JP7653698B2 - Information processing device, information processing system, information processing method, and computer program - Google Patents

Description

The technology disclosed in this specification relates to an information processing device for predicting the saturation deficit of a farm field.

In recent years, while the number of people employed in agriculture has been decreasing overall, the number of new entrants to agriculture has been increasing. New entrants to agriculture are often technically inexperienced, and it is desirable to provide them with technical assistance (support) to ensure stable production. A technology is known that formalizes knowledge about plant (crop) cultivation (e.g., the best time to perform each agricultural task) by describing it in a tabular format, thereby providing support to agricultural workers who lack this knowledge (see, for example, Patent Document 1).

JP 2015-177758 A

In plant cultivation, it is important to appropriately adjust the vapor pressure deficit of the field. Here, the vapor pressure deficit is a value obtained by subtracting the actual amount of water vapor from the saturated water vapor amount. The vapor pressure deficit of the field affects the amount of transpiration of the cultivated plants and determines the photosynthetic efficiency. For example, if the vapor pressure deficit of the field suddenly decreases, the incidence of plant diseases and fruit cracking may increase, and conversely, if the vapor pressure deficit of the field suddenly increases, the stomata of the plants may close and the plants may wither. Therefore, in order to improve the yield and quality of cultivated plants, it is desirable to avoid sudden increases and decreases in the vapor pressure deficit of the field, and to do so, it is necessary to accurately predict the vapor pressure deficit of the field at future times. Conventional technology has the problem that it is not possible to accurately predict the vapor pressure deficit of the field at future times.

This specification discloses a technology that can solve the above-mentioned problems.

The technology disclosed in this specification can be realized, for example, in the following forms:

(1) The information processing device disclosed in this specification is a device for predicting a saturation deficit of a field, and includes an environmental prediction information acquisition unit, a model acquisition unit, and a saturation deficit prediction unit. The environmental prediction information acquisition unit acquires environmental prediction information indicating predicted values of environmental indicators including a field air temperature, which is the air temperature of the field, at a future time. The model acquisition unit acquires a saturation deficit prediction model. The saturation deficit prediction model is a model generated by performing machine learning using a predetermined learning method, with the environmental indicators as explanatory variables and the field saturation deficit, which is the saturation deficit of the field, as an objective variable. The saturation deficit prediction unit predicts the field saturation deficit at the future time using the environmental prediction information and the saturation deficit prediction model. According to this information processing device, the field saturation deficit at a future time is predicted using a saturation deficit prediction model generated by performing machine learning using a predetermined learning method using environmental indicators including the field temperature as explanatory variables and the field saturation deficit, which is the saturation deficit of the field, as the objective variable, and environmental prediction information indicating the predicted values of the environmental indicators including the field temperature. This makes it possible to accurately predict the field saturation deficit at a future time.

(2) The information processing device may further include a prediction information transmission unit that transmits saturation deficit prediction value information indicating the predicted value of the field saturation deficit at the future time to an external device. According to this information processing device, the predicted value of the field saturation deficit at the future time can be made perceptible to the farm worker by the external device, thereby realizing the efficiency of farm work.

(3) The information processing device may further include an adjusted field temperature setting unit that sets multiple adjusted field temperatures based on the predicted value of the field temperature at the future time when the predicted value of the field saturation deficit at the future time is not within a predetermined appropriate range, and a target field temperature setting unit that predicts an adjusted field saturation deficit, which is the adjusted field saturation deficit at the future time, for each adjusted field temperature using the saturation deficit prediction model, and sets a target field temperature that is a target value of the field temperature at the future time based on the predicted value of each adjusted field saturation deficit. According to this information processing device, it is possible to set a target field temperature that is a target value of the field temperature for bringing the field saturation deficit at the future time closer to the appropriate range, making it possible to easily adjust the field saturation deficit and improving the efficiency of agricultural work.

(4) The information processing device may further include a temperature adjustment instruction unit that transmits a temperature adjustment instruction to an external device to bring the field temperature at the future time closer to the target field temperature. This information processing device can realize automatic adjustment (automatic control) of the field saturation deficit, thereby achieving further efficiency in agricultural work.

(5) In the above information processing device, the target field temperature setting unit may be configured to preferentially set as the target field temperature, among the multiple adjusted field temperatures, the predicted value of the adjusted field saturation deficit for each adjusted field temperature having a small absolute deviation from the appropriate range. According to this information processing device, it is possible to set a target field temperature that brings the field saturation deficit at a future time as close as possible to the appropriate range, thereby realizing optimization of agricultural work.

(6) In the above information processing device, the target field temperature setting unit may be configured to preferentially set, as the target field temperature, one of the multiple adjusted field temperatures that has a small absolute difference from the predicted value of the field temperature at the future time. According to this information processing device, it is possible to set a target field temperature that has as little variation as possible from the current predicted value of the field temperature at the future time, thereby realizing optimization of agricultural work.

(7) In the above information processing device, the adjusted field temperature setting unit may be configured to set the multiple adjusted field temperatures by increasing or decreasing the predicted value of the field temperature at the future time at a preset pitch. According to this information processing device, the adjusted field temperature, which is a candidate for the target field temperature, can be set with relatively simple processing, thereby realizing optimization and efficiency of agricultural work.

(8) In the above information processing device, the adjusted field air temperature setting unit may be configured to set the multiple adjusted field air temperatures within a preset temperature range. According to this information processing device, it is possible to avoid the adjusted field air temperatures, which are candidates for the target field air temperature, being set to an unrealistic value, thereby realizing optimization of agricultural work.

