Support system and support method

Abstract

To provide a support system that can efficiently grasp the occurrence status of pests in a cultivation facility.SOLUTION: This support system 1 is provided with: a first storage unit 12 for storing data indicating the environment of a cultivation facility 60 and data indicating the occurrence status of pests; and a first control unit 11 that performs machine learning using the data indicating the environment and the data indicating the occurrence status of pests acquired from the first storage unit 12 to generate a prediction model for predicting the probability of occurrence of pests in the cultivation facility 60.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a support system and a support method for supporting a cultivation manager. [Background technology]

[0002] Conventionally, a technology for grasping the occurrence status of pests in a cultivation facility is disclosed, for example, in Patent Document 1. In the technology disclosed in Patent Document 1, when a worker discovers a pest during work such as harvesting, the worker uses a mobile terminal to transmit information about the discovery of the pest to a management terminal. Next, the management terminal creates a pest management table based on the information about the discovery of the pest received from the mobile terminal and outputs the table to a monitor or the like. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2018-85981 Summary of the Invention [Problem to be solved by the invention]

[0004] With the technology disclosed in Patent Document 1, unless an operator checks whether or not pests or diseases have occurred in all areas of the cultivation facility, it is not possible to comprehensively grasp the pest and disease occurrence status of the cultivation facility. Therefore, the technology disclosed in Patent Document 1 has room for improvement in terms of efficiently grasping the pest and disease occurrence status of the cultivation facility.

[0005] Therefore, an object of the present disclosure is to provide a support system and support method that can efficiently grasp the occurrence status of pests and diseases in cultivation facilities. [Means for solving the problem]

[0006] In some embodiments, the support system includes a first memory unit that stores data indicating the environment of at least one cultivation facility and data indicating the occurrence status of pests and diseases, and a first control unit that generates a predictive model for predicting the probability of occurrence of pests and diseases in the cultivation facility by performing machine learning using the data indicating the environment and the data indicating the occurrence status of pests and diseases acquired from the first memory unit.

[0007] This allows the cultivation manager to efficiently grasp the occurrence of pests and diseases in the cultivation facility.

[0008] In one embodiment, the support system may further include a group of sensors that measure data indicating the environment of the cultivation facility, a second control unit that inputs the data indicating the environment obtained from the group of sensors into the predictive model and predicts the probability of occurrence of the pests and diseases in the cultivation facility, and an output unit that outputs the probability of occurrence of the pests and diseases predicted by the second control unit.

[0009] This allows the cultivation manager to prioritize management of areas within each cultivation facility that are more likely to be affected by pests and diseases.

[0010] In one embodiment, the output unit may further output data indicating the actual occurrence status of the pests in the cultivation facility.

[0011] This allows the cultivation manager to efficiently understand the prediction accuracy of the prediction model.

[0012] In one embodiment, the second control unit may update the predictive model by performing machine learning using data indicating the environment measured by the sensor group and data indicating the actual occurrence status of the pests.

[0013] This allows for improved prediction accuracy of the prediction model for each cultivation facility.

[0014] In one embodiment, the second control unit may control the environmental control device installed in the cultivation facility based on input from the cultivation manager so as to reduce the probability of occurrence of the pests and diseases in the cultivation facility.

[0015] This makes it possible to suppress damage caused by pests.

[0016] In one embodiment, the second control unit may generate a pest control plan based on the probability of occurrence of the pests in the cultivation facility, and the output unit may output the pest control plan generated by the second control unit.

[0017] This allows the cultivation manager to efficiently manage the cultivation facility based on the pest control plan.

[0018] In one embodiment, the second control unit may optimize the position and / or number of sensors included in the sensor group based on data indicating the environment of the cultivation facility, and the output unit may output the optimized position and / or number of the sensors.

[0019] This reduces the costs required for managing the cultivation facility.

[0020] In some embodiments, the support method is a computer-based support method, and includes a step in which the computer performs machine learning using data indicating the environment of at least one cultivation facility and data indicating the occurrence status of pests and diseases, thereby generating a predictive model for predicting the probability of pest and disease occurrence in the cultivation facility.

[0021] This allows the cultivation manager to efficiently grasp the occurrence of pests and diseases in the cultivation facility.

