Information processing device, information processing system, logic determination method, and logic determination program

Abstract

The present invention automatically determine control content corresponding to a target facility, and improve the ease of managing the control content determination logic. An information processing device (1) comprises: a target identifying unit (102) that identifies a target facility; and a logic determining unit (106) that, on the basis of logic allocation information (112) in which common logic (41) and individual logic (42) are allocated to each of a plurality of facilities (61), determines the common logic (41) and the individual logic (42) to be used in the target facility.

Description

Information processing device, information processing system, logic determination method, and logic determination program

[0001] The present invention relates to an information processing system or the like that performs control in accordance with the results of predictions made by a prediction model.

[0002] A technique for determining appropriate control content for various facilities by utilizing prediction results from a prediction model is known as a conventional technique. For example, Patent Document 1 below describes a waste treatment plant facility in which a predicted value of the amount of steam generated is calculated using a learning model that predicts the amount of steam generated after a predetermined time. It also describes controlling a primary combustion air damper by applying an operating value that makes the difference between the predicted amount of steam generated and a target amount of steam generated zero.

[0003] Japanese Patent Application Publication No. 2005-242524

[0004] Although the above-described conventional technology assumes that there is only one control target, in reality, there may be multiple control targets. For example, a waste treatment plant may be equipped with multiple incinerators. Furthermore, when there are multiple incinerators, each incinerator may have different characteristics, so it may not be appropriate to determine the control content for each incinerator using the same decision logic. However, if control content were to be determined using individual decision logics for each incinerator, it would be necessary to manage each of the multiple decision logics individually, which would reduce manageability.

[0005] In view of the above, when there are multiple facilities, a technology is desired that enables automatic determination of appropriate control content according to the target facility to be controlled while improving the manageability of the decision logic of the control content, but such a technology has not existed in the past. One aspect of the present invention aims to realize an information processing device or the like that enables determination of control content according to the target facility while also improving the manageability of the decision logic of the control content.

[0006] In order to solve the above problem, an information processing device according to one embodiment of the present invention includes a target identification unit that identifies equipment to be controlled from among a specified number of pieces of equipment as target equipment, and a logic determination unit that determines, for each of the plurality of pieces of equipment, the common logic that is a determination logic for control content that can be commonly applied to the plurality of pieces of equipment, and the individual logic for determining individual control content for each piece of equipment, based on logic assignment information that assigns the common logic and the individual logic to be used in the target equipment.

[0007] In order to solve the above problem, an information processing system according to one embodiment of the present invention includes an information processing device that determines, based on logic assignment information that assigns, for each of a specified number of pieces of equipment, a common logic that is a determination logic for control content that can be applied to the plurality of pieces of equipment, and an individual logic for determining individual control content for each piece of equipment, the common logic and the individual logic to be used for a target piece of equipment that is to be controlled among the plurality of pieces of equipment, and a control content determination device that determines the control content for the target equipment using the common logic and the individual logic determined by the information processing device.

[0008] In order to solve the above problems, a logic determination method according to one aspect of the present invention is a logic determination method executed by one or more information processing devices, and includes a target identification step of identifying equipment to be controlled from among a specified number of pieces of equipment as target equipment, and a logic determination step of determining, for each of the plurality of pieces of equipment, the common logic, which is a determination logic for control content that can be commonly applied to the plurality of pieces of equipment, and the individual logic for determining individual control content for each piece of equipment, based on logic assignment information that assigns the common logic and the individual logic to be used in the target equipment.

[0009] According to one aspect of the present invention, it is possible to determine control content according to the target facility, and also to improve the manageability of the control content determination logic.

[0010] FIG. 1 is a block diagram showing an example of the configuration of a main part of an information processing device according to an embodiment of the present invention. FIG. 2 is a diagram showing an example of the configuration of an information processing system according to an embodiment of the present invention. FIG. 3 is a diagram showing examples of model allocation information and logic allocation information. FIG. 4 is a diagram showing an example of time series data that is the basis of teacher data. FIG. 5 is a diagram showing examples of pseudo-time series data and teacher data. FIG. 6 is a flowchart showing an example of a model determination method / logic determination method according to an embodiment of the present invention.

[0011] 2 is a diagram showing an example of the configuration of an information processing system 9 according to one embodiment of the present invention. The information processing system 9 is a system having a function of predicting predetermined events related to the control of equipment 61 included in the facility 6 and determining the control content for the equipment 61 based on the prediction results.

[0012] As shown in the figure, the information processing system 9 includes an information processing device 1, a prediction device 2a, a prediction device 2b, a control content determination device 3, a storage device 4, a storage device 5, and a facility 6. When there is no need to distinguish between the prediction device 2a and the prediction device 2b, they will be simply referred to as the "prediction device 2." The same applies to prediction models 51a, 51b, equipment 61a to 61c, individual logic 42a to 42c, and setting information 43a to 43c, which will be described later.

[0013] The information processing device 1 identifies equipment to be controlled among the equipment 61 as target equipment, and determines the prediction model 51 to be used for the above-mentioned prediction of the target equipment. The information processing device 1 also determines decision logic for determining control content for the target equipment. Note that the decision logic defines a method for determining control content, and can also be referred to as control content decision logic. For example, a computer program for determining control content may be used as the decision logic. Furthermore, the above "control content" includes at least one of what type of control to perform and what control amount to apply.

[0014] The prediction device 2 makes the above-mentioned prediction for the target equipment using the prediction model 51 determined by the information processing device 1. Since the prediction model 51 is stored in the storage device 5, the prediction device 2 acquires the prediction model 51 determined by the information processing device 1 from the storage device 5 and makes a prediction using the acquired prediction model 51.

