Information processing device, information processing method, and program

Abstract

To provide technology that can provide justification for driving plans. [Solution] An information processing device according to one aspect of the present disclosure includes: a selection unit that selects a planned value to be the subject of justification presentation from among the planned values ​​included in the operation plan of the target equipment; a calculation unit that calculates the change in the value of the subject of justification presentation for each of the plurality of time series data used in calculating the operation plan, when at least the value at the same time as the subject of justification presentation is set to the value at the previous time; and a presentation unit that presents justification information representing the justification for the subject of justification presentation using the change in the value of the subject of justification presentation.

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus, an information processing method, and a program.

Background Art

[0002] There is known a technique for determining an optimal operation plan for facilities such as various plants (e.g., power generation plants, steel plants, chemical plants, oil plants, energy plants, etc.) by an optimization method. For example, Patent Document 1 discloses a technique for calculating an optimal operation plan for a heat source system using an optimization method.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, since the basis of the operation plan obtained by using the optimization method is unclear, optimal operation may not be achieved because it is not adopted when it contradicts the user's intuition or rules of thumb.

[0005] The present disclosure has been made in view of the above points, and an object thereof is to provide a technique capable of presenting the basis for an operation plan.

Means for Solving the Problems

[0006] An information processing device according to one aspect of the present disclosure includes: a selection unit that selects a planned value to be used as the basis for providing justification from among the planned values ​​included in the operation plan of the target equipment; a calculation unit that calculates the change in the value of the basis for providing justification for each of the multiple time series data used in calculating the operation plan, by setting the value of at least the value at the same time as the basis for providing justification from among the values ​​included in the time series data to the value at the previous time; and a presentation unit that presents justification information representing the basis for the basis for providing justification using the change in the value of the basis for providing justification. [Effects of the Invention]

[0007] They can provide justification for their driving plan. [Brief explanation of the drawing]

[0008] [Figure 1] This diagram shows an example of the configuration of the target equipment. [Figure 2] This figure shows an example of the hardware configuration of a evidence presentation device according to one embodiment. [Figure 3] This figure shows an example of the functional configuration of a evidence presentation device according to one embodiment. [Figure 4] This flowchart shows an example of pretreatment according to one embodiment. [Figure 5] A flowchart illustrating an example of the evidence presentation process according to one embodiment. [Figure 6] This figure shows an example of prediction data converted into an array format. [Figure 7] This figure shows an example of optimal design data converted into an array format. [Figure 8] This flowchart shows an example of a contribution calculation process according to one embodiment. [Figure 9] This figure shows an example of evidence information represented in a causal diagram. [Modes for carrying out the invention]

[0009] One embodiment of the present invention will be described in detail below with reference to the drawings.

[0010] <Background and Conventional Technology> Generally, facilities such as plants are operated in a way that ensures their state (e.g., the temperature of the furnace in the facility reaches a predetermined temperature) is maintained based on the operating environment and the operator's knowledge. Examples of plants include power plants, steel plants, chemical plants, oil plants, and energy plants.

[0011] Traditionally, a technology called logic processing has been known and used in operational settings to automate or support plant operation and control.

[0012] However, it is difficult to find a mathematically exact solution (i.e., the best driving method) using logic processing. For this reason, for example, if the quality of a driving method is evaluated by its cost and CO2 emissions, it is difficult to achieve driving that minimizes both cost and CO2 emissions using logic processing.

[0013] For example, consider a plant consisting of one gas turbine, one boiler, one turbo chiller, and two steam absorption chillers, and determine the operation plan for 24 hours. In this case, it is necessary to determine the values ​​of 5 × 24 = 120 discrete variables (start: 1, stop: 0), so 2 120 There are established operating patterns (Reference 1). Therefore, it is impossible to determine the best operating method through logic processing.

[0014] On the other hand, in recent years, techniques have been used in plant operations to determine the optimal operation plan by treating the plant as a mathematical model, defining objective functions and constraints, and solving optimization problems using optimization methods. Since such techniques can obtain mathematically exact solutions, it is expected that the best operation that minimizes the objective function value (for example, operation that minimizes costs and CO2 emissions) can be achieved. An example of a technique that can determine the optimal operation plan using optimization methods is the technique disclosed in Patent Document 1.

[0015] <Problems with conventional technology> However, technologies that use optimization methods to determine the optimal operating plan have a problem in that the basis for that operating plan (for example, the basis for starting or stopping equipment or devices that make up the plant) is unclear. Therefore, the user may not adopt the operating plan based on their own judgment, and optimal operation may not be achieved. This basis may also be called, for example, "reasoning."