[0022] In one embodiment, the support method may further include the steps of the computer acquiring data indicating the environment of the cultivation facility, inputting the acquired data indicating the environment into the prediction model to predict the probability of occurrence of the pests and diseases in the cultivation facility, and outputting the predicted probability of occurrence of the pests and diseases.

[0023] This allows the cultivation manager to prioritize management of areas within each cultivation facility that are more likely to be affected by pests and diseases. [Effects of the Invention]

[0024] According to the present disclosure, it is possible to provide a support system and a support method that are capable of efficiently understanding the occurrence status of pests and diseases in a cultivation facility. [Brief explanation of the drawings]

[0025] [Figure 1] 1 is a functional block diagram showing the configuration of a support system according to an embodiment of the present invention. [Figure 2] 10 is a flowchart illustrating a first operation example of the support system according to the present embodiment. [Figure 3] FIG. 3 is a schematic diagram illustrating the process of step S101 in FIG. 2. [Figure 4] FIG. 3 is a schematic diagram illustrating the process of step S102 in FIG. 2. [Figure 5] FIG. 3 is a schematic diagram illustrating the process of step S103 in FIG. 2. [Figure 6] FIG. 3 is a schematic diagram illustrating the process of step S104 in FIG. 2. [Figure 7] 10 is a flowchart illustrating a second operation example of the support system according to the present embodiment. [Figure 8] FIG. 8 is a schematic diagram illustrating the process of step S203 in FIG. 7. [Figure 9] FIG. 8 is a schematic diagram illustrating the process of step S204 in FIG. 7. [Figure 10] 10 is a flowchart illustrating a third operation example of the support system according to the present embodiment. [Figure 11] FIG. 11 is a schematic diagram illustrating the process of step S302 in FIG. [Figure 12] FIG. 11 is a schematic diagram illustrating the process of step S303 in FIG. [Figure 13] 11 is a schematic diagram illustrating an example of output from the process of step S306 in FIG. 10. FIG. [Figure 14] FIG. 10 is a schematic diagram illustrating an example of a screen of a mobile terminal. [Figure 15] 10 is a flowchart illustrating a fourth operation example of the support system according to the present embodiment. [Figure 16] FIG. 16 is a schematic diagram illustrating an example of output from the process of step S407 in FIG. [Figure 17] This is an image analyzed by a spectral camera. DETAILED DESCRIPTION OF THE INVENTION

[0026] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In each drawing, the same reference numerals indicate the same or equivalent components.

[0027] With reference to FIG. 1, a support system 1 according to an embodiment of the present disclosure will be described.

[0028] The support system 1 includes a server device 10, a client device 20, a sensor group 30, an environmental control device 40, and a mobile terminal 50. The server device 10 is capable of communicating with the client device 20 via a network 70 such as the Internet. The server device 10 may also be capable of communicating with the sensor group 30, the environmental control device 40, and the mobile terminal 50 via the network 70. The client device 20, the sensor group 30, and the environmental control device 40 are each installed in at least one cultivation facility 60, for example, cultivation facilities 60 across the country. The mobile terminal 50 is carried by the cultivation manager.

[0029] The server device 10 is a server that belongs to a cloud computing system or the like.

[0030] The server device 10 includes a first control unit 11, a first storage unit 12, and a first communication unit 13.

[0031] The first control unit 11 includes one or more processors. The processor may be, for example, a general-purpose processor such as a central processing unit (CPU) or a dedicated processor specialized for a specific process. The first control unit 11 controls each unit of the server device 10 and performs processes related to the operation of the server device 10.

[0032] The first storage unit 12 includes one or more semiconductor memories, one or more magnetic memories, one or more optical memories, or a combination thereof. The semiconductor memory is, for example, a random access memory (RAM) or a read only memory (ROM). The first storage unit 12 stores information used in the operation of the server device 10 and information obtained by the operation of the server device 10.

[0033] The first communication unit 13 includes one or more communication interfaces that can communicate via, for example, the Internet, a gateway, a LAN (local area network), etc. The first communication unit 13 receives information used in the operation of the server device 10 and transmits information obtained by the operation of the server device 10.

[0034] The operation of the server device 10 is realized by executing a program stored in the first storage unit 12 by a processor included in the first control unit 11.

[0035] The client device 20 is a mobile phone, a smartphone, a tablet, a PC (personal computer), or the like.

[0036] The client device 20 includes a second control unit 21, a second storage unit 22, a second communication unit 23, an input unit 24, and an output unit 25.