[0015] 2 shows two prediction devices 2, prediction devices 2a and 2b. The prediction device 2a makes predictions using a prediction model 51a that is commonly used for predictions in multiple pieces of equipment 61. On the other hand, the prediction device 2b makes predictions using a prediction model 51b that is used for predictions in a specific piece of equipment 61. The number of prediction devices 2 included in the information processing system 9 is arbitrary.

[0016] The control content determination device 3 determines the control content for the target equipment based on the prediction result by the prediction device 2. The control content is determined using the decision logic determined by the information processing device 1. Since the control content decision logic is stored in the storage device 4, the control content determination device 3 acquires the decision logic determined by the information processing device 1 from the storage device 4 and determines the control content using the acquired decision logic. Note that the information processing device 1 may be configured to acquire the decision logic from the storage device 4 and transfer the acquired decision logic to the control content determination device 3.

[0017] The storage device 4 stores common logic 41, individual logic 42, and setting information 43. Both the common logic 41 and the individual logic 42 are logic for determining control content (which can also be referred to as control content determination logic). As will be described in detail later, the common logic 41 is applicable to multiple pieces of equipment 61 in common, while the individual logic 42 is used to determine individual control content for each piece of equipment 61. The setting information 43 indicates settings (for example, constant values ​​included in a program or formula that determines the control content) to be applied when performing processing to determine the control content using the common logic 41 and the individual logic 42 in each piece of equipment 61.

[0018] A single decision logic is completed by combining the common logic 41 and the individual logic 42 and applying the settings indicated in the setting information 43. Then, by inputting the prediction results of the prediction model 51 into the completed decision logic, control content is output. Such common logic 41 can be generated by generating one decision logic for each piece of equipment 61 (a decision logic adapted to each piece of equipment 61) and then extracting the parts of the created decision logic that are common to each piece of equipment 61. The remaining parts that are not extracted can then be used as the individual logic 42 for each piece of equipment 61.

[0019] Although details will be described later, the allocation of the common logic 41, individual logic 42, and setting information 43 to each piece of equipment 61 is regulated by logic allocation information. Therefore, the information processing device 1 determines the common logic 41, individual logic 42, and setting information 43 to be used in the target equipment based on this logic allocation information.

[0020] The storage device 5 stores the prediction model 51. As will be described in detail later, a prediction model 51 is assigned to each facility 61, and the assignment of the prediction model 51 to each facility 61 is defined by model assignment information. Therefore, the information processing device 1 determines the prediction model 51 to be used for prediction of the target facility based on this model assignment information.

[0021] 2 shows two prediction models 51, namely, prediction models 51a and 51b. The prediction model 51a is a prediction model 51 that is commonly used for predictions about a plurality of pieces of equipment 61. On the other hand, the prediction model 51b is a prediction model 51 that is used for predictions about a specific piece of equipment 61. For example, the prediction model 51a may be used for predictions about the pieces of equipment 61a and 61b, and the prediction model 51b may be used for predictions about the piece of equipment 61c. There is no particular limitation on the number of types of prediction models 51 that can be stored in the storage device 5.

[0022] The facility 6 is a facility that includes multiple pieces of equipment 61 that are the target of control by the information processing system 9. Although FIG. 2 shows three pieces of equipment 61, 61a to 61c, any number of pieces of equipment may be included in one facility 6. The information processing system 9 can also determine the control content of each piece of equipment 61 included in each facility 6 for multiple facilities 6. The equipment 61 may be a single device or apparatus, or may include multiple devices or apparatuses. The pieces of equipment 61 included in the facility 6 may operate independently or may cooperate with each other.

[0023] (Example of Processing) For example, in the information processing system 9, when controlling the equipment 61a, prediction data acquired by the equipment 61a and an equipment ID (IDentification) that is identification information for the equipment 61a are input to the information processing device 1. The prediction data is data used for prediction using the prediction model 51. For example, time-series measurement values ​​measured by a sensor or the like installed in the equipment 61a may be used as the prediction data. Furthermore, the process of transmitting the prediction data and the equipment ID to the information processing device 1 may be performed by, for example, an information processing device (not shown) installed in the facility 6. Furthermore, the user of the information processing system 9 may input these data to the information processing device 1.

[0024] Next, the information processing device 1 determines the prediction model 51 to be used for prediction of the facility 61a, and also determines the prediction device 2 to make the prediction, based on the above-mentioned model allocation information. For example, assume that the prediction model 51 assigned to the facility 61a is the prediction model 51a. In this case, the information processing device 1 determines to use the prediction model 51a for prediction of the facility 61a, and also determines to make the prediction by the prediction device 2a that is capable of making predictions using the prediction model 51a.

[0025] The information processing device 1 also performs predetermined preprocessing on the prediction data of the equipment 61a to generate input data to be input to the prediction model 51a, and transmits the generated input data to the prediction device 2a. The prediction device 2a obtains the prediction model 51a from the storage device 5, inputs the input data received from the information processing device 1 into the obtained prediction model 51a, obtains a prediction result, and notifies the information processing device 1 of the prediction result.

[0026] Furthermore, the information processing device 1 determines the common logic 41, the individual logic 42, and the setting information 43 to be used in determining the control content in the equipment 61a based on the logic allocation information described above. For example, if the common logic 41, the individual logic 42a, and the setting information 43a are assigned to the equipment 61a, the information processing device 1 determines that these determination logics and setting information will be used in determining the control content in the equipment 61a.

[0027] Then, the information processing device 1 notifies the control content determination device 3 of the logic ID indicating the common logic 41, the logic ID indicating the individual logic 42a, the setting information ID indicating the setting information 43a, and the prediction result of the prediction device 2a.