(9) In the above information processing device, the predetermined learning method may be a learning method using a decision tree model. According to this information processing device, it is possible to effectively improve the prediction accuracy of the field saturation deficit at a future time.

(10) In the above information processing device, the predetermined learning method may be a learning method using a random forest. According to this information processing device, it is possible to further effectively improve the prediction accuracy of the field saturation deficit at a future time.

(11) In the above information processing device, the environmental index may further include at least one of the following: external temperature, external solar radiation, daily average temperature, daily range, sunshine hours, humidity, and weather. According to this information processing device, it is possible to predict the field saturation deficit at a future time with even greater accuracy.

(12) In the above information processing device, the field may be configured as a semi-open plant factory. According to this information processing device, even in a semi-open plant factory where the internal environment cannot be completely controlled because it is premised on the use of sunlight, it is possible to accurately predict the field saturation deficit at a future time, thereby realizing optimization and efficiency of agricultural work.

(13) The information processing system disclosed in this specification comprises the information processing device and a terminal device capable of communicating with the information processing device via a communication network, the information processing device further comprising a prediction information transmission unit that transmits saturation deficit prediction value information indicating the predicted value of the field saturation deficit at the future time to the terminal device via the communication network, and the terminal device comprises a display unit that displays the predicted value of the field saturation deficit at the future time based on the saturation deficit prediction value information received via the communication network. According to this information processing system, the predicted value of the field saturation deficit at the future time can be visually confirmed by a farm worker on the display unit of the terminal device, thereby making it possible to improve the efficiency of farm work.

The technology disclosed in this specification can be realized in various forms, such as an information processing device, an information processing system, an information processing method, a computer program that realizes these methods, and a non-transitory recording medium on which the computer program is recorded.

FIG. 1 is an explanatory diagram showing a schematic configuration of a farm land management system 10 according to an embodiment of the present invention; FIG. 1 is a block diagram showing a schematic configuration of a farm management server 100. 1 is a flowchart showing the contents of a farm field management process according to an embodiment of the present invention. FIG. 2 is an explanatory diagram conceptually illustrating an example of a saturation deficit prediction model MO. FIG. 2 is an explanatory diagram showing an example of a screen displayed on the display unit 210 of the terminal device 200 during a farm field management process in this embodiment. FIG. 2 is an explanatory diagram showing an example of a screen displayed on the display unit 210 of the terminal device 200 during a farm field management process in this embodiment. An explanatory diagram showing the evaluation results for each learning method An explanatory diagram showing the accuracy of the saturation deficit prediction model MO created using Random Forest.

A. Embodiments:
A-1. Configuration of farm land management system 10:
FIG. 1 is an explanatory diagram showing a schematic configuration of a farm field management system 10 in this embodiment. The farm field management system 10 is a system for managing a farm field, and more specifically, a system for improving the efficiency and optimizing plant cultivation by predicting and controlling the saturation deficit in the farm field. In this embodiment, the farm field is a semi-open type plant factory PF. Here, the semi-open type plant factory PF is a farm field surrounded by a material that does not block sunlight, such as a vinyl greenhouse. However, the farm field management system 10 can also be applied to other types of farm fields (for example, other types of plant factories such as closed and open types, fields for outdoor cultivation, etc.). Note that, although only one plant factory PF is shown in FIG. 1, the farm field management system 10 can be applied to the management of each of multiple plant factories PF. The following describes the management of a specific one plant factory PF by the farm field management system 10. Furthermore, the plants to be cultivated in the farm field may be any type of plant, such as leafy vegetables (e.g., spinach), fruit trees (e.g., tomatoes), ornamental plants, etc. The farm field management system 10 is an example of an information processing system in the claims.

The farm field management system 10 comprises a farm field management server 100, a terminal device 200 used by a farm worker P1, sensors 310 and air conditioners 320 installed in the plant factory PF, and a weather data server 500. Each device constituting the farm field management system 10 is connected to each other so as to be able to communicate with each other via a communication network NET.

The sensor 310 installed in the plant factory PF is a device for measuring the environmental index values of the plant factory PF. In this embodiment, the sensor 310 measures the air temperature, vapor pressure deficit, and external solar radiation in the plant factory PF at a predetermined time interval (e.g., every minute). The sensor 310 has a communication interface (not shown) and transmits the measured values of the environmental indexes to the field management server 100 via the communication network NET, for example, each time a measurement is made. The air conditioner 320 installed in the plant factory PF is a device for adjusting the air temperature in the plant factory PF by supplying cool air or warm air into the plant factory PF.

The terminal device 200 is, for example, a smartphone or a tablet terminal. The terminal device 200 has a display unit 210 configured, for example, by a liquid crystal display or the like (see Figures 5 and 6). In addition, an application program for field management is installed in the terminal device 200. As will be described in more detail later, a user of the terminal device 200 (farmer P1) can use the application program to display the vapor pressure deficit in the plant factory PF on the display unit 210.

The weather data server 500 is a server device that functions as a database that provides weather data indicating environmental index values. As the weather data, for example, 1 km square mesh weather data is used. In addition, examples of the environmental index values indicated by the weather data include outside temperature, daily average temperature (minimum daily average temperature, maximum daily average temperature, etc.), daily range, hours of sunshine, humidity, weather, etc. The weather data used in this embodiment includes actual values (actual measured values) for the past period and forecast values for the future period.