[0037] The second control unit 21 includes one or more processors. The processor may be, for example, a general-purpose processor such as a central processing unit (CPU) or a dedicated processor specialized for a specific process. The second control unit 21 controls each unit of the client device 20 and performs processes related to the operation of the client device 20.

[0038] The second storage unit 22 includes one or more semiconductor memories, one or more magnetic memories, one or more optical memories, or a combination thereof. The semiconductor memory is, for example, a random access memory (RAM) or a read only memory (ROM). The second storage unit 22 stores information used in the operation of the client device 20 and information obtained by the operation of the client device 20.

[0039] The second communication unit 23 includes one or more communication interfaces capable of communicating via, for example, the Internet, a gateway, a local area network (LAN), etc. The second communication unit 23 receives information used in the operation of the client device 20 and transmits information obtained by the operation of the client device 20.

[0040] The input unit 24 includes one or more input interfaces. The input interfaces are, for example, physical keys, capacitance keys, a pointing device, a touch screen integrated with a display, a microphone, or a combination thereof. The input unit 24 accepts an operation to input information used in the operation of the client device 20.

[0041] The output unit 25 includes one or more output interfaces. The output interfaces are, for example, a display such as an LCD (liquid crystal display) or an organic EL (electro luminescence) display, a speaker, or a combination thereof. The output unit 25 outputs information obtained by the operation of the client device 20.

[0042] The operation of the client device 20 is realized by executing a program stored in the second storage unit 22 by a processor included in the second control unit 21.

[0043] The sensor group 30 includes one or more sensors, such as, but not limited to, a CO2 concentration meter, a pyranometer, a thermometer, a hygrometer, or a combination thereof. The sensor group 30 can communicate with the client device 20 via a LAN or the like.

[0044] The environmental control device 40 includes, but is not limited to, a sprinkler system, a lighting system, an air conditioner, a curtain, a skylight, a CO2 generator, or a combination thereof. The environmental control device 40 can communicate with the client device 20 via a LAN or the like. The environmental control device 40 is controlled by a second control unit 21 of the client device 20.

[0045] The mobile terminal 50 is, but is not limited to, a mobile phone, a smartphone, a tablet, etc. The mobile terminal 50 is capable of communicating with the client device 20 via Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0046] The cultivation facility 60 may be, but is not limited to, a farm field, a greenhouse, a vinyl house, or the like.

[0047] In the above-described configuration, the second control unit 21 of the client device 20 may be responsible for some or all of the processing of the first control unit 11 of the server device 10. Alternatively, the first control unit 11 of the server device 10 may be responsible for some or all of the processing of the second control unit 21 of the client device 20.

[0048] A first operation example of the support system 1 according to this embodiment will be described with reference to Fig. 2. The first operation example corresponds to an embodiment of the support method according to the present disclosure.

[0049] In step S101, the first control unit 11 of the server device 10 acquires data indicating the environment of each cultivation facility 60 and data indicating the occurrence of pests and diseases at predetermined time intervals from the client devices 20 installed in each of the cultivation facilities 60, such as cultivation facilities 60 across the country, preferably multiple cultivation facilities 60, via the first communication unit 13. The first control unit 11 also stores the acquired data indicating the environment and data indicating the occurrence of pests and diseases in the first memory unit 12. As shown in FIG. 3, the first control unit 11 may classify the acquired data by the location of the cultivation facility 60 (e.g., Kanto, Hokuriku), the variety (A, B, etc.) of the plant or the like cultivated in the cultivation facility 60, and the type of pest or disease (e.g., gray mold, powdery mildew, etc.), and store the classified data in the first memory unit 12.

[0050] In this embodiment, the "environment" includes, but is not limited to, CO2 concentration, solar radiation, temperature, humidity, or a combination thereof. Data indicating the environment is acquired by a sensor group 30 installed in each cultivation facility 60. In addition, in this embodiment, the "pests" include, but are not limited to, gray mold, powdery mildew, leaf mold, or a combination thereof. The occurrence status of pests is determined by visual inspection by the cultivation manager or based on images captured by a camera or the like. Details will be described in the third operation example. In addition, in this embodiment, the "predetermined time interval" can be appropriately set by the cultivation manager and can be, for example, several tens of minutes to several hours, but is not limited to this.