[0028] Next, the control content determination device 3 obtains the common logic 41, the individual logic 42a, and the setting information 43a indicated by the notified logic ID and setting information ID from the storage device 4, and uses these to determine control content according to the prediction result of the prediction device 2a.The control content determination device 3 then notifies the equipment 61a (more precisely, a control device (not shown) that controls the equipment 61a) of the determined control content.In this way, control of the equipment 61a is performed.It is to be noted that one control device may control one equipment 61, or one control device may control multiple equipment 61.

[0029] (Regarding the Information Processing System) As described above, the information processing system 9 includes the information processing device 1, the prediction device 2, and the control content determination device 3. The information processing device 1 determines the prediction model 51 to be used to predict a predetermined event related to the control of a predetermined plurality of pieces of equipment 61, based on model allocation information that assigns, to each of the predetermined plurality of pieces of equipment 61, a prediction model 51 to be used to predict the above-mentioned event in a piece of target equipment to be controlled among the plurality of pieces of equipment 61. The prediction device 2 predicts the above-mentioned event in the target equipment using the prediction model 51 determined by the information processing device 1. The control content determination device 3 determines the control content for the target equipment based on the prediction result by the prediction device 2. Therefore, the information processing system 9 makes it possible to perform state prediction using an appropriate prediction model 51 according to the target equipment, and to determine appropriate control content based on the prediction result.

[0030] The information processing system 9 also includes a storage device 5 that stores a prediction model 51 assigned to each of the multiple pieces of equipment 61, and the prediction device 2 obtains the prediction model 51 determined by the information processing device 1 from the storage device 5 and performs predictions using the prediction model 51. This configuration makes it possible to collectively manage the prediction models 51 used by each piece of equipment 61. Furthermore, by storing a prediction model 51a that is commonly used by multiple pieces of equipment 61 (for example, pieces of equipment 61a and 61b) in the storage device 5, efficient operation can be achieved by reusing the prediction model 51a among those pieces of equipment 61.

[0031] Furthermore, when there are multiple pieces of target equipment and a common prediction model 51 is assigned to the multiple pieces of target equipment, the prediction device 2 may execute prediction processes using the prediction model 51 in parallel. This makes it possible to shorten the time until prediction results for each piece of target equipment are output. The information processing system 9 uses a prediction model 51a that is commonly used for predictions for multiple pieces of equipment 61. Therefore, when a prediction using the prediction model 51a is being performed or is about to be performed for one piece of target equipment, it may be necessary to also perform predictions using the prediction model 51a for other pieces of target equipment. In such cases, the prediction device 2 executes prediction processes using the prediction model 51a in parallel (in other words, simultaneously in parallel), thereby making it possible to quickly obtain prediction results for each piece of target equipment.

[0032] As described above, the information processing system 9 includes the information processing device 1 that determines the common logic 41 and the individual logic 42 to be used for target equipment among the plurality of pieces of equipment 61 to be controlled, based on logic assignment information that assigns, for each of the plurality of pieces of equipment 61, the common logic 41, which is a decision logic for control content applicable to the plurality of pieces of equipment 61 in common, and the individual logic 42 for determining individual control content for each piece of equipment 61, and the control content determination device 3 that determines the control content for the target equipment using the common logic 41 and the individual logic 42 determined by the information processing device 1. Thus, the information processing system 9 makes it possible to determine control content appropriate for the target equipment by combining the common logic 41 and the individual logic 42. Furthermore, since the common logic 41 and the individual logic 42 can each be managed according to their characteristics, it is also possible to improve the manageability of the decision logic.

[0033] As described above, the information processing system 9 includes the storage device 4 that stores the common logic 41 that can be commonly applied to the plurality of pieces of equipment 61, and the individual logic 42 that is applied to each of the plurality of pieces of equipment 61. The control content determination device 3 acquires the common logic 41 and the individual logic 42 determined by the information processing device 1 from the storage device 4. This configuration realizes efficient operation in which the common logic 41 stored in the storage device 4 is reused among the plurality of pieces of equipment 61.

[0034] (Application Example) The information processing system 9 can be applied to the control of various facilities and equipment. For example, the facility 6 may be a plant equipped with a plurality of devices. In this case, each device equipped in the plant is equipment 61. Here, a "plant" refers to a facility used industrially, equipped with a plurality of devices, which performs predetermined processing such as the production of products or the treatment of objects. For example, a waste treatment facility is also included in the category of a plant.

[0035] In the following, an example will be described in which facility 6 is a waste treatment facility that incinerates waste (e.g., combustible waste) and generates electricity using the exhaust heat, and equipment 61a-61c are incinerators. Equipment 61a-61c collects prediction data related to waste incineration and power generation, uses the prediction data to perform predictions using prediction model 51, and controls equipment 61a-61c based on the prediction results. This allows for stable automatic control of the waste treatment facility.

[0036] Furthermore, the subject of prediction by the prediction model 51 may be anything related to the control of the facility 61, and is not particularly limited. For example, the prediction model 51 may be a model that predicts the quality of the combustion state in an incinerator. In order to efficiently generate power in the facility 61, it is necessary to stabilize the amount of steam used to turn the power-generating turbine. However, the combustion state in the incinerator changes with fluctuations in the quality and quantity of the waste to be incinerated, and the amount of steam also changes accordingly. By using the prediction model 51 that predicts the quality of the combustion state, it is possible to predict the quality of the combustion state in the facility 61, which is an incinerator, and appropriately control the facility 61 in accordance with the prediction results, thereby enabling stable incineration of waste and power generation.