The field management server 100 is a server device for managing the plant factory PF. FIG. 2 is a block diagram showing a schematic configuration of the field management server 100. The field management server 100 includes a control unit 110, a memory unit 120, a display unit 130, an operation input unit 140, and an interface unit 150. These units are connected to each other via a bus 190 so as to be able to communicate with each other. The field management server 100 is an example of an information processing device within the scope of the claims.

The display unit 130 of the farm field management server 100 is, for example, a liquid crystal display, and displays various images and information. The operation input unit 140 is, for example, a keyboard, mouse, buttons, microphone, and the like, and accepts operations and instructions from the administrator. The display unit 130 may be equipped with a touch panel so as to function as the operation input unit 140. The interface unit 150 is, for example, a LAN interface, USB interface, and the like, and communicates with other devices via wired or wireless connection.

The storage unit 120 of the field management server 100 is composed of, for example, a ROM, a RAM, a hard disk drive (HDD), etc., and is used to store various programs and data, and as a working area when executing various programs, and as a temporary storage area for data. For example, the storage unit 120 stores a field management program CP, which is a computer program for executing the field management process described below. The field management program CP is provided in a state stored in a computer-readable recording medium (not shown), such as a CD-ROM, DVD-ROM, or USB memory, and is stored in the storage unit 120 by installing it in the field management server 100.

The storage unit 120 of the field management server 100 also stores learning data LD including environmental history information EH, a saturation deficit prediction model MO, environmental prediction information EI, and saturation deficit prediction value information SI in the field management process described below. These data and models will be explained in conjunction with the explanation of the field management process described below.

The control unit 110 of the field management server 100 is configured, for example, by a CPU, and controls the operation of the field management server 100 by executing a computer program read from the memory unit 120. For example, the control unit 110 functions as a field management unit 111 that executes the field management process described below by reading and executing a field management program CP from the memory unit 120. The field management unit 111 includes a model acquisition unit 113, an environmental prediction information acquisition unit 114, a saturation deficit prediction unit 115, a prediction information transmission unit 116, an adjusted field temperature setting unit 117, a target field temperature setting unit 118, and a temperature adjustment instruction unit 119. The functions of each of these units will be explained in conjunction with the explanation of the field management process described below.

A-2. Field management treatment:
Next, the field management process executed by the field management server 100 of this embodiment will be described. Fig. 3 is a flowchart showing the contents of the field management process in this embodiment. The field management process is a process for managing the plant factory PF, and more specifically, a process for predicting a vapor pressure deficit in the plant factory PF at a future time t1, and instructing adjustment of the air temperature in the plant factory PF so that the vapor pressure deficit approaches the appropriate range R1 when the predicted value of the vapor pressure deficit is not included in a predetermined appropriate range R1. The field management process is started in response to, for example, an administrator operating the operation input unit 140 of the field management server 100 to input a start instruction.

In the field management process, first, the model acquisition unit 113 (FIG. 2) of the field management server 100 acquires the saturation deficit prediction model MO (S110). FIG. 4 is an explanatory diagram conceptually showing an example of the saturation deficit prediction model MO. As shown in FIG. 4, the saturation deficit prediction model MO is a model for estimating the objective variable from the explanatory variables, with the environmental index of the plant factory PF as the explanatory variable and the saturation deficit in the plant factory PF as the objective variable. In this embodiment, the temperature in the plant factory PF, external solar radiation, external temperature, average minimum daily temperature, average maximum daily temperature, average daily range, and average sunshine hours are used as the explanatory variables.

The model acquisition unit 113 acquires (creates) a saturation deficit prediction model MO by performing a predetermined machine learning using the learning data LD stored in the memory unit 120. As shown in FIG. 2, the learning data LD used to acquire the saturation deficit prediction model MO includes environmental history information EH. The environmental history information EH is information indicating the history of environmental index values (index values used as explanatory variables and objective variables of the saturation deficit prediction model MO) in the plant factory PF. Of the environmental index values included in the environmental history information EH, the saturation deficit in the plant factory PF, the temperature in the plant factory PF, and the external solar radiation are measured values obtained from the sensor 310 installed in the plant factory PF, and the other environmental index values are actual values acquired from the weather data server 500.

The model acquisition unit 113 creates a saturation deficit prediction model MO by performing a predetermined machine learning process using the environmental index values other than the saturation deficit in the plant factory PF as explanatory variables and the saturation deficit in the plant factory PF as the objective variable among the environmental index values indicated by the environmental history information EH included in the learning data LD. For example, known machine learning algorithms including linear models such as multiple regression analysis, ridge regression, and lasso regression, and decision tree models such as Random Forest and XGBoost can be used to create the saturation deficit prediction model MO. As described later, it is preferable to use a decision tree model to create the saturation deficit prediction model MO, and it is particularly preferable to use Random Forest among the decision tree models. The created saturation deficit prediction model MO is stored in the memory unit 120 of the farm management server 100.

In this embodiment, the model acquisition unit 113 of the farm management server 100 acquires (creates) the saturation deficit prediction model MO by performing machine learning using the learning data LD, but the model acquisition unit 113 may also acquire the saturation deficit prediction model MO created by another device via the communication network NET.