[0016] Therefore, in order to solve the above problems, the following describes a justification presentation device 10 that can present the justification for the operation plan obtained using an optimization method. By using the justification presentation device 10 described below, users (e.g., operators of equipment such as plants) will be able to trust the operation plan obtained using an optimization method and achieve optimal operation in accordance with that operation plan.

[0017] The evidence presentation device 10 described below can be implemented using, for example, an information processing device (computer) such as a PC (Personal Computer) or a general-purpose server, or an information processing system (computer system) composed of these information processing devices.

[0018] <Example of equipment configuration> Below, we will consider the energy plant shown in Figure 1 as an example of equipment for which an optimal operation plan will be created using an optimization method (hereinafter also referred to as "target equipment"). The energy plant shown in Figure 1 consists of a gas turbine (+ exhaust gas boiler), a once-through boiler, N turbo chillers, and N steam absorption chillers. The gas turbine (+ exhaust gas boiler) consists of a gas turbine that generates electricity from city gas and an exhaust gas boiler that outputs steam from the gas discharged from the gas turbine. The once-through boiler outputs steam from city gas. The turbo chillers output cooling using electricity generated by the gas turbine and electricity purchased as needed. The steam absorption chillers output cooling using steam output from the exhaust gas boiler and steam output from the once-through boiler. The electricity, cooling, and steam output from the energy plant are used as power load, heat load, and steam load, respectively.

[0019] Note that the configuration of the energy plant shown in Figure 1 is just one example and is not limited to this configuration.

[0020] <Example of hardware configuration for evidence presentation device 10> Figure 2 shows an example of the hardware configuration of the evidence presentation device 10 according to one embodiment. As shown in Figure 2, the evidence presentation device 10 according to one embodiment includes an input device 101, a display device 102, an external I / F 103, a communication I / F 104, a RAM (Random Access Memory) 105, a ROM (Read Only Memory) 106, an auxiliary storage device 107, and a processor 108. Each of these hardware components is connected to each other via a bus 109 so as to be able to communicate.

[0021] The input device 101 is, for example, a keyboard, mouse, touch panel, or physical button. The display device 102 is, for example, a display or display panel. The evidence presentation device 10 does not necessarily have to have at least one of the input device 101 and the display device 102.

[0022] External I / F 103 is an interface with external devices such as recording media 103a. Examples of recording media 103a include CD (Compact Disc), DVD (Digital Versatile Disk), SD memory card (Secure Digital memory card), and USB (Universal Serial Bus) memory card.

[0023] The communication interface 104 is an interface for connecting to a communication network. The RAM 105 is a volatile semiconductor memory (storage device) that temporarily holds programs and data. The ROM 106 is a non-volatile semiconductor memory (storage device) that can retain programs and data even when the power is turned off. The auxiliary storage device 107 is a non-volatile storage device such as an HDD (Hard Disk Drive), SSD (Solid State Drive), or flash memory. The processor 108 is a processing unit such as a CPU (Central Processing Unit) or GPU (Graphics Processing Unit).

[0024] Note that the hardware configuration shown in Figure 2 is just one example, and the hardware configuration of the evidence presentation device 10 is not limited to this. The evidence presentation device 10 may, for example, have multiple auxiliary storage devices 107 or multiple processors 108, or it may not have some of the hardware shown, or it may have various hardware other than the hardware shown.

[0025] <Example of Functional Configuration of Evidence Presentation Device 10> Figure 3 is a diagram showing an example of the functional configuration of the evidence presentation device 10 according to one embodiment. As shown in Figure 3, the evidence presentation device 10 according to one embodiment includes a collection unit 110, a prediction unit 111, an optimal plan calculation unit 112, an acquisition unit 113, an information organization unit 114, a target selection unit 115, a contribution calculation unit 116, and an evidence presentation unit 117. Each of these units is realized, for example, by processing that one or more programs installed in the evidence presentation device 10 are executed by a processor 108 or the like. The evidence presentation device 10 according to one embodiment also has a database 118. The database 118 is realized, for example, by a storage area such as an auxiliary storage device 107. The database 118 may also be realized by a storage area such as a storage device (e.g., a storage device of a database server) that is communicably connected to the evidence presentation device 10.

[0026] The data collection unit 110 collects characteristic information, operating information, and sensor information related to the target equipment, as well as information used to predict future values ​​of the characteristic information and sensor information (hereinafter also referred to as "predictive information"). The data collection unit 110 also stores this characteristic information, operating information, sensor information, and predictive information in the database 118.