[0037] When using the prediction model 51 to predict the quality of the combustion state in an incinerator, the prediction data may be various data related to the combustion state. For example, time-series sensing data related to automatic combustion control (ACC) may be used as the prediction data. By using such data as explanatory variables and a value indicating the quality of the combustion state after a predetermined time has elapsed after the data is obtained as a target variable, the relationship between these data is learned through machine learning, thereby generating the prediction model 51 that predicts the quality of the combustion state after a predetermined time has elapsed.

[0038] Similar prediction data can also be applied when using a prediction model 51 that predicts a numerical value indicating the state of the equipment 61 after a predetermined time. Examples of numerical values ​​indicating the state of the equipment 61 include the amount of steam, the pressure inside the furnace, and the temperature inside the furnace. Such a prediction model 51 can be generated by machine learning the relationship between various data used for prediction as explanatory variables and the numerical value to be predicted as a target variable.

[0039] Furthermore, the prediction model 51 may predict the content of control to be executed on the equipment 61. For example, various data acquired when an operator performs appropriate intervention control on the equipment 61 (or a predetermined time before that) are used as explanatory variables, and a value indicating the content of the intervention control is used as a target variable, and such a prediction model 51 can be generated by machine learning the relationship between these. Then, by using such a prediction model 51, it is possible to predict which of the intervention controls executed by the operator in the past should be executed.

[0040] The machine learning algorithm is not particularly limited. For example, the prediction model 51 may be a neural network prediction model, or a prediction model such as a support vector machine, linear regression, or random forest.

[0041] [Device Configuration] The configuration of the information processing device 1 will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the main parts of the information processing device 1. As shown in the figure, the information processing device 1 includes a control unit 10 that controls each part of the information processing device 1 in an integrated manner, and a storage unit 11 that stores various data used by the information processing device 1. The information processing device 1 also includes a communication unit 12 that enables the information processing device 1 to communicate with other devices, an input unit 13 that accepts input of various data to the information processing device 1, and an output unit 14 that enables the information processing device 1 to output various data.

[0042] The control unit 10 includes a data acquisition unit 101, a target identification unit 102, a model determination unit 103, an input data generation unit 104, a prediction device determination unit 105, a logic determination unit 106, and a management unit 107. The storage unit 11 stores model assignment information 111 and logic assignment information 112.

[0043] The data acquisition unit 101 acquires various data used by the information processing device 1. For example, the data acquisition unit 101 acquires time-series data acquired by the equipment 61 as prediction data to be used for prediction using the prediction model 51. The time-series data may be acquired from a sensor or the like arranged in the equipment 61 via the communication unit 12 or the input unit 13, or may be input by a user of the information processing device 1 via the input unit 13. The data acquisition unit 101 may also acquire process data obtained by performing predetermined preprocessing on the time-series data.

[0044] The target identification unit 102 identifies, as the target equipment, the equipment 61 to be controlled from among a predetermined plurality of pieces of equipment 61. The method for identifying the target equipment may be any method, as long as it is determined in advance. For example, when data necessary for prediction using the prediction model 51 is acquired by the data acquisition unit 101, the target identification unit 102 may identify, as the target equipment, the equipment 61 from which the data is acquired.

[0045] The model determination unit 103 determines, based on the model allocation information 111, the prediction model 51 to be used for predicting an event in the target equipment identified by the target identification unit 102. The model allocation information 111 is information indicating, for each of a plurality of pieces of equipment 61, the allocation of the prediction model 51 to be used for predicting a predetermined event related to the control of the piece of equipment 61.

[0046] The input data generation unit 104 generates input data to be input to the prediction model 51. More specifically, the input data generation unit 104 performs standardization or normalization as preprocessing on the time-series data acquired by the data acquisition unit 101, thereby generating input data that can be input to the prediction model 51. Note that when the data acquisition unit 101 acquires preprocessed process data, the input data generation unit 104 is omitted.

[0047] The prediction device determination unit 105 determines a prediction device 2 that will perform prediction for the target facility from among a plurality of prediction devices 2 that are capable of performing prediction using the prediction model 51. By providing the information processing device 1 with the prediction device determination unit 105, it becomes possible to avoid concentration of the processing load on one prediction device 2. The method for determining the prediction device 2 will be described later.

[0048] The logic determination unit 106 determines the common logic 41 and the individual logic 42 to be used in determining the control content for the target equipment identified by the target identification unit 102, based on the logic assignment information 112. The logic assignment information 112 is information indicating the assignment of the common logic 41 and the individual logic 42 for each of the multiple pieces of equipment 61. Furthermore, the logic determination unit 106 may determine setting information 43 to be used for the target equipment, in addition to the common logic 41 and the individual logic 42 to be used for the target equipment.

[0049] The management unit 107 manages the authority for each piece of data stored in the storage device 4 (for example, authority to add, delete, and change data, authority to access data, etc.). Specifically, the management unit 107 individually sets a user to whom authority for the common logic 41 is to be granted, a user to whom authority for the individual logic 42 is to be granted, and a user to whom authority for the setting information 43 is to be granted. The management unit 107 then identifies the user of the information processing device 1 and allows processing of each piece of data stored in the storage device 4 within the scope of the authority granted to that user.

[0050] As described above, the information processing device 1 includes the target identification unit 102 and the model determination unit 103. The target identification unit 102 identifies, as target equipment, equipment 61 to be controlled from among a predetermined plurality of pieces of equipment 61. The model determination unit 103 determines, based on model allocation information 111 that assigns, to each of the plurality of pieces of equipment 61, a prediction model 51 to be used for predicting a predetermined event related to the control of the piece of equipment 61. Thus, it is possible to determine an appropriate prediction model 51 according to the target equipment.