Next, the environmental prediction information acquisition unit 114 (FIG. 2) of the farm management server 100 acquires the environmental prediction information EI (S120). The environmental prediction information EI is information indicating the predicted value of the environmental index at the future time t1, which is the predicted time of the vapor pressure deficit in the plant factory PF. As described above, in this embodiment, the temperature in the plant factory PF, external solar radiation, external temperature, average minimum daily temperature, average maximum daily temperature, average daily range, and average sunshine hours are used as explanatory variables of the vapor pressure deficit prediction model MO. Of these, forecast values acquired from the weather data server 500 are used for the external solar radiation and external temperature, actual values acquired from the weather data server 500 are used for the average minimum daily temperature, average maximum daily temperature, average daily range, and average sunshine hours, and a predicted value predicted from these information using a known method is used for the temperature in the plant factory PF. The known method may be, for example, a regression analysis method, a machine learning method such as support vector regression (SVR) or random forest, or the like.

Next, the saturation deficit prediction unit 115 (FIG. 2) of the farm management server 100 predicts the saturation deficit in the plant factory PF at future time t1 using the environmental prediction information EI and the saturation deficit prediction model MO (S130). That is, the saturation deficit prediction unit 115 inputs the environmental index value (explanatory variable value) indicated by the environmental prediction information EI to the saturation deficit prediction model MO, and obtains the predicted value of the saturation deficit in the plant factory PF output from the saturation deficit prediction model MO.

In this embodiment, when the prediction of the vapor pressure deficit in the plant factory PF at future time t1 (S130) is completed, the prediction information transmission unit 116 (FIG. 2) of the farm land management server 100 transmits vapor pressure deficit prediction value information SI indicating the prediction value to the terminal device 200 via the communication network NET. As a result, as shown in FIG. 5, the prediction value of the vapor pressure deficit in the plant factory PF at future time t1 (2.7 g/m ³ in the example of FIG. 5) is displayed on the display unit 210 of the terminal device 200. In the example of FIG. 5, in addition to the prediction value of the vapor pressure deficit, the display unit 210 of the terminal device 200 also displays the plant factory name, the crop name, the current value of the vapor pressure deficit (5.2 g/m ³ in the example of FIG. 5), and a graph showing the progress of the vapor pressure deficit from the past to the present to the future. By visually checking the information displayed on the display unit 210 of the terminal device 200, the farmer P1 can understand the current performance and future predictions regarding the saturation deficit within the plant factory PF without having to go to the plant factory PF.

Next, the field management unit 111 (FIG. 2) of the field management server 100 judges whether the predicted value of the vapor pressure deficit in the plant factory PF at the future time t1 is included in a preset appropriate range R1 of vapor pressure deficit (S140). The appropriate range R1 of the vapor pressure deficit in the plant factory PF is, for example, 3.0 to 7.0 g/ ^m3 . The appropriate range R1 of the vapor pressure deficit in the plant factory PF may be changed depending on the cultivated variety. If it is judged that the predicted value of the vapor pressure deficit in the plant factory PF at the future time t1 is included in the appropriate range R1 (S140: YES), the field management unit 111 concludes that there is no problem with the vapor pressure deficit in the plant factory PF at the future time t1 and ends the field management process. Note that, thereafter, when the future time t1 is reached or at a timing around that time, the process from S120 onwards is repeatedly executed.

On the other hand, if it is determined that the predicted value of the vapor pressure deficit in the plant factory PF at the future time t1 is not included in the appropriate range R1 (S140: NO), it is determined that there is a problem with the vapor pressure deficit in the plant factory PF at the future time t1, and the field management server 100 executes the process described below to bring the vapor pressure deficit in the plant factory PF at the future time t1 closer to the appropriate range R1. Specifically, the adjusted field temperature setting unit 117 (FIG. 2) of the field management server 100 sets multiple adjusted field temperatures Tx based on the predicted value (temperature To) of the temperature in the plant factory PF at the future time t1 (S150). The multiple adjusted field temperatures Tx are candidates for the target value of the temperature in the plant factory PF at the future time t1. In this embodiment, the adjusted field air temperature setting unit 117 sets multiple adjusted field air temperatures Tx by increasing or decreasing the predicted value (air temperature To) of the air temperature in the plant factory PF at future time t1 at a preset interval (in this embodiment, in increments of 1°C). However, the upper limit of the absolute value of the increase or decrease is set to i (e.g., i=10), and a minimum temperature Tmin (e.g., Tmin=16°C) and a maximum temperature Tmax (e.g., Tmax=28°C) that define the effective temperature range are set. Therefore, a set of multiple adjusted field air temperatures Tx is expressed as follows:
Tx=(To-i, To-i+1, To-i+2,..., To+i-1, To+i)
However, Tmin≦Tx≦Tmax

Next, the target field temperature setting unit 118 (FIG. 2) of the field management server 100 predicts the adjusted field saturation deficit Sx, which is the saturation deficit in the plant factory PF after adjustment at future time t1, for each adjusted field temperature Tx using the saturation deficit prediction model MO (S160). The prediction of the adjusted field saturation deficit Sx is performed using the environmental prediction information EI and the saturation deficit prediction model MO, as in the process of S130. However, at this time, among the input values (explanatory variables) to the saturation deficit prediction model MO, the above-mentioned adjusted field temperature Tx is used for the temperature in the plant factory PF instead of the temperature To shown in the environmental prediction information EI. The prediction of the adjusted field saturation deficit Sx is repeatedly performed the number of times equal to the number of adjusted field temperatures Tx. This allows the predicted value of the vapor pressure deficit in the plant factory PF (adjusted field vapor pressure deficit Sx) to be obtained assuming that the temperature in the plant factory PF at future time t1 will be the adjusted field temperature Tx, rather than the current predicted value of temperature To.