[0051] In step S102, the first control unit 11 acquires data indicating the environment and data indicating the occurrence status of pests and diseases from the first memory unit 12 and performs preprocessing. In this embodiment, "preprocessing" includes, but is not limited to, missing value processing, outlier processing, normalization processing, discretization processing, or a combination thereof. For example, "preprocessing" may include creating new variables based on sensor values ​​such as differential or integral values ​​over a predetermined period (e.g., 1 to 6 hours). Discretization processing includes, but is not limited to, equal frequency division, equal number division, chi-merge, or a combination thereof. These processes are appropriately selected depending on the clustering method described below. Figure 4 shows the discretized data. When it is desired to predict the occurrence probability of a specific pest or disease, appropriate weighting may be applied to the objective variable corresponding to the presence or absence of the specific pest or disease.

[0052] In step S103, the first control unit 11 generates a prediction model for predicting the probability of pest occurrence by performing machine learning using the data preprocessed in step S102. Step S103 will be described in detail below.

[0053] For example, the first control unit 11 can use clustering and Bayesian network techniques to generate a predictive model in which variables corresponding to environmental data are used as explanatory variables and variables corresponding to the presence or absence of pests are used as objective variables. A Bayesian network is a directed acyclic graph with variables as nodes and includes a conditional probability table between the nodes. Clustering is a technique for classifying similar variables into the same cluster. Examples of clustering include, but are not limited to, hard clustering such as k-means or soft clustering such as probabilistic latent semantic analysis. In step S103, the first control unit 11 first performs clustering on the time series using the data preprocessed in step S102. This classifies similar explanatory variables into the same cluster and reduces the dimensionality. Note that while clustering is a so-called unsupervised learning method, if there is a particularly focused objective variable, such as one indicating the occurrence of pests, the influence of similarity on that variable may be appropriately adjusted by weighting that variable. Figure 5 shows the clustering results. Next, the first control unit 11 generates a prediction model for predicting the probability of pest or disease occurrence by performing structural learning using the clustering results. Figure 6 shows an example of a prediction model including a parent node corresponding to data indicating the environment, a child node whose dimensionality has been reduced by clustering, and a grandchild node corresponding to the presence or absence of a specified pest or disease occurrence in a specified variety. Note that the parent node and grandchild node do not necessarily need to be linked via the child node; they may be directly linked. In this way, by using a prediction model that combines dimensionality reduction by clustering with a Bayesian network that visualizes the dependencies between nodes, cultivation managers can efficiently interpret complex biological phenomena in plants, etc.

[0054] In step S103, the first control unit 11 can perform machine learning using data indicating the environment of the cultivation facility 60 up to the present and data indicating the current occurrence status of pests and diseases. This allows the first control unit 11 to generate a prediction model that can predict the current occurrence probability of pests and diseases. In addition, the first control unit 11 can perform machine learning using data indicating the environment of the cultivation facility 60 in the past (for example, up to three days ago) and data indicating the current occurrence status of pests and diseases. This allows the first control unit 11 to generate a prediction model that can predict the occurrence probability of pests and diseases in the future (for example, three days from now). Furthermore, the first control unit 11 can generate a prediction model that can predict the current and / or future occurrence probability of pests and diseases for each variety of plants, etc., and each type of pest and disease in the cultivation facility 60.

[0055] In step S104, the first control unit 11 stores the prediction model generated in step S103 in the first storage unit 12.

[0056] It is preferable that the first control unit 11 periodically updates the prediction model by performing steps S101 to S104 in response to changes in the climate or season, thereby maintaining the accuracy of prediction of the probability of pest occurrence even when the climate or season changes.

[0057] According to the first operation example, it becomes possible to manage the cultivation facility 60 using the prediction model, thereby making it possible to efficiently grasp the occurrence status of pests and diseases in the cultivation facility 60. Furthermore, as will be described in detail later, by using this prediction model as an initial model to predict the probability of pest and disease occurrence in each cultivation facility 60, it is possible to reduce initial costs.

[0058] A second operation example of the support system 1 according to this embodiment will be described with reference to Fig. 7. The second operation example corresponds to an embodiment of the support method according to the present disclosure.

[0059] The second operation example starts when the second control unit 21 of the client device 20 installed in each cultivation facility 60 acquires the prediction model generated in the first operation example from the server device 10 via the second communication unit 23 and stores it in the second memory unit 22. However, the second operation example does not need to be executed in all cultivation facilities 60 nationwide, and the second operation example may be executed in a specific cultivation facility 60.