[0051] In the model allocation information 111, a common prediction model 51 may be assigned to a plurality of pieces of equipment 61. This allows the prediction model 51 to be used by a plurality of pieces of equipment 61 in a common manner. This makes it possible to simplify the system configuration related to prediction and reduce costs for generating and updating the prediction model 51, compared to when different prediction models 51 are used for each piece of equipment 61.

[0052] As described above, the information processing device 1 includes a target identification unit 102 that identifies, as target equipment, equipment 61 to be controlled from among a specified number of pieces of equipment 61, and a logic determination unit 106 that determines, for each of the pieces of equipment 61, the common logic 41 that is a determination logic for control content that can be commonly applied to the plurality of pieces of equipment 61 and the individual logic 42 for determining individual control content for each piece of equipment 61, based on logic assignment information 112 that assigns, to each piece of equipment 61, the common logic 41 and the individual logic 42 to be used to determine the control content for the target equipment.

[0053] According to the above configuration, the control content for a target facility can be determined using the common logic 41 and the individual logic 42 that correspond to the target facility. Furthermore, according to the above configuration, the common logic 41 and the individual logic 42 can be managed according to their respective characteristics, thereby improving the manageability of the decision logic. For example, according to the above configuration, by simply modifying one common logic 41, the modification can be reflected in the determination of the control content for multiple facilities 61. Furthermore, for example, it is possible to grant only certain people the authority to modify the common logic 41 that affects multiple facilities 61, and grant more people the authority to modify the individual logic 42 that affects only a specific facility 61.

[0054] As described above, the logic assignment information 112 may assign setting information 43 indicating settings to be applied when performing processing to determine control content using the common logic 41 and the individual logic 42 in each piece of equipment 61 to each piece of equipment 61. Then, based on the logic assignment information 112, the logic determination unit 106 may determine the setting information 43 to be used in the target equipment in addition to the common logic 41 and individual logic 42 to be used in the target equipment.

[0055] According to this configuration, the setting information 43, which indicates the settings to be applied when performing a process of determining control content using the common logic 41 and the individual logic 42, is separate data from the common logic 41 and the individual logic 42. This improves manageability compared to a case where the determination logic and the setting information are inseparably integrated. For example, it may be possible to allow even someone who is not familiar with the determination logic of the control content to modify the setting information 43.

[0056] As described above, the information processing device 1 also includes the management unit 107 that individually sets the persons to whom authority regarding the common logic 41 is to be granted, the persons to whom authority regarding the individual logic 42 is to be granted, and the persons to whom authority regarding the setting information 43 is to be granted. This configuration makes it possible to achieve the above-described improvement in manageability.

[0057] For example, the management unit 107 can grant only certain individuals the authority to modify both the common logic 41 and the individual logic 42, while granting other individuals the authority to modify the individual logic 42. This makes it possible to easily modify the individual logic 42 in accordance with the actual conditions of each piece of equipment 61, while reducing the possibility of problems affecting multiple pieces of equipment 61 due to modifications to the common logic 41.

[0058] [Examples of Model Allocation Information and Logic Allocation Information] The model allocation information 111 and the logic allocation information 112 may be, for example, information as shown in Fig. 3. Fig. 3 is a diagram showing examples of the model allocation information 111 and the logic allocation information 112.

[0059] 3 is information in which an equipment ID, which is identification information of the equipment 61, a model ID, which is identification information of the prediction model 51, and a prediction device ID, which is identification information of the prediction device 2, are associated with each other. By using such model allocation information 111, the model determination unit 103 can determine the prediction model 51 to be used for prediction of the target equipment identified by the target identification unit 102.

[0060] For example, assume that the equipment ID of the target equipment identified by the target identification unit 102 is A0002. In this case, the model determination unit 103 determines the prediction model 51 associated with A0002 and having a model ID of B0001 as the prediction model 51 to be used for prediction of A0002. In this case, the prediction device determination unit 105 determines that the prediction device 2 associated with A0002 and having a prediction device ID of C0001 will perform prediction using the prediction model 51 with a model ID of B0001.

[0061] 3 is information in which an equipment ID, which is identification information of the equipment 61, a logic ID, which is identification information of the common logic 41 or the individual logic 42, and a setting information ID, which is identification information of the setting information 43, are associated with each other. By using such logic assignment information 112, the logic determination unit 106 can determine the common logic 41 and the individual logic 42, and the setting information 43, to be used in determining the control content for the target equipment identified by the target identification unit 102.

[0062] For example, assume that the equipment ID of the target equipment identified by the target identification unit 102 is A0003. In this case, the logic determination unit 106 determines that the common logic 41 with a logic ID of D1001 and the individual logic 42 with a logic ID of D2001, which are associated with A0003, will be used to determine the control content for A0003. In addition, in this case, the logic determination unit 106 determines that the setting information 43 with a setting information ID of E0003 will be applied when determining the control content using the common logic 41 with a logic ID of D1001 and the individual logic 42 with a logic ID of D2001.

[0063] [Method of Determining a Prediction Device] As described above, if the allocation of the prediction devices 2 is specified in the model allocation information 111, the prediction device determination unit 105 can determine the prediction device 2 that will perform prediction by referring to the model allocation information 111. In this case, the model allocation information 111 also serves as prediction device allocation information that indicates the allocation of the prediction devices 2. The prediction device allocation information may be data separate from the model allocation information 111.

[0064] Furthermore, the prediction device determination unit 105 may determine the prediction device 2 to perform prediction so that the periods in which predictions are performed by the prediction devices 2 do not overlap or the overlapping periods are minimized. For example, a plurality of prediction devices 2 that can perform predictions using a prediction model 51a that is commonly used for a plurality of pieces of equipment 61 may be provided. In this case, the prediction device determination unit 105 may determine, from among the plurality of prediction devices 2, one that is not performing prediction at that time as the prediction device 2 to perform prediction.