Next, the target field temperature setting unit 118 ( FIG. 2 ) of the field management server 100 sets a target field temperature Tt, which is a target value for the temperature in the plant factory PF at future time t1, based on the predicted value of the adjusted field saturation deficit Sx for each adjusted field temperature Tx (S170). In this embodiment, the target field temperature setting unit 118 sorts a list of multiple adjusted field temperatures Tx according to the following priorities, and sets the adjusted field temperature Tx located at the top of the sorted list as the target field temperature Tt.
(1) In ascending order of the absolute value of the deviation of the predicted value of the adjusted field vapor deficit Sx from the appropriate range R1 of the vapor deficit; (2) In ascending order of the absolute value of the difference with the predicted value (temperature To) of the temperature in the plant factory PF at a future time t1; (3) In descending order of the adjusted field temperature Tx.

For example, if all of the adjusted field saturation deficits Sx calculated for each adjusted field temperature Tx are not included in the appropriate saturation deficit range R1, the adjusted field temperature Tx corresponding to the predicted adjusted field saturation deficit Sx with the smallest absolute deviation from the appropriate saturation deficit range R1 is set as the target field temperature Tt according to the above priority order (1). In this case, if there are multiple predicted adjusted field saturation deficits Sx with the smallest absolute deviation from the appropriate saturation deficit range R1, they are sorted according to the above priority orders (2) and (3). Also, if multiple predicted adjusted field saturation deficits Sx calculated for each adjusted field temperature Tx are included in the appropriate saturation deficit range R1, the corresponding multiple predicted adjusted field temperatures Tx with the smallest absolute difference from the predicted value (temperature To) of the temperature in the plant factory PF at future time t1 is set as the target field temperature Tt according to the above priority order (2). In this case, if there are multiple values with the smallest absolute difference from the predicted value (temperature To) of the temperature in the plant factory PF at future time t1, they are sorted according to the above priority order (3). By setting the target field temperature Tt in this way, it is possible to bring the saturation deficit in the plant factory PF at future time t1 as close as possible to the appropriate range R1 among the multiple adjusted field temperatures Tx, and a temperature that has as little variation as possible from the current predicted value (temperature To) of the temperature in the plant factory PF at future time t1 and is as high as possible is set as the target field temperature Tt.

Next, the field management unit 111 (FIG. 2) of the field management server 100 determines whether the set target field temperature Tt is the same as the predicted value (temperature To) of the temperature in the plant factory PF at the future time t1 (S180). If it is determined that the target field temperature Tt is not the same as the predicted value (temperature To) of the temperature in the plant factory PF at the future time t1 (S180: NO), the temperature adjustment instruction unit 119 of the field management server 100 transmits a temperature adjustment instruction to the air conditioner 320 of the plant factory PF via the communication network NET to bring the temperature in the plant factory PF at the future time t1 closer to the target field temperature Tt (S190). For example, if the target field temperature Tt is higher than the predicted value (temperature To) of the temperature in the plant factory PF at future time t1, the temperature adjustment instruction unit 119 issues a temperature adjustment instruction to the air conditioner 320 to raise the temperature in the plant factory PF. Conversely, if the target field temperature Tt is lower than the predicted value (temperature To) of the temperature in the plant factory PF at future time t1, the temperature adjustment instruction unit 119 issues a temperature adjustment instruction to the air conditioner 320 to lower the temperature in the plant factory PF. The air conditioner 320 that receives the temperature adjustment instruction operates in accordance with the instruction, so that the temperature in the plant factory PF at future time t1 can be brought closer to the target field temperature Tt, and as a result, the vapor pressure deficit in the plant factory PF at future time t1 can be brought closer to the appropriate vapor pressure deficit range R1.

In this embodiment, when the issuance of the temperature adjustment instruction (S190) is completed, the prediction information transmission unit 116 ( FIG. 2 ) of the farm land management server 100 transmits, via the communication network NET, to the terminal device 200, the predicted value information SI indicating the predicted value of the vapor deficit when the air conditioner 320 operates in accordance with the temperature adjustment instruction (after the internal air temperature is adjusted). As a result, as shown in FIG 6 , the display unit 210 of the terminal device 200 displays the predicted value of the vapor deficit in the plant factory PF at a future time t1 after the internal air temperature is adjusted (3.5 g/ ^m3 in the example of FIG 6 , which is within the appropriate range R1 of the vapor deficit). In the example of Figure 6, the graph showing the trend in saturation deficit shows the trend in saturation deficit before and after the internal air temperature adjustment, and by visually checking the information shown on the display unit 210 of the terminal device 200, the farmer P1 can understand that the internal air temperature adjustment by the air conditioner 320 is expected to cause the saturation deficit in the plant factory PF at a future time t1 to fall within an appropriate range R1.

If it is determined that the target field temperature Tt is the same as the predicted temperature (temperature To) in the plant factory PF at the future time t1 (S180: YES), the process of S190 is skipped. This ends the field management process. Thereafter, when the future time t1 arrives or at a timing around that time, the processes from S120 onwards are repeatedly executed.

