data stored in the growth information storage unit 23, and a coefficient calculation unit 26 that calculates an optimal value by varying coefficients corresponding to various elements so that the first predicted value and the second predicted value match or approximate within a predetermined range, and an important element extraction unit 27 that extracts elements with high importance from the calculation results of the coefficient calculation unit 26, and an output control unit 28 that outputs information on the extracted important elements.

[0014] In the growth information storage unit 23, information on various elements measured during a predetermined period in plant growth is accumulated. For example, when growing a certain plant, the ambient temperature change is measured every second by a temperature sensor, and the temperature information measured by the temperature sensor is transmitted to the growth estimation device 1 every second and accumulated in the growth information storage unit 23. In addition to this, various sensors measure for each of the various elements as described above, and are accumulated in the growth information storage unit 23 via the input unit 22.

[0015] Note that the data input via the input unit 22 may be transmitted using the communication function for the information measured by various sensors as described above, or may be manually input and registered by the user using an input device. For example, since it is difficult to detect the type and amount of fertilizer applied by a sensor, it may be manually input. That is, elements that can be easily measured by a sensor are acquired by the sensor, and the data is registered in the growth information storage unit 23 using communication or a portable memory or the like, and elements that are difficult to acquire by the sensor may be registered in the growth information storage unit 23 by the user performing manual input or the like.

[0016] The first calculation unit 24 calculates a first predicted value using the linear regression model of the following formula (1) based on various data stored in the growth information storage unit 23.

[0017] [Number]

[0018] In formula (1), P1 is a predicted value representing the probability and timing of transitioning to each process in plant growth such as germination, leaf emergence, flowering, fruiting, and withering, and αi is a coefficient indicating the degree of influence (weight) of various elements (calculation variables). Note that as the linear regression model, for example, a multiple regression model, a simple regression model, a logistic regression model, a ridge regression model, a lasso regression model, an Elastic Net regression model, a robust regression model, a quantile regression model, a Bayesian linear regression model, etc. may be used.

[0019] Further, the second calculation unit 25 calculates a second predicted value using the non-linear regression model of the following formula (2) based on various data stored in the growth information storage unit 23.

[0020]

Number

[0021] In formula (2), P2 is a predicted value representing the probability and timing of transitioning to each process in plant growth such as germination, leaf emergence, flowering, fruiting, and withering, and βi is a coefficient indicating the degree of influence (weight) of various factors (calculation variables). As the non-linear regression model, for example, a hyperbolic regression model, a polynomial regression model, an exponential regression model, a logistic curve regression model, a Gaussian curve regression model, a spline regression model, etc. may be used.

[0022] The coefficient calculation unit 26 calculates the optimal value of the coefficient based on the calculation of formula (1) in the first calculation unit 24 and the calculation of formula (2) in the second calculation unit 25. Specifically, it calculates αi and βi such that P1 = P2 or P1 ≒ P2 (P1 - P2 < a predetermined threshold). For example, formula (1) is defined as the objective function (fixed), formula (2) is defined as the function to be adjusted (coefficient adjustment), and the coefficients of formula (2) are adjusted so that formula (2) approaches formula (1). To detect the difference between the respective functions, a loss function such as the mean squared error or the mean absolute error is defined and minimized by an analytical method or numerical optimization (gradient descent method, quadratic programming method, library-based optimization, etc.). Based on the coefficient βi obtained in the minimization process, formula (2) is plotted on a graph to confirm whether it approaches the graph of formula (1).

[0023] An example of the processing of the above-described first calculation unit 24, second calculation unit 25, and coefficient calculation unit 26 will be described below with reference to the following schematic diagrams. FIG. 3 is a schematic diagram for explaining the processing of the first calculation unit in the growth estimation device according to the present embodiment, FIG. 4 is a schematic diagram for explaining the processing of the second calculation unit in the growth estimation device according to the present embodiment, and FIG. 5 is a schematic diagram for explaining the processing of the coefficient calculation unit in the growth estimation device according to the present embodiment. First, in the processing of the first calculation unit 24, the data for each of the various elements registered in the growth information storage unit 23 is plotted on a graph with the horizontal axis representing time and the vertical axis representing the measured value (see FIG. 3(A)). At this time, the values obtained for each of the various elements are normalized and aggregated into one dimension. That is, the points plotted in FIG. 3(A) are plot data in which the measured values of the various elements are mixed. For these plotted data, the first calculation unit 24 performs an operation to draw a linear graph as shown in FIG. 3(B) using the linear regression model of equation (1).