[0065] [Regarding Prediction Models Assigned to Multiple Facilities] The prediction models 51 assigned to multiple facilities 61 may be models generated by machine learning using training data generated by performing standardization or normalization processing on pseudo-time series data obtained by concatenating time series data related to the event to be predicted.

[0066] By performing machine learning using the above-described training data, the characteristics of the time-series data for each facility 61 are learned, thereby generating a prediction model 51 that can output valid prediction results for each facility 61. Therefore, with the above-described configuration, it is possible to standardize the prediction model 51 used by multiple facilities 61 while also ensuring the validity of the prediction results.

[0067] The above-mentioned teacher data will be described below with reference to Fig. 4 and Fig. 5. Fig. 4 is a diagram showing an example of time series data that is the basis of the teacher data, and Fig. 5 is a diagram showing an example of pseudo-time series data generated from the time series data shown in Fig. 4 and teacher data generated from the pseudo-time series data.

[0068] 4 shows an example in which time-series data 114a to 114c used to generate teacher data are generated from steam volume data 113a to 113c, each of which includes a measurement value of the steam volume measured by a steam volume sensor that measures the volume of steam generated from a boiler provided in facility 61. Note that the steam volume data 113a to 113c indicate the steam volumes measured in facilities 61a to 61c in FIG. 2, respectively.

[0069] As shown in Fig. 4, the steam volume data 113a to 113c include measured values ​​(PV: Process Variable) for each time. The steam volume data 113a to 113c shown in Fig. 4 also include set values ​​corresponding to each measured value. The set values ​​(SV: Set Variables) indicate target values ​​(steam volume in this example) at the times associated with the set values. In other words, in the equipment 61a to 61c from which the steam volume data 113a to 113c are obtained, control is performed so that the steam volume approaches the set value.

[0070] From such steam volume data 113a-113c, time series data 114a-114c can be generated, each of which has as its element the PV / SV value at each time, i.e., the ratio of the measured value to the set value. When the PV / SV value is close to 1, the facility condition can be said to be normal. Note that the difference between PV and SV, i.e., the difference between the measured value and the set value, may be used instead of PV / SV.

[0071] The time-series data 114a to 114c also include an element called an abnormality flag. The value of the abnormality flag at each time indicates whether the state of the equipment 61a to 61c is normal or not. Specifically, in the example of Fig. 4, the value of the abnormality flag is set to 0 if the state is normal, and set to 1 if the state is not normal. The criteria for determining whether the state is normal or not can be determined as appropriate.

[0072] The value of the abnormality flag is correct data in machine learning, in other words, the objective variable of the generated prediction model 51. In other words, the prediction model 51 generated using the time-series data 114a to 114c is a model that predicts whether or not the state of the equipment 61 is normal. Of course, the objective variable of the prediction model 51 is arbitrary and is not limited to whether or not the state of the equipment 61 is normal.

[0073] The difference or ratio between a measured value and a set value is a more versatile index than the measured value measured at the equipment 61. Therefore, by using such a versatile index, the pseudo-time series data can be made more natural. In other words, the above configuration can reduce the variation in data caused by differences in the equipment 61. Furthermore, by linking the data collected for each of the multiple pieces of equipment 61, the bias that affects the time-series fluctuations can be reduced. Therefore, the above configuration can generate a prediction model 51 with high prediction accuracy.

[0074] FIG. 5 shows pseudo-time series data 115 generated from the time series data 114a to 114c shown in FIG. 4, and teacher data 116 generated from the pseudo-time series data 115.

[0075] The pseudo-time series data 115 is generated by concatenating the time series data 114a to 114c in this order. However, simply concatenating the time series data 114a to 114c as is will result in overlaps and inconsistencies in the "time" values.

[0076] Therefore, when generating the pseudo-time series data 115, the time values ​​in the time series data 114a to 114c are replaced with consecutive values ​​from 1 to 15. These values ​​indicate the order of each data element included in the time series data 114a to 114c, and can be called order information. By providing new order information in place of the original "time" values, it is possible to prevent inconsistencies in the "time" values ​​while maintaining the continuity of the data.

[0077] Generally, when creating a predictive model using multiple variables with different data distributions, a highly accurate predictive model can be created by standardizing or normalizing the variables. Here, performing the standardization or normalization process requires statistics for the target data set, but these statistics are specific to the data set. Therefore, with conventional techniques, it is impossible to combine time-series data after standardization or normalization into a single data set, or even if it were possible to combine the data, learning using such data significantly reduces the performance of the inference model.

[0078] In contrast, with the above configuration, multiple time series data are concatenated in advance to form a single pseudo-time series data, so that standardization or normalization processing and concatenation processing of multiple time series data can be performed simultaneously.

[0079] 5, the teacher data 116 is generated by performing a standardization process on the pseudo-time series data 115. Specifically, the teacher data 116 is generated by performing a standardization process on each of three data elements included in the pseudo-time series data 115, namely, the measured value, the set value, and the PV / SV.

[0080] Specifically, the difference between the value of a data element included in the pseudo-time series data 115 and the average value is divided by the standard deviation, and this process is performed for each data element to generate the teacher data 116. For example, in the pseudo-time series data 115, the average value of the measured values ​​is 9.387 and the standard deviation is 0.519. Therefore, the measured value of 9.8 at time = 1 in the pseudo-time series data 115 is standardized to a value of (9.8 - 9.387) / 0.519 = 0.795.