A-3. Example:
A saturation deficit prediction model MO was created using various learning methods, and was evaluated. FIG. 7 is an explanatory diagram showing the evaluation results for each learning method. In the evaluation, environmental index values were collected from a certain plant factory PF during the period from April 2018 to December of the same year. The "average years" for the environmental index values were set to 1981 to 2010. All 244 collected data were randomly divided, and 70% (171 cases) were used as learning data, and the remaining 30% (73 cases) were used as verification data. Using the learning data, machine learning was performed using each learning method (multiple regression analysis, ridge regression, lasso regression, Random Forest, XGBoost) to create a saturation deficit prediction model MO, and the created saturation deficit prediction model MO was evaluated using the verification data. The root mean square error (RMSE) and correlation coefficient (R) were used as evaluation items. The root mean square error takes a value of 0 or more, and the closer to 0, the better the accuracy. The correlation coefficient takes a value of -1 or more and 1 or less, and the closer to 1, the better the accuracy.

As shown in FIG. 7, when a decision tree model (Random Forest, XGBoost) was used, better evaluation results were obtained compared to when a linear model (multiple regression analysis, ridge regression, lasso regression) was used. When Random Forest, one of the decision tree models, was used, particularly good evaluation results were obtained. FIG. 8 is an explanatory diagram showing the accuracy of the saturation deficit prediction model MO created using Random Forest. As shown in FIG. 8, the saturation deficit prediction model MO created using Random Forest showed generally good accuracy in predicting the saturation deficit in the plant factory PF.

A-4. Advantages of this embodiment:
As described above, the field management server 100 of the present embodiment is an information processing device for predicting the saturation deficit of the field (plant factory PF), and includes an environmental prediction information acquisition unit 114, a model acquisition unit 113, and a saturation deficit prediction unit 115. The environmental prediction information acquisition unit 114 acquires environmental prediction information EI indicating predicted values of environmental indicators including the field temperature, which is the temperature of the field, at a future time t1. The model acquisition unit 113 acquires a saturation deficit prediction model MO. The saturation deficit prediction model MO is a model generated by performing machine learning using a predetermined learning method, with the environmental indicators including the field temperature as explanatory variables and the field saturation deficit, which is the saturation deficit of the field, as an objective variable. The saturation deficit prediction unit 115 predicts the field saturation deficit at a future time t1 using the environmental prediction information EI and the saturation deficit prediction model MO.

In this way, according to the field management server 100 of this embodiment, the field saturation deficit at future time t1 is predicted using a saturation deficit prediction model MO generated by performing machine learning using a predetermined learning method using environmental indicators including the field temperature as explanatory variables and the field saturation deficit, which is the saturation deficit of the field, as the objective variable, and using environmental prediction information EI indicating predicted values of the environmental indicators including the field temperature. Therefore, the field saturation deficit at future time t1 can be predicted with high accuracy.

The farm field management server 100 of this embodiment further includes a prediction information transmission unit 116 that transmits saturation deficit prediction value information SI indicating the predicted value of the field saturation deficit at future time t1 to the terminal device 200 as an external device. Therefore, according to the farm field management server 100 of this embodiment, the predicted value of the field saturation deficit at future time t1 can be made perceptible to the farm worker P1 by the terminal device 200, thereby realizing the efficiency of farm work.

In addition, the field management server 100 of this embodiment further includes an adjusted field temperature setting unit 117 and a target field temperature setting unit 118. When the predicted value of the field saturation deficit at future time t1 is not included in the predetermined appropriate range R1, the adjusted field temperature setting unit 117 sets multiple adjusted field temperatures Tx based on the predicted value of the field temperature at future time t1 (temperature To). The target field temperature setting unit 118 predicts the adjusted field saturation deficit Sx, which is the adjusted field saturation deficit at future time t1, for each adjusted field temperature Tx using the saturation deficit prediction model MO, and sets the target field temperature Tt, which is the target value of the field temperature at future time t1, based on the predicted value of each adjusted field saturation deficit Sx. Therefore, according to the field management server 100 of this embodiment, it is possible to set a target field temperature Tt, which is a target value of the field temperature for bringing the field saturation deficit at a future time t1 closer to the appropriate range R1, and it is possible to easily adjust the field saturation deficit and realize the efficiency of agricultural work.

The field management server 100 of this embodiment further includes a temperature adjustment instruction unit 119 that transmits a temperature adjustment instruction to the air conditioner 320 as an external device to bring the field temperature at future time t1 closer to the target field temperature Tt. Therefore, the field management server 100 of this embodiment can realize automatic adjustment (automatic control) of the field saturation deficit, thereby achieving further efficiency in agricultural work.

In addition, in the field management server 100 of this embodiment, the target field temperature setting unit 118 preferentially sets, as the target field temperature Tt, the predicted value of the adjusted field saturation deficit Sx for each adjusted field temperature Tx, among the multiple adjusted field temperatures Tx, which has a small absolute deviation from the appropriate range R1. Therefore, according to the field management server 100 of this embodiment, it is possible to set a target field temperature Tt that brings the field saturation deficit at future time t1 as close as possible to the appropriate range R1, thereby realizing optimization of agricultural work.

In addition, in the field management server 100 of this embodiment, the target field temperature setting unit 118 preferentially sets, as the target field temperature Tt, the adjusted field temperature Tx having the smallest absolute difference from the predicted field temperature at future time t1 (temperature To). Therefore, according to the field management server 100 of this embodiment, it is possible to set a target field temperature Tt that varies as little as possible from the current predicted field temperature at future time t1 (temperature To), thereby optimizing agricultural work.

In addition, in the field management server 100 of this embodiment, the adjusted field air temperature setting unit 117 sets multiple adjusted field air temperatures Tx by increasing or decreasing the predicted value of the field air temperature (air temperature To) at future time t1 at a preset pitch. Therefore, according to the field management server 100 of this embodiment, the adjusted field air temperature Tx, which is a candidate for the target field air temperature Tt, can be set with relatively simple processing, thereby realizing optimization and efficiency of agricultural work.