[0024] Similarly, in the processing of the second calculation unit 25, the data for each of the various elements registered in the growth information storage unit 23 is plotted on a graph with the horizontal axis representing time and the vertical axis representing the measured value (see FIG. 4(A)). At this time, the values obtained for each of the various elements are normalized and aggregated into one dimension. For these plotted data, the second calculation unit 25 performs an operation to draw a non-linear graph as shown in FIG. 4(B) using the non-linear regression model of equation (2).

[0025] When the graphs of FIGS. 3(B) and 4(B) are obtained, the coefficient calculation unit 26 calculates the coefficients αi and βi so that the respective graphs match or approximate each other as shown in FIG. 5. FIGS. 5(A) to (C) show how the graph of FIG. 4(B) approximates the graph of FIG. 3(B) step by step. In FIG. 5, the state of obtaining the optimal value of βi by fixing αi and adjusting βi is shown. However, it is also possible to make the graphs of FIGS. 3(B) and 4(B) match or approximate each other by changing αi with βi as a reference, or to perform an operation to make the graphs of FIGS. 3(B) and 4(B) match or approximate each other while changing both αi and βi.

[0026] The method for obtaining the optimal value of βi for the graphs of FIGS. 3(B) and 4(B) to match or approximate can use the solution methods and algorithms of the generally known optimization problems as described above. At this time, for example, image processing by AI (artificial intelligence) may be combined to obtain the optimal value. Specifically, the shapes of the graphs of FIG. 3(B) and FIG. 4(B) may be analyzed by image processing, and the optimal value of βi for which the graph shapes are approximated may be calculated by AI.

[0027] In the process of calculating the optimal value of βi, the important element extraction unit 27 identifies and extracts βi for which the degree of shrinkage of the difference between P1 and P2 with respect to the change in βi is the largest. In other words, βi that brings P2 closest to P1 when βi is changed. Then, it is estimated that the element of the calculation variable corresponding to the extracted βi is the element that has the greatest influence on the target plant.

[0028] When αi is adjusted with βi fixed, in the process of calculating the optimal value of αi, the important element extraction unit 27 identifies and extracts αi for which the degree of shrinkage of the difference between P1 and P2 with respect to the change in αi is the largest. In other words, αi that brings P1 closest to P2 when αi is changed. Then, it is estimated that the element of the calculation variable corresponding to the extracted αi is the element that has the greatest influence on the target plant.

[0029] When the optimal values are calculated by changing both αi and βi with respect to each other, in the process of calculating the optimal values of αi and βi, the important element extraction unit 27 identifies αi for which the degree of shrinkage of the difference between P1 and P2 with respect to the change in αi is the largest. In other words, αi that brings P1 closest to P2 when αi is changed, and at the same time, identifies and extracts βi for which the degree of shrinkage of the difference between P1 and P2 with respect to the change in βi is the largest. In other words, βi that brings P2 closest to P1 when βi is changed. Then, it is estimated that the elements of the calculation variables corresponding to the extracted αi and βi are the elements that have the greatest influence on the target plant. That is, in this case, the two elements obtained from equations (1) and (2) may be identified as important elements.

[0030] The output control unit 28 outputs the information of the elements with a large influence extracted by the important element extraction unit 27 to a display unit such as a display. The user of the cultivation estimation device 1 (i.e., the grower of the plant) can adjust the work during cultivation according to the growth of the plant or create an efficient work plan based on the output element information.

[0031] Each of the above calculations can be performed for the location, type, and cultivation process of the plant to be cultivated. For example, even when the same plant (e.g., vanilla of any variety) is cultivated under the same conditions in adjacent greenhouses, in reality, there may be differences in the subtle length and timing of sunlight due to the influence of shadows, or the amount (pressure) of water supplied may differ between the greenhouses, resulting in differences in the growth rate and quality. That is, even a slight environmental difference will result in different influencing factors. Therefore, as described above, it is desirable to use the data for each measured element for each location (even for adjacent cases, for each greenhouse, and more precisely, for each ridge, etc.) as the input information 21 for the calculation.

[0032] Also, when cultivating different varieties of plants in the same greenhouse, it is desirable to use the data for each measured element for each variety as the input information 21 for the calculation.

[0033] Furthermore, it is desirable to use the data for each measured element for each cultivation process such as germination, leaf emergence, flowering, fruiting, and withering as the input information 21 for the calculation. By doing so, it becomes possible to identify important elements for each cultivation process, such as elements with a large influence on "germination" and elements with a large influence on "flowering".