[0081] Alternatively, training data may be generated by performing a normalization process instead of the standardization process. In this case, the maximum and minimum values ​​of the data elements included in the pseudo-time-series data 115 are calculated. Then, the difference between the value of the data element and the calculated minimum value is divided by the data range, i.e., the difference between the maximum and minimum values, to generate training data for each data element.

[0082] As described above, by generating the teacher data 116 by performing the standardization process or normalization process on the pseudo-time-series data 115, it is possible to generate a prediction model 51 that absorbs the characteristics of the time-series data for each facility 61. Note that the conditions for the standardization process or normalization process (statistics such as the mean, standard deviation, maximum value, and minimum value) are also used when making predictions using the generated prediction model 51.

[0083] A prediction model 51 is generated by performing machine learning using the teacher data 116 generated as described above. The objective variable of the prediction model 51 may be the value of the anomaly flag. The explanatory variables of the prediction model 51 may be at least one of the measured values, the set value, and the PV / SV values. Values ​​other than these may also be included in the explanatory variables. For example, at least one of the measured values, the set value, and the PV / SV values ​​of the temperature time series may also be included in the explanatory variables. At least one of the values ​​obtained by standardizing or normalizing these values ​​may also be included in the explanatory variables.

[0084] By using training data 116 generated based on pseudo-time-series data 115 obtained by concatenating time-series data 114a-114c collected from multiple facilities 61, a more versatile prediction model 51 can be generated compared to when time-series data from a single source is used as training data. For example, the prediction model 51a may be generated using training data generated from time-series data originating from facilities 61a and 61b. This allows the prediction model 51a to perform highly accurate predictions for both facilities 61a and 61b.

[0085] Furthermore, the prediction model 51a generated in this manner can also be used for predictions in facilities 61 other than the facilities 61a and 61b, as long as they are the same type of facility. In other words, the prediction model 51a generated in this manner can also be used for predictions in facilities 61 that do not provide the time-series data that is the basis of the teacher data 116. Note that the time-series data to be linked is not limited to data collected in multiple different facilities 61. For example, the prediction model 51 can also be generated using teacher data 116 that links time-series data collected for multiple devices in one facility 61.

[0086] When using the prediction model 51 generated using the above-described teacher data 116, the data acquisition unit 101 may acquire time-series data related to an event to be predicted in the target facility (for example, time-series steam volume data, time-series temperature data, or time-series pressure data). Then, the input data generation unit 104 may generate input data that can be input to the prediction model 51 by performing standardization or normalization on the acquired time-series data as preprocessing.

[0087] Furthermore, the input data generation unit 104 may generate input data by performing common preprocessing for target equipment to which the assigned prediction model 51 is commonly assigned. This makes it possible to efficiently perform preprocessing when making predictions for target equipment to which the assigned prediction model 51 is commonly assigned.

[0088] For example, assume that a prediction model 51a commonly used for the facilities 61a and 61b shown in Fig. 2 is generated using the teacher data 116. In this case, the input data generation unit 104 can share the preprocessing of the time-series data, which is the prediction data acquired from the facility 61a, with the preprocessing of the time-series data, which is the prediction data acquired from the facility 61b.

[0089] [Flow of Processing Executed by Information Processing Apparatus 1] The flow of processing executed by the information processing apparatus 1 will be described with reference to Fig. 6. Fig. 6 is a flowchart showing an example of processing (model determination method / logic determination method) executed by the information processing apparatus 1.

[0090] In S1, the data acquiring unit 101 acquires prediction data. The prediction data acquired in S1 may include, for example, time-series data measured in any of the equipment 61 included in the facility 6. In this case, the data acquiring unit 101 may also acquire identification information (for example, the equipment ID shown in FIG. 3 ) of the equipment 61 from which the time-series data was measured.

[0091] In S2 (target identification step), the target identification unit 102 identifies, as target equipment, equipment 61 to be controlled from among a predetermined plurality of pieces of equipment 61. For example, when identification information of equipment 61a is acquired in S1, the target identification unit 102 identifies equipment 61a as target equipment.

[0092] In S3 (model determination step), the model determination unit 103 determines, based on the model allocation information 111, the prediction model 51 to be used for predicting a predetermined event in the target facility identified in S2.

[0093] In S4, the prediction device determination unit 105 determines a prediction device 2 to perform prediction for the target equipment from among a plurality of prediction devices 2 that are capable of performing prediction using the prediction model 51 determined in S3.

[0094] In S5, the input data generation unit 104 performs preprocessing on the prediction data acquired in S1 to generate input data that can be input to the prediction model 51 determined in S3. Note that preprocessed process data may be acquired as prediction data in S1, and in this case, the processing in S5 is omitted.

[0095] In S6, the input data generation unit 104 notifies the prediction device 2 determined in S4 of the prediction model 51 determined in S3, sends the input data generated in S5, causes the prediction device 2 to perform a prediction using the prediction model 51, and receives the prediction result from the prediction device 2.

[0096] In S7 (logic determination step), the logic determination unit 106 determines the common logic 41 and the individual logic 42 to be used in the target equipment identified in S2 based on the logic assignment information 112. The logic determination unit 106 also determines the setting information 43 to be applied when using these determined logics. Note that the processing of S7 can be performed at any timing after S2, and is not limited to the example shown in the figure.

[0097] In S8, the logic determination unit 106 notifies the control content determination device 3 of the content determined in S7 and the prediction result received in S6, thereby completing the processing in FIG.