[0060] In step S201, the second control unit 21 of the client device 20 acquires data indicating the environment of the cultivation facility 60 from the sensor group 30 at predetermined time intervals and stores the data in the second storage unit 22.

[0061] In step S202, the second control unit 21 performs preprocessing by acquiring data indicating the environment from the second storage unit 22. This preprocessing is similar to step S102.

[0062] In step S203, the second control unit 21 performs clustering on the sensor group 30 using the data preprocessed in step S202. As a result, the sensor group 30 is classified into clusters based on the environmental similarity between areas of the cultivation facility 60. Figure 8 shows the clustering results. In Figure 8, "explanatory variable 1" and "explanatory variable 2" correspond to data indicating the environment, respectively. Furthermore, "t1" and "t2" correspond to their changes over time. Note that if cost is a priority, the cultivation manager may forcibly set the number of clusters via the input unit 24.

[0063] In step S204, the second control unit 21 optimizes the positions and / or number of sensors included in the sensor group 30 based on the clustering results of step S203. The second control unit 21 also outputs the optimized positions and / or number of sensors to the cultivation manager via the output unit 25. In FIGS. 8 and 9, the sensor group 30 is classified into nine clusters (Clusters 1 to 9). The second control unit 21 can optimize the sensors by selecting one of the sensors belonging to the same cluster as a representative and eliminating the other sensors. For example, the representative sensor (hereinafter also referred to as the "representative sensor") may be determined so that the distances between the representative sensors belonging to each cluster are equal. The excluded sensor may be used in another cultivation facility 60 or may be returned to the leasing company.

[0064] According to the second operation example, it is possible to determine the minimum positions and / or number of sensors required to detect differences in the environment of the cultivation facility 60, thereby reducing the costs required for managing the cultivation facility 60. From the viewpoint of improving the accuracy of optimization, steps S201 to S204 may be performed using data indicating the occurrence status of pests and diseases determined based on visual inspection by a cultivation manager or images captured by a fixed camera, etc., in addition to data indicating the environment of the cultivation facility 60.

[0065] A third operation example of the support system 1 according to this embodiment will be described with reference to Fig. 10. The third operation example corresponds to an embodiment of the support method according to the present disclosure.

[0066] The third operation example starts from a state in which the second control unit 21 of the client device 20 installed in each cultivation facility 60 acquires the prediction model generated in the first operation example from the server device 10 via the second communication unit 23 and stores it in the second memory unit 22. However, the third operation example does not need to be executed in all cultivation facilities 60 across the country, and the third operation example may be executed in a specific cultivation facility 60.

[0067] In step S301, the second control unit 21 of the client device 20 acquires data indicating the environment of the cultivation facility 60 from the sensor group 30 at predetermined time intervals and stores the data in the second storage unit 22.

[0068] In step S302, the second control unit 21 acquires data indicating the environment from the second storage unit 22 and performs preprocessing on the data. This preprocessing is similar to step S102. Fig. 11 shows the discretized data.

[0069] In step S303, the second control unit 21 acquires a prediction model from the second memory unit 22 and predicts the current and future occurrence probabilities of pests and diseases. FIG. 12 shows an example of predicting the current occurrence probability of pests and diseases using a prediction model using a Bayesian network and clustering. In this example, the second control unit 21 sets the evidence for the parent node corresponding to the data indicating the environment of the cultivation facility 60 acquired in step S301 to "1," and sets the evidence for other parent nodes to "0." As a result, for example, the current occurrence probability of pests and diseases is predicted to be 85% for the grandchild node corresponding to the presence or absence of occurrence of Botrytis cinerea in variety A.