[0027] Characteristic information refers to information representing the characteristics of the equipment or devices constituting the target facility (e.g., output characteristics), information representing the characteristics of the energy used in the target facility (e.g., energy purchase price), etc. Operational information refers to information representing events of the target facility (e.g., operating days, inspection days, shutdown periods, etc.), information representing the normal or abnormal range of sensor values ​​collected from the equipment or devices constituting the target facility, etc. Sensor information refers to information representing sensor values, which are values ​​measured by sensors regarding the state of the equipment or devices constituting the target facility (e.g., load), input, output, etc.

[0028] The collection unit 110 may collect characteristic information, operating information, sensor information, and prediction information from any information source (e.g., sensors installed on equipment or devices constituting the target equipment, control devices that control the operation of the target equipment, etc.). In addition, all or part of the characteristic information, operating information, sensor information, and prediction information may be acquired sequentially in real time, or may be input from an input device 101, etc.

[0029] Here, as an example, if we consider the energy plant shown in Figure 1 as the target facility, examples of characteristic information, operational information, sensor information, and predictive information include the following:

[0030] Characteristic information: Electricity purchase price, city gas purchase price, upper and lower output limits and output efficiency of each component of the target facility (gas turbine (+ exhaust gas boiler), once-through boiler, each turbo chiller, each steam absorption chiller). Operational Information: Shutdown periods for gas turbines (+ exhaust gas boilers), once-through boilers, turbo chillers, and steam absorption chillers. Sensor information: Electricity purchase amount, city gas purchase amount, city gas consumption of gas turbine (+ exhaust gas boiler), power output of gas turbine (+ exhaust gas boiler), steam output of gas turbine (+ exhaust gas boiler), electricity consumption of turbo chiller, heat output of turbo chiller, steam consumption of steam absorption chiller, heat output of steam absorption chiller, city gas consumption of once-through boiler, steam output of once-through boiler, power load, heat load, steam load, surplus steam Predictive information: Weather information (e.g., weather, temperature, humidity, etc.)

[0031] The prediction unit 111 uses characteristic information, operating information, sensor information, and prediction information stored in the database 118, a specified period representing a future period specified by the user, and a pre-created prediction model to predict information for creating an operational plan for the specified period. Hereinafter, this information will be referred to as "prediction information," and the data composed of each prediction information for the specified period (i.e., time-series data composed of the values ​​of each prediction information at each time point included in the specified period) will be referred to as "prediction data." For example, if the target facility is the energy plant shown in Figure 1, examples of prediction information include electricity purchase price, city gas purchase price, electricity load, heat load, steam load, etc. Other examples of prediction information include the upper and lower output limits and output efficiency of each piece of equipment constituting the target facility.

[0032] Furthermore, the prediction unit 111 stores prediction data consisting of prediction information for the specified period in the database 118.

[0033] A prediction model is a function or algorithm that takes characteristic information, driving information, sensor information, prediction information, and a specified period as input and outputs prediction data consisting of prediction information for that specified period. For example, statistical and machine learning models, including neural networks, can be used as prediction models. However, neural networks are just one example and are not the only ones that can be used; any function or algorithm used in regression tasks can be used as a prediction model. If the prediction model is a neural network, it is trained, for example, using supervised learning methods to improve the accuracy of predictions for the specified period.

[0034] The optimal plan calculation unit 112 uses the prediction data stored in the database 118 (or prediction data in which some values ​​included in the prediction data have been changed) to calculate the optimal operation plan for the specified period using an optimization method. The optimal plan calculation unit 112 also stores time-series data representing the optimal operation plan (hereinafter also referred to as "optimal plan data") in the database 118. The optimal plan data consists of, for example, values ​​such as sensor information at each time point included in the specified period (hereinafter also referred to as "plan values").

[0035] The objective function of the optimization problem to be solved using the optimization method is predetermined, but for example, cost or CO2 emissions can be used. Furthermore, constraints on the optimization problem may include, for example, upper and lower limits on the output or output efficiency of the equipment or devices that make up the target facility.

[0036] The acquisition unit 113 acquires prediction data and optimal plan data calculated from that prediction data from the database 118.

[0037] The information processing unit 114 converts the prediction data and optimal plan data acquired by the acquisition unit 113 into a predetermined format (e.g., array format).