[0098] The process of FIG. 6 described above includes the model determination method according to this embodiment. That is, the model determination method according to this embodiment includes a target identification step (S2) and a model determination step (S3). In the target identification step (S2), a piece of equipment 61 to be controlled is identified as the target equipment from among a predetermined number of pieces of equipment 61. In the model determination step (S3), a prediction model 51 to be used for predicting an event in the target equipment is determined based on model assignment information 111 that assigns, to each of the plurality of pieces of equipment 61, a prediction model 51 to be used for predicting a predetermined event related to the control of that piece of equipment 61. Thus, it is possible to determine an appropriate prediction model 51 according to the target equipment.

[0099] 6 described above also includes a logic determination method according to this embodiment. That is, the logic determination method according to this embodiment includes a target identification step (S2) of identifying, as target equipment, equipment 61 to be controlled from among a predetermined plurality of pieces of equipment 61, and a logic determination step (S7) of determining, for each of the plurality of pieces of equipment 61, common logic 41, which is determination logic for control content commonly applicable to the plurality of pieces of equipment 61, and individual logic 42, which is used to determine individual control content for each piece of equipment 61, based on logic assignment information 112. This makes it possible to determine control content appropriate for the target equipment and also improves the manageability of the determination logic.

[0100] [Modifications] The execution entity of each process described in each of the above embodiments is arbitrary and is not limited to the above examples. In other words, as long as the execution entity of each process described in each of the above embodiments can be executed, the devices constituting the information processing system 9 can be changed as appropriate.

[0101] For example, in the information processing system 9 of FIG. 2 , the information processing device 1 and the control content determination device 3 are separate devices, but they may be integrated into a single device. Similarly, the storage device 4 and the storage device 5 may be integrated into a single storage device. Alternatively, for example, the information processing device 1 may have the functions of the prediction device 2a. In this case, the information processing device 1 performs predictions using the prediction model 51a itself, and has the prediction device 2b perform predictions using the prediction model 51b. Furthermore, the information processing device 1 determines both the prediction model 51 and the common logic 41 and the individual logic 42. However, it is also possible to provide an information processing device that determines the prediction model 51 and an information processing device that determines the common logic 41 and the individual logic 42.

[0102] 6 (model determination method / logic determination method) does not necessarily have to be performed by a single device, but can be performed by a plurality of arbitrary information processing devices (computers). In other words, the above-described model determination method and logic determination method can be performed by a single information processing device 1, or by a plurality of information processing devices.

[0103] [Example of implementation by software] The functions of the information processing device 1 can be realized by a program (model determination program / logic determination program) that causes a computer to function as the information processing device 1, and that causes a computer to function as each control block of the information processing device 1 (particularly each part included in the control unit 10).

[0104] In this case, the information processing device 1 includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The functions described in the above embodiment are realized by executing the program using the control device and storage device.

[0105] The program may be stored non-transitory on one or more computer-readable recording media. The recording media may or may not be included in the information processing device 1. In the latter case, the program may be supplied to the information processing device 1 via any wired or wireless transmission medium.

[0106] Furthermore, some or all of the functions of the control blocks can be realized by logic circuits. For example, an integrated circuit in which a logic circuit that functions as each of the control blocks is formed is also included in the scope of the present invention. In addition, the functions of the control blocks can also be realized by, for example, a quantum computer.

[0107] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

[0108] REFERENCE SIGNS LIST 1 Information processing device 102 Target identification unit 103 Model determination unit 104 Input data generation unit 105 Prediction device determination unit 106 Logic determination unit 107 Management unit 2 Prediction device 3 Control content determination device 4, 5 Storage device 61 Facility 9 Information processing system

Claims

1. An information processing device comprising: a target identification unit that identifies equipment to be controlled from among a specified number of pieces of equipment as target equipment; and a logic determination unit that determines the common logic and individual logic to be used in the target equipment based on logic assignment information that assigns, for each of the multiple pieces of equipment, a common logic that is a determination logic for control content that can be commonly applied to the multiple pieces of equipment, and an individual logic for determining individual control content for each piece of equipment.

2. The information processing device of claim 1, wherein the logic assignment information assigns setting information indicating settings to be applied when performing processing to determine control content in each piece of equipment using the common logic and the individual logic, and the logic determination unit determines the setting information to be used in the target equipment, in addition to the common logic and individual logic to be used in the target equipment, based on the logic assignment information.

3. The information processing device according to claim 2, further comprising a management unit that individually sets the persons to whom authority regarding the common logic is to be given, the persons to whom authority regarding the individual logic is to be given, and the persons to whom authority regarding the setting information is to be given.

4. An information processing system including: an information processing device that determines, based on logic assignment information that assigns, for each of a specified number of pieces of equipment, a common logic that is a decision logic for control content that can be commonly applied to the plurality of pieces of equipment, and an individual logic for deciding individual control content for each piece of equipment, the common logic and the individual logic to be used in a target piece of equipment to be controlled among the plurality of pieces of equipment; and a control content determination device that determines the control content for the target equipment using the common logic and the individual logic determined by the information processing device.

5. An information processing system as described in claim 4, further comprising a storage device that stores the common logic applicable to a plurality of the facilities and the individual logic applied to each of the plurality of the facilities, and the control content determination device acquires the common logic and the individual logic determined by the information processing device from the storage device.

6. A logic determination method executed by one or more information processing devices, comprising: a target identification step of identifying equipment to be controlled from among a specified number of pieces of equipment as target equipment; and a logic determination step of determining the common logic and individual logic to be used in the target equipment based on logic assignment information that assigns, for each of the multiple pieces of equipment, a common logic that is a determination logic for control content commonly applicable to the multiple pieces of equipment, and an individual logic for determining individual control content for each piece of equipment.

7. A logic determination program for causing a computer to function as the information processing device according to claim 1, the logic determination program causing a computer to function as the target identification unit and the logic determination unit.

Images