[0070] In step S304, the second control unit 21 outputs the current and future pest occurrence probabilities predicted in step S303 to the cultivation manager via the output unit 25. An example of the output is shown in FIG. 13. In FIG. 13, the current and future occurrence probabilities of gray mold are shown. However, in step S304, indicators such as "severe," "mild," and "none" are not displayed in FIG. 13. In this example, the occurrence probability is defined as "low" between 0% and 40%, "medium" between 40% and 70%, and "high" between 70% and 100%, but not limited thereto. Here, the second control unit 21 may transmit data indicating the occurrence probabilities of pests to the mobile terminal 50 via the second communication unit 23 and display the data on the screen of the mobile terminal 50. This allows the cultivation manager to visually confirm the actual pest infestation status, prioritizing areas of the cultivation facility 60 with a high current probability of pest infestation, and register the status on the mobile terminal 50. The actual pest infestation status is determined appropriately according to the severity of the infestation level, and is represented by indicators such as "severe," "light," and "none," but is not limited to these. Furthermore, the cultivation manager can include information indicating the location of pest infestation in the information indicating the actual pest infestation status by reading barcodes or QR codes (registered trademark) that indicate location information attached to each area of ​​the cultivation facility 60 with the mobile terminal 50. FIG. 14 shows an example screen of the mobile terminal 50. FIG. 14 shows that the current probability of pest infestation in an area (lane A, number 6) of the cultivation facility 60 is "Botrytis: high, powdery mildew: medium, leaf mold: low," and also shows that the actual pest infestation status is "Botrytis: severe, powdery mildew: mild, leaf mold: absent." In this example, if the cultivation manager is an apprentice, the actual pest infestation status is determined appropriately not by visual confirmation by the cultivation manager, but by analyzing images of plants captured by a camera or the like on the mobile terminal 50 using any technology such as deep learning. The future probability of pest infestation will be described in detail in the fourth operation example.

[0071] In step S305, the second control unit 21 acquires data indicating the actual occurrence status of pests from the mobile terminal 50 via the second communication unit 23.

[0072] In step S306, the second control unit 21 further outputs data indicating the actual occurrence status of pests and diseases acquired in step S305 to the cultivation manager via the output unit 25. Preferably, the second control unit 21 outputs the data indicating the actual occurrence status of pests and diseases acquired in step S305 together with the current occurrence probability of pests and diseases predicted in step S303 to the cultivation manager via the output unit 25. An example of the output is shown on the left side of Figure 13.

[0073] Here, the prediction model generated in the first operation example is a prediction model generated using data indicating the environment of cultivation facilities 60 nationwide and data indicating the occurrence status of pests and diseases, and therefore is not necessarily customized for each cultivation facility 60. For this reason, the current occurrence probability of pests and diseases predicted in step S303 does not necessarily accurately reproduce the actual occurrence status of pests and diseases obtained in step S305. Therefore, it is preferable to improve the prediction accuracy of the occurrence probability of pests and diseases in each cultivation facility 60 by customizing the prediction model generated in the first operation example to each cultivation facility 60.

[0074] Therefore, in step S307, the second control unit 21 determines whether to update the prediction model acquired from the second memory unit 22 in step S303 based on a comparison between the current probability of pest occurrence predicted in step S303 and the actual pest occurrence status acquired in step S305. If the second control unit 21 determines to update the prediction model (step S307: YES), the process proceeds to step S308. If the second control unit 21 determines not to update the prediction model (step S307: NO), the process terminates. Note that since the cultivation manager can visually check the output example of FIG. 13, in step S307, the second control unit 21 may determine whether to update the prediction model based on the cultivation manager's input. For example, the second control unit 21 outputs a question dialog asking whether to update the prediction model to the cultivation manager via the output unit 25. The second control unit 21 also accepts an operation by the cultivation manager to input a response regarding whether to update the prediction model via the input unit 24.

[0075] In step S308, the second control unit 21 adds the data indicating the environment of the cultivation facility 60 acquired in step S301 and the data indicating the actual occurrence status of pests and diseases acquired in step S305 to the existing data, and performs machine learning in the same manner as in steps S102 and S103. When the machine learning by the second control unit 21 is completed, the prediction model is updated. In addition, the second control unit 21 replaces the prediction model stored in the second memory unit 22 with the updated prediction model.

[0076] It should be noted that repeating steps S303 to S308 further improves the prediction accuracy of the probability of pest occurrence in each cultivation facility 60. In addition, if the environment of a specific cultivation facility 60 changes due to changes in climate or season, the prediction accuracy of the prediction model may decrease. Therefore, the second control unit 21 may perform clustering similar to that of step S203 after step S307 and before step S308. However, from the viewpoint of calculation time and amount of calculation, it is preferable that this clustering be performed approximately once every few months to once a year.