[0038] The target selection unit 115 selects one or more plan values ​​from among the plan values ​​that constitute the optimal plan data converted by the information organization unit 114 to be the target of providing justification. Hereinafter, the plan values ​​that are the target of providing justification will be referred to as "targets for providing justification." The targets for providing justification may be selected by the user on a UI (user interface) that visualizes the plan values ​​that constitute the optimal plan data, or they may be selected by other methods (e.g., selecting all plan values ​​for the final time of the optimal plan data as targets for providing justification).

[0039] The contribution calculation unit 116 uses the prediction data and optimal plan data converted by the information organization unit 114 to calculate the contribution of each piece of prediction information to the subject of evidence presentation.

[0040] The evidence presentation unit 117 uses the contribution level calculated by the contribution level calculation unit 116 and the predicted data converted by the information organization unit 114 to create and output information representing the evidence for the subject of evidence presentation (hereinafter also referred to as "evidence information"). For example, the evidence presentation unit 117 uses the predicted information with the highest contribution level and its contribution level to output a natural language sentence representing the evidence for the subject of evidence presentation as evidence information.

[0041] Database 118 stores various types of information (characteristic information, driving information, sensor information, prediction information), prediction data, optimal planning data, etc.

[0042] <Example of pre-processing> Figure 4 is a flowchart showing an example of preprocessing according to one embodiment. Below, we will describe the preprocessing, which involves collecting various types of information (characteristic information, driving information, sensor information, and prediction information) and calculating optimization plan data. Note that if the prediction data and optimal plan data are stored in database 118, preprocessing is not required.

[0043] The data collection unit 110 collects various types of information (characteristic information, driving information, sensor information, and prediction information) (step S101).

[0044] The collection unit 110 stores the various information collected in step S101 above in the database 118 (step S102).

[0045] The prediction unit 111 calculates prediction data using characteristic information, driving information, sensor information, and prediction information stored in the database 118, a specified period representing a future period specified by the user, and a pre-created prediction model (step S103). That is, the prediction unit 111 inputs characteristic information, driving information, sensor information, prediction information, and the specified period into the prediction model and outputs prediction data consisting of the values ​​of the prediction information at each time point included in the specified period.

[0046] The prediction unit 111 saves the prediction data calculated in step S103 above to the database 118 (step S104).

[0047] The optimal plan calculation unit 112 uses the prediction data saved in step S104 to calculate optimal plan data representing the optimal operation plan for the specified period using an optimization method (step S105). This calculates optimal plan data corresponding to the prediction data saved in step S104.

[0048] For example, predict the data Y={y m Let (s)|m=1,···,M, s=1,···,t}. Here, m is the label number representing the forecast information (e.g., the name of the forecast information), M is the number of labels representing the forecast information, [1,t] is the specified period, y m (s) is the value of the prediction information with the m-th label (hereinafter also referred to as "label m") at time s ∈ [1, t]. In this case, for example, the objective function C(·) = f(·; Y) is the sum of the electricity purchase amount and the city gas purchase amount during the specified period, and the upper output limit, lower output limit, and output efficiency of the equipment constituting the target facility are used as constraints. Note that the upper output limit and lower output limit of a certain equipment are out, respectively. max and out min It can be expressed as a constant such as [this]. Furthermore, the output efficiency of a certain device can be expressed as a function such as out = α × in + β (a linear function in this example), where in is the input and out is the output.

[0049] At this time, the optimal plan calculation unit 112 uses known optimization methods to minimize the objective function value C(X) by X={x k The optimal plan data is calculated as (s)|k=1,···,K, s=1,···,t}. Here, k is the label number representing the sensor information etc. that constitutes the optimal plan data (e.g., the name of the sensor information etc.), K is the number of labels representing the sensor information etc. that constitute the optimal plan data, x k(s) is the value (planned value) of sensor information, etc., with the k-th label (hereinafter also referred to as "label k") at time s ∈ [1, t]. Any optimization method capable of finding an exact solution can be used as the optimization method.

[0050] The optimal plan calculation unit 112 saves the optimal plan data calculated in step S105 above to the database 118 (step S106).

[0051] <Example of evidence presentation process> Figure 5 is a flowchart showing an example of the evidence presentation process according to one embodiment. Hereinafter, it is assumed that prediction data Y and the corresponding optimal plan data X are stored in the database 118.

[0052] The acquisition unit 113 acquires the prediction data Y and the corresponding optimal plan data X from the database 118 (step S201).