[0034] Furthermore, it is also possible to identify important elements for each time period. For example, by obtaining the important elements obtained when the data measured from 7:00 to 19:00 during the day is used for calculation as input information 21 and the important elements obtained when the data measured from 19:00 to 7:00 is used for calculation as input information 21 respectively, it is possible to identify the important elements during the daytime (time period with sunlight) and the important elements during the nighttime (time period without sunlight) respectively, and it becomes possible to devise such as changing or modifying the cultivation method according to the time period.

[0035] Furthermore, when calculating the coefficients αi and βi, for example, in the case where the graph shape is particularly approximated in some time periods, it may be estimated that any element in that time period is a time period that has a great influence on the growth of plants.

[0036] FIG. 6 is a flowchart showing the processing of the growth estimation device according to the present embodiment. When the growth estimation device 1 performs processing, it is assumed that the performance data of various elements (temperature, humidity, light irradiation, nutrients, moisture, ethylene concentration, drying period, etc.) measured in the previous year are registered in advance in the growth information storage unit 23. The user of the growth estimation device 1 operates the input device to specify the item to be processed (S1). The item to be processed here refers to, for example, the type of plant, the growth process, the location, the period, etc. that are the objects of the growth estimation process by the growth estimation device 1. Specifically, for example, for the "XX variety of vanilla", an item such as "greenhouse A" in the growth process until the vanilla flower "blooms" (or a specific period such as "XX month to XX month") is specified. Based on these input information and specified information, the information extraction unit (not shown) reads out the data to be calculated from the growth information storage unit 23 (S2).

[0037] Note that it is not necessary to specify all of the above items in S1, and only some items may be specified. Also, in addition to this, a configuration may be adopted such that items related to the growth of plants, such as season, weather, precipitation, disaster-related, etc., can be specified.

[0038] The first calculation unit 24 calculates the coefficient αi based on the data extracted in S2 (S3). The calculation performed here is, for example, a calculation using a linear regression model such as a multiple linear regression model. The second calculation unit 25 calculates the coefficient βi based on the data extracted in S2 (S4). The calculation performed here is, for example, a calculation using a non-linear regression model such as a hyperbolic regression model. The coefficient calculation unit 26 calculates the coefficient αi and / or βi so that the predicted values or graphs obtained in S3 and S4 match or approximate (S5). The calculation of the optimal values of the coefficients αi and βi for which the predicted values match may be, for example, to analyze the shape of the graph obtained from the calculation results of the first calculation unit 24 and the second calculation unit 25 by image processing and obtain the optimal values of the coefficients αi and βi while looking at their similarity.

[0039] The important element extraction unit 27 extracts, as important elements, the elements corresponding to αi or βi that contributed the most to making P1 and P2 match or approximate in the process of obtaining the optimal values of αi and βi (S6). The output control unit 28 displays the information of the extracted important elements on the display (S7) and ends the process. The user of the growth estimation device 1 will make a growth plan or take measures for quality improvement based on the information of the important elements displayed on the display.

[0040] As described above, in the growth estimation device 1 according to the present embodiment, a first calculation unit 24 that uses a first model to calculate a first predicted value (P1) regarding the growth state of a plant based on calculation variables for each of a plurality of types of elements representing the plurality of types of elements, a second calculation unit 25 that uses a second model to calculate a second predicted value (P2) different from the first model regarding the growth state of the plant based on the same calculation variables as the calculation variables, and a coefficient calculation unit 26 that calculates a plurality of first coefficients (αi) corresponding to each of the plurality of calculation variables in the first model and a plurality of second coefficients (βi) corresponding to each of the plurality of calculation variables in the second model so that the first predicted value and the second predicted value are the same or approximate within a predetermined range, and in the coefficient calculation unit 26, according to the change amounts of the predicted values P1 and P2 with respect to the change amounts of the first coefficient (αi) and / or the second coefficient (βi), an important element extraction unit 27 that extracts important elements having a large influence degree in the growth of the plant from the elements represented by the calculation variables corresponding to each of the first coefficient (αi) and / or the second coefficient (βi) is provided. Therefore, elements that have an important influence on the growth of the plant can be mathematically obtained using two calculation models, and can greatly contribute to the plant growth operation.

[0041] Further, if necessary, since the important element extraction unit 27 extracts important elements according to the change amounts of the differences between the first predicted value (P1) and the second predicted value (P2) with respect to the change amounts of the first coefficient (αi) or the second coefficient (βi) in the process where the coefficient calculation unit 26 calculates the first coefficient (αi) and the second coefficient (βi), elements that have an important influence on the growth of the plant can be extracted with high accuracy.