[0077] According to the third operation example, the cultivation manager can prioritize management of areas in each cultivation facility 60 where the probability of pests and diseases occurring is high, thereby enabling efficient management of the cultivation facility 60 with less effort. Furthermore, according to the third operation example, a prediction model is generated that takes into account the actual occurrence of pests and diseases in each cultivation facility 60, thereby improving the prediction accuracy of the prediction model for each cultivation facility 60.

[0078] A fourth operation example of the support system 1 according to this embodiment will be described with reference to Fig. 15. The fourth operation example corresponds to an embodiment of the support method according to the present disclosure.

[0079] The fourth operation example starts from a state in which the prediction model customized for each cultivation facility 60 in the third operation example is stored in the second memory unit 22 of the client device 20. However, the fourth operation example does not need to be executed in all cultivation facilities 60 across the country, and the fourth operation example may be executed in a specific cultivation facility 60.

[0080] In step S401, the second control unit 21 of the client device 20 acquires data indicating the environment of the cultivation facility 60 from the sensor group 30 at predetermined time intervals and stores the data in the second storage unit 22.

[0081] In step S402, the second control unit 21 performs preprocessing by acquiring data indicating the environment from the second storage unit 22. This preprocessing is similar to step S102.

[0082] In step S403, the second control unit 21 obtains the prediction model from the second storage unit 22 and predicts the current and future occurrence probabilities of pests. Step S403 is similar to step S303.

[0083] In step S404, the second control unit 21 outputs, via the output unit 25, the current and future occurrence probabilities of the pests predicted in step S403 to the cultivation manager.

[0084] In step S405, the second control unit 21 determines whether to control the environmental control device 40 installed in the cultivation facility 60 based on the input of the cultivation manager. For example, the second control unit 21 outputs a question dialog asking whether to control the environmental control device 40 to the cultivation manager via the output unit 25. The second control unit 21 also accepts an operation by the cultivation manager to input a response as to whether to control the environmental control device 40 via the input unit 24. At this time, the cultivation manager can comprehensively determine the current and future occurrence probabilities of pests and diseases output in step S404. If the second control unit 21 determines to control the environmental control device 40 (step S405: YES), the process proceeds to step S406. If the second control unit 21 determines not to control the environmental control device 40 (step S405: NO), the process proceeds to step S407.

[0085] In step S406, the second control unit 21 acquires a target value for the value indicated by the environmental control device 40 based on the input of the cultivation manager. For example, the second control unit 21 accepts an operation in which the cultivation manager inputs a target value via the input unit 24. At this time, the cultivation manager can determine the target value so as to reduce the probability of pest and disease occurrence. The second control unit 21 also controls the environmental control device 40 so that the value indicated by the environmental control device 40 reaches the acquired target value. The second control unit 21 also predicts the future probability of pest and disease occurrence when the value indicated by the environmental control device 40 reaches the target value. This prediction is performed in a manner similar to step S303, by the second control unit 21 setting the evidence of the parent node corresponding to the target value to "1."

[0086] Here, in the fourth processing example, the name of the pesticide capable of controlling pests and the time of spraying are linked in advance to the type of pest and its occurrence probability based on past performance data, etc., and stored in the second memory unit 22.

[0087] In step S407, the second control unit 21 references the second memory unit 22 and generates a control plan including the name and spraying time of a pesticide for controlling pests and diseases based on the future probability of pests and diseases occurring predicted in step S404. Alternatively, the second control unit 21 references the second memory unit 22 and generates a control plan including the name and spraying time of a pesticide for controlling pests and diseases based on the future probability of pests and diseases occurring when the environmental control device 40 is controlled, predicted in step S406. The second control unit 21 also outputs the generated control plan to the cultivation manager via the output unit 25. The upper part of Figure 16 shows a control plan generated based on the future probability of pests and diseases occurring when the environmental control device 40 is controlled, predicted in step S406. The lower part of Figure 16 shows a control plan generated based on the future probability of pests and diseases occurring when the environmental control device 40 is controlled, predicted in step S404. The control plan in Figure 16 includes, for example, a control plan for when the level of pest infestation in the area where the sensor group (No. 2) is located is "light." If the environmental control device 40 is controlled, the cultivation manager can understand from the control plan in Figure 16 that it is sufficient to spray chemical B (effect: sterilization) within two days. On the other hand, if the environmental control device 40 is not controlled, the cultivation manager can understand from the control plan in Figure 16 that it is necessary to spray chemical B (effect: sterilization) immediately.