[0053] The information processing unit 114 converts the prediction data Y and optimal plan data X obtained in step S201 into a predetermined format (e.g., array format) (step S202). The following explanation assumes that the prediction data Y and optimal plan data X have been converted into array format.

[0054] As an example, Figure 6 shows the predicted data Y converted to array format. The predicted data Y shown in Figure 6 is y m (s) is represented as a two-dimensional array of size M × t with array elements. In the example shown in Figure 6, the label m=1 is "Electricity purchase price (yen / kWh)", the label m=2 is "City gas purchase price (yen / Nm3)", the label m=3 is "Electricity load (kWh)", the label m=4 is "Heat load (kWh)", and the label m=6 is "Steam load (kWh)". Here, the values ​​for heat load (kWh) and steam load (kWh) are the values ​​obtained by converting the heat load and steam load to electricity, respectively. This will continue to be the case hereafter.

[0055] Similarly, as an example, the optimal plan data X converted into an array format is shown in FIG. 7. The optimal plan data X shown in FIG. 7 has x k (s) represented as a two-dimensional array of K×t with array elements. In the example shown in FIG. 7, the label for k = 1st is "Power purchase amount (kWh)", the label for k = 2nd is "City gas purchase amount (Nm3)", the label for k = 3rd is "City gas usage amount of gas turbine (Nm3)", the label for k = 4th is "Power usage amount of turbo chiller 1 (kWh)", the label for k = 5th is "Power generation output amount of gas turbine (kWh)", the label for k = 6th is "Steam output amount of gas turbine (kWh)", etc.

[0056] The target selection unit 115 selects one or more planned values x k (s) that are the basis presentation targets among the planned values x k' (s) that constitute the optimal plan data X converted in step S202 above (step S203). Hereinafter, as an example, it is assumed that the planned value x

[0057] The basis presentation device 10 executes a contribution degree calculation process described later (step S204). In the contribution degree calculation process, the contribution degree of each prediction information y k' (s') to the basis presentation target x m (m = 1, ···, M) selected in step S203 above is calculated. Hereinafter, the contribution degree of the prediction information having the label m is Δ m is used.

[0058] The basis presentation unit 117 creates and outputs basis information (step S205) using the prediction data Y converted in step S202 above and the contribution degrees Δ m (m = 1, ···, M) calculated in step S204 above. As a result, the basis information is presented to the user. Here, the basis presentation unit 117 may create the basis information, for example, according to the following procedure 1 to procedure 2.

[0059] Procedure 1: The basis presentation unit 117 has the contribution degree Δ mThis identifies the highest label m'. In other words, the evidence presentation unit 117 states that m'=argmax m {Δ m Let m = 1, ..., M.

[0060] Step 2: The evidence presentation unit 117 presents the label m' and the value y of the prediction information associated with label m'. m' Using (s)(s=1,···,t), the object of evidence presentation x k' Create supporting information that represents the basis for (s'). For example, the supporting information unit 117 states, "Label m' is y m' (s'-1) to y m' (s') has changed, so at time s' the label k' is x k' Create a natural language sentence that says "(s') became."

[0061] For example, label k' is "gas turbine steam output (kWh)", x k' (s')=0.0, label m' is "City gas purchase price (yen / Nm3)", y m' (s'-1)=73.11, y m' Let (s') = 83.11. In this case, the supporting information "The price of city gas purchased (yen / Nm3) changed from 73.11 to 83.11, so the steam output of the gas turbine (kWh) became 0.0 at time s'" is created.

[0062] For example, y m' (s')-y m' Using the sign and absolute value of (s'-1), if the label m' is |y m' (s')-y m' (s'-1)|{y m' (s')-y m' If the sign of (s'-1) is +, it "increased"; if it is -, it "decreased". Therefore, at time s', label k' is x k'The supporting information may be created as "(s') occurred". In this case, for example, the supporting information "The price of city gas purchased per unit (yen / Nm3) increased by 10, so the steam output of the gas turbine (kWh) became 0.0 at time s'" would be created. Here, {"string1" if the condition is met, "string2" if the condition is not met} is a function that outputs string1 if the condition is met, and string2 if the condition is not met.

[0063] ≪Example of contribution calculation process≫ Figure 8 is a flowchart showing an example of the contribution calculation process according to one embodiment.

[0064] The contribution calculation unit 116 selects the label r∈{1,···,M} (step S301).

[0065] The contribution calculation unit 116 calculates the value y of the prediction information with label r at time s'. r For (s'), the value of s'-1 one time step earlier is y. r (s'-1) is set (step S302). That is, the contribution calculation unit 116 