In addition, in the field management server 100 of this embodiment, the adjusted field air temperature setting unit 117 sets multiple adjusted field air temperatures Tx within a preset temperature range. Therefore, according to the field management server 100 of this embodiment, it is possible to avoid setting the adjusted field air temperature Tx, which is a candidate for the target field air temperature Tt, to an unrealistic value, thereby realizing optimization of agricultural work.

The specified learning method used to obtain the saturation deficit prediction model MO is preferably a learning method using a decision tree model. By adopting such a configuration, it is possible to effectively improve the prediction accuracy of the field saturation deficit at future time t1. The specified learning method used to obtain the saturation deficit prediction model MO is preferably a learning method using a random forest. By adopting such a configuration, it is possible to extremely effectively improve the prediction accuracy of the field saturation deficit at future time t1.

In addition, in the farm field management server 100 of this embodiment, the environmental index further includes at least one of the following: external temperature, external solar radiation, daily average temperature, daily range, sunshine hours, humidity, and weather. Therefore, according to the farm field management server 100 of this embodiment, it is possible to predict the field saturation deficit at future time t1 with even greater accuracy.

In addition, in this embodiment, the field is a semi-open plant factory. According to the field management server 100 of this embodiment, even in a semi-open plant factory where the internal environment cannot be completely controlled because it is based on the use of sunlight, it is possible to accurately predict the field saturation deficit at future time t1, thereby optimizing and streamlining agricultural work.

The farm field management system 10 of this embodiment also includes a farm field management server 100 and a terminal device 200 capable of communicating with the farm field management server 100 via a communication network NET. The farm field management server 100 further includes a prediction information transmission unit 116 that transmits saturation deficit prediction value information SI indicating a prediction value of the field saturation deficit at a future time t1 to the terminal device 200 via the communication network NET. The terminal device 200 includes a display unit 210 that displays a prediction value of the field saturation deficit at a future time t1 based on the saturation deficit prediction value information SI received via the communication network NET. According to the farm field management system 10 of this embodiment, the prediction value of the field saturation deficit at a future time t1 can be visually confirmed by the display unit 210 of the terminal device 200 to the farm worker P1, thereby realizing the efficiency of farm work.

B. Variations:
The technology disclosed in this specification is not limited to the above-described embodiments, and can be modified in various forms without departing from the spirit of the invention. For example, the following modifications are also possible.

The configuration of the farm field management system 10 in the above embodiment is merely an example and can be modified in various ways. For example, in the above embodiment, the farm field management system 10 may not be equipped with the weather data server 500. In addition, in the above embodiment, the plant factory PF is provided with an air conditioner 320 that adjusts the temperature in the plant factory PF by supplying cold air or warm air into the plant factory PF, but instead of the air conditioner 320 or together with the air conditioner 320, another device (external device) that adjusts the temperature in the plant factory PF may be provided. Examples of such other devices include a fan that adjusts the temperature in the plant factory PF by adjusting the air flow in the plant factory PF, and a shading device that adjusts the temperature in the plant factory PF by adjusting the amount of solar radiation into the plant factory PF.

The configuration of the field management server 100 in the above embodiment is merely an example, and various modifications are possible. For example, in the above embodiment, the learning data LD including the environmental history information EH and the vapor deficit prediction model MO are stored in the memory unit 120 of the field management server 100, but the learning data LD and the vapor deficit prediction model MO may be stored in a separate server, and the field management server 100 may refer to the learning data LD and the vapor deficit prediction model MO stored in the separate server.

The content of the field management process in the above embodiment is merely an example and can be modified in various ways. For example, the method of setting the appropriate range R1, the adjusted field temperature Tx, and the target field temperature Tt in the above embodiment is merely an example and can be modified in various ways. In addition, in the above embodiment, after predicting the field saturation deficit at future time t1 (S130), the process from determining whether the predicted value is included within R1 (S140) onwards is executed, but the process from S140 onwards does not have to be executed.

In addition, in the above embodiment, a plant factory PF is used as an example of a farm field, but the present invention can be similarly applied to other types of farm fields (e.g., fields for open-field cultivation).

In addition, in each of the above embodiments, some of the configurations realized by hardware may be replaced by software, and conversely, some of the configurations realized by software may be replaced by hardware.

10: Field management system 100: Field management server 110: Control unit 111: Field management unit 113: Model acquisition unit 114: Environmental prediction information acquisition unit 115: Saturation deficit prediction unit 116: Prediction information transmission unit 117: Adjusted field temperature setting unit 118: Target field temperature setting unit 119: Temperature adjustment instruction unit 120: Memory unit 130: Display unit 140: Operation input unit 150: Interface unit 190: Bus 200: Terminal device 210: Display unit 310: Sensor 320: Air conditioner 500: Weather data server CP: Field management program EH: Environmental history information EI: Environmental prediction information LD: Learning data MO: Saturation deficit prediction model NET: Communication network P1: Farm worker PF: Plant factory R1: Appropriate range SI: Saturation deficit prediction information Sx: Adjusted field saturation deficit Tt: Target field temperature Tx: Adjusted field temperature

Claims (14)