[0088] According to the fourth operation example, the environmental control device 40 is controlled to reduce the probability of pests and diseases occurring in each cultivation facility 60, thereby suppressing damage caused by pests and diseases. Furthermore, the cultivation manager can efficiently determine the work schedule and the timing and quantity of pesticide orders based on the pest control plan, thereby enabling efficient management of the cultivation facility 60.

[0089] Although the present disclosure has been described above based on the drawings and embodiments, it should be noted that those skilled in the art can easily make various modifications or alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included in the scope of the present disclosure. For example, the functions included in each step can be rearranged so as not to cause logical inconsistencies, and multiple steps can be combined or divided into one.

[0090] For example, the sensor group 30 shown in FIG. 1 may include a spectral camera capable of measuring the amount of chlorophyll, which is important for photosynthesis. When a plant or the like is stressed by a pest or disease, photosynthesis is inhibited in the stressed area (hereinafter, "stressed area"). Therefore, the area of ​​the stressed area or the time change in the area calculated by analyzing the image captured by the spectral camera based on the reflectance intensity or the normalized vegetation index may be added to the explanatory variables of the prediction model. In FIG. 17, the black area among the white areas (corresponding to "leaves") corresponds to the stressed area. Furthermore, a prediction model for detecting the onset or potential occurrence of a specific pest or disease may be generated based on the reflectance intensity of characteristic wavelengths for each pixel of the captured image using machine learning such as random forests. This improves the accuracy of predicting the probability of pest or disease occurrence. [Industrial Applicability]

[0091] According to the present disclosure, it is possible to provide a support system and a support method that are capable of efficiently understanding the occurrence status of pests and diseases in a cultivation facility. [Explanation of symbols]

[0092] 1. Support System 10 Server device 11 First control section 12 First storage unit 13 First Communications Department 20 Client Device 21 Second control section 22 Second memory unit 23 Second Communications Department 24 Input section 25 Output section 30 sensors 40 Environmental Control Device 50 Mobile Devices 60 Cultivation Facilities 70 Network

Claims

1. a first memory unit that stores data indicating the environment of at least one cultivation facility and data indicating the occurrence status of pests; a first control unit that generates a prediction model for predicting the probability of pests and diseases occurring in the cultivation facility by performing machine learning using the data indicating the environment acquired from the first storage unit as an explanatory variable and the data indicating the pest and disease occurrence status as a target variable; A group of sensors that measure data indicating the environment of the cultivation facility; a second control unit that inputs data indicating the environment acquired from the group of sensors into the prediction model and predicts the probability of occurrence of the pests in the cultivation facility; an output unit that outputs the probability of pest occurrence predicted by the second control unit; Equipped with The first control unit performing preprocessing on the data indicating the environment and the data indicating the occurrence status of pests; performing clustering on a time series basis using the data indicating the environment that has been preprocessed and the data indicating the occurrence status of pests; performing structural learning using the clustering results to generate, as the prediction model, a Bayesian network including a parent node corresponding to data indicating the environment, a child node corresponding to the clustering results, and a child node corresponding to the probability of occurrence of the pest; The second control unit clustering the sensors included in the sensor group based on data indicating the environment of the cultivation facility; optimizing the positions and / or the number of sensors included in the sensor group based on the clustering results; The output unit outputs the optimized positions and / or number of the sensors.

2. 1. A support method using a first computer and a second computer, comprising: a step in which the first computer performs machine learning using data indicating the environment of at least one cultivation facility as an explanatory variable and data indicating the occurrence status of pests and diseases as a target variable to generate a prediction model for predicting the occurrence probability of pests and diseases in the cultivation facility; The second computer clustering sensors included in a group of sensors that measure data indicative of the environment of the cultivation facility based on the data indicative of the environment of the cultivation facility; optimizing the positions and / or the number of sensors included in the sensor group based on the clustering results; outputting the optimized positions and / or numbers of the sensors from an output unit; Including, In the step of generating the prediction model by the first computer, performing preprocessing on the data indicating the environment and the data indicating the occurrence status of pests; performing clustering on a time series basis using the data indicating the environment that has been preprocessed and the data indicating the occurrence status of pests; performing structural learning using the clustering results to generate, as the prediction model, a Bayesian network including a parent node corresponding to the data indicating the environment, a child node corresponding to the clustering results, and a child node corresponding to the probability of occurrence of the pest; How to help.

Images