[0042] Furthermore, if necessary, since the first model is a linear regression model and the second model is a non - linear regression model, elements that have an important influence on the growth of the plant can be mathematically obtained from two different calculation models.

[0043] Furthermore, if necessary, in the process of the coefficient calculation unit 26 calculating the first coefficient (αi) and the second coefficient (βi), the similarity / dissimilarity between the graph image obtained by the first calculation unit 24 and the graph image obtained by the second calculation unit 25 is calculated by image analysis, and the first coefficient (αi) and / or the second coefficient (βi) is calculated so that the respective graphs are similar. Therefore, the operation of the coefficient calculation unit 26 can be easily and quickly performed by image processing.

Explanation of Signs

[0044] 1 Cultivation Estimation Device 11 CPU 12 RAM 13 ROM 14 Hard Disk 15 Input / Output I / F 16 Communication I / F 21 Input Information 22 Input Unit 23 Cultivation Information Storage Unit 24 First Calculation Unit 25 Second Calculation Unit 26 Coefficient Calculation Unit 27 Important Element Extraction Unit 28 Output Control Unit

Claims

1. A growth estimation device for estimating elements having a large influence on plant growth, comprising: a first calculation unit that uses a first model for calculating a first predicted value related to the growth state of a plant based on calculation variables for each of a plurality of types of said elements; a second calculation unit that uses a second model for calculating a second predicted value different from the first model related to the growth state of a plant based on the same calculation variables as the calculation variables; a coefficient calculation unit that calculates a plurality of first coefficients corresponding to each of the plurality of calculation variables in the first model and a plurality of second coefficients corresponding to each of the plurality of calculation variables in the second model so that the first predicted value and the second predicted value are the same or approximate within a predetermined range; an important element extraction unit that extracts, from the elements represented by the calculation variables corresponding to the respective first coefficients and / or second coefficients, important elements having a large influence on plant growth according to the change amount of the predicted value with respect to the change amount of the first coefficient and / or the second coefficient in the coefficient calculation unit; A growth estimation device, characterized by comprising the above.

2. In the growth estimation device according to Claim 1, the important element extraction unit, in the process where the coefficient calculation unit calculates the first coefficient and the second coefficient, extracts the important element according to the change amount of the difference between the first predicted value and the second predicted value with respect to the change amount of each of the first coefficient or the second coefficient. A growth estimation device characterized by this.

3. In the growth estimation device according to Claim 1 or 2, the first model is a linear regression model, and the second model is a non - linear regression model. A growth estimation device characterized by this.

4. In the growth estimation device according to Claim 3, in the process where the coefficient calculation unit calculates the first coefficient and the second coefficient, calculates the similarity / dissimilarity of the graph image obtained by the first calculation unit and the graph image obtained by the second calculation unit by image analysis, and the first coefficient and / or the second coefficient so that the respective graphs are similar A growth estimation device, characterized by calculating the above.

5. A growth estimation method for estimating elements having a large influence on plant growth, comprising: a first calculation step of using a first model for calculating a first predicted value related to the growth state of a plant based on calculation variables for each of a plurality of types of said elements; A second calculation step of using a second model that calculates a second predicted value different from the first model regarding the growth state of the plant based on the same calculation variables as the calculation variables; A coefficient calculation step of calculating a plurality of first coefficients corresponding to each of the plurality of calculation variables in the first model and a plurality of second coefficients corresponding to each of the plurality of calculation variables in the second model so that the first predicted value and the second predicted value are the same or approximate within a predetermined range; An important factor extraction step of extracting, as an important factor having a large influence on plant growth, the factor represented by the calculation variable corresponding to the first coefficient and / or the second coefficient whose relative change amount is equal to or greater than a predetermined value in the coefficient calculation step; A growth estimation method characterized by including the above.

6. A growth estimation program for estimating factors having a large influence on plant growth, comprising: A first calculation means using a first model that calculates a first predicted value regarding the growth state of the plant based on calculation variables for each of a plurality of types of the factors; A second calculation means using a second model that calculates a second predicted value different from the first model regarding the growth state of the plant based on the same calculation variables as the calculation variables; A coefficient calculation means for calculating a plurality of first coefficients corresponding to each of the plurality of calculation variables in the first model and a plurality of second coefficients corresponding to each of the plurality of calculation variables in the second model so that the first predicted value and the second predicted value are the same or approximate within a predetermined range; An important factor extraction means for extracting, as an important factor having a large influence on plant growth, the factor represented by the calculation variable corresponding to the first coefficient and / or the second coefficient whose relative change amount is equal to or greater than a predetermined value in the coefficient calculation means; A growth estimation program characterized by causing a computer to function as above.

Images