An information processing device for predicting a saturation deficit of a farm field,
an environmental prediction information acquisition unit that acquires environmental prediction information indicating predicted values of environmental indices including a field temperature, which is the temperature of the field , an external temperature, external solar radiation, a daily average temperature, a daily range, and sunshine hours at a future time;
a model acquisition unit that acquires a saturation deficit prediction model, the saturation deficit prediction model being a model generated by performing machine learning using a predetermined learning method with the environmental index as an explanatory variable and a field saturation deficit, which is a saturation deficit of the field, as an objective variable;
a saturation deficit prediction unit that predicts the field saturation deficit at the future time by using the environmental prediction information and the saturation deficit prediction model;
an adjusted field air temperature setting unit that sets a plurality of adjusted field air temperatures based on the predicted value of the field air temperature at the future time when the predicted value of the field saturation deficit at the future time is not included in a predetermined appropriate range;
a target field temperature setting unit that uses the saturation deficit prediction model to predict an adjusted field saturation deficit, which is the field saturation deficit after adjustment at the future time, for each of the adjusted field air temperatures, and sets a target field temperature, which is a target value of the field air temperature at the future time, based on the predicted value of each adjusted field saturation deficit;
An information processing device comprising: The information processing device according to claim 1 , further comprising:
an information processing device comprising: a prediction information transmitting unit configured to transmit, to an external device, saturation deficit prediction value information indicating a prediction value of the field saturation deficit at the future time. 3. The information processing device according to claim 1 , further comprising:
an information processing device comprising: a temperature adjustment instruction unit that transmits a temperature adjustment instruction to an external device to bring the field temperature at the future time closer to the target field temperature; The information processing device according to any one of claims 1 to 3 ,
The target field temperature setting unit preferentially sets, as the target field temperature, the adjusted field temperature for which the predicted value of the adjusted field saturation deficit for each of the adjusted field temperatures has a small absolute deviation from the appropriate range, among the multiple adjusted field temperature. The information processing device according to any one of claims 1 to 4 ,
The target field temperature setting unit preferentially sets, as the target field temperature, one of the adjusted field temperatures having a small absolute value of difference from the predicted value of the field temperature at the future time, according to the information processing device. The information processing device according to any one of claims 1 to 5 ,
The adjusted field air temperature setting unit sets the plurality of adjusted field air temperatures by increasing or decreasing the predicted value of the field air temperature at the future time at a preset pitch. The information processing device according to any one of claims 1 to 6 ,
The adjusted field air temperature setting unit sets the plurality of adjusted field air temperatures within a preset temperature range. The information processing device according to any one of claims 1 to 7 ,
The information processing device, wherein the predetermined learning method is a learning method using a decision tree model. 9. The information processing device according to claim 8 ,
The information processing device, wherein the predetermined learning method is a learning method using a random forest. The information processing device according to any one of claims 1 to 9 ,
The information processing device, wherein the environmental indicators further include at least one of external humidity and weather. The information processing device according to any one of claims 1 to 10 ,
The information processing device, wherein the field is a semi-open plant factory. An information processing device according to any one of claims 1 to 11 ,
a terminal device capable of communicating with the information processing device via a communication network;
An information processing system comprising:
The information processing device further includes a prediction information transmission unit that transmits saturation deficit prediction value information indicating a prediction value of the field saturation deficit at the future time to the terminal device via the communication network,
The terminal device includes a display unit configured to display the predicted value of the field saturation deficit at the future time based on the saturation deficit predicted value information received via the communication network. An information processing method for predicting a saturation deficit in a farm field, comprising:
acquiring environmental prediction information indicating predicted values of environmental indices including a field temperature, which is the temperature of the field , an external temperature, external solar radiation, a daily average temperature, a daily range, and sunshine hours at a future time;
A step of acquiring a saturation deficit prediction model, the saturation deficit prediction model being a model generated by performing machine learning using a predetermined learning method with the environmental index as an explanatory variable and a field saturation deficit, which is a saturation deficit of the field, as a target variable;
predicting the field saturation deficit at the future time by using the environmental prediction information and the saturation deficit prediction model;
setting a plurality of adjusted field temperatures based on the predicted value of the field air temperature at the future time when the predicted value of the field saturation deficit at the future time is not included in a predetermined appropriate range;
A step of predicting an adjusted field saturation deficit, which is the field saturation deficit after adjustment at the future time, for each of the adjusted field air temperatures using the saturation deficit prediction model, and setting a target field air temperature, which is a target value of the field air temperature at the future time, based on the predicted value of each adjusted field saturation deficit;
An information processing method comprising: A computer program for predicting a saturation deficit in a field, comprising:
On the computer,
A process of acquiring environmental prediction information indicating predicted values of environmental indices including a field temperature, which is the temperature of the field , an external temperature, external solar radiation, a daily average temperature, a daily range, and sunshine hours at a future time;
A process of acquiring a saturation deficit prediction model, the saturation deficit prediction model being a model generated by performing machine learning using a predetermined learning method with the environmental index as an explanatory variable and a field saturation deficit, which is a saturation deficit of the field, as a target variable;
A process of predicting the field saturation deficit at the future time by using the environmental prediction information and the saturation deficit prediction model;
a process of setting a plurality of adjusted field temperatures based on the predicted value of the field air temperature at the future time when the predicted value of the field saturation deficit at the future time is not included within a predetermined appropriate range;
a process of predicting an adjusted field saturation deficit, which is the field saturation deficit after adjustment at the future time, for each of the adjusted field air temperatures using the saturation deficit prediction model, and setting a target field air temperature, which is a target value of the field air temperature at the future time, based on the predicted value of each adjusted field saturation deficit;
A computer program that executes the following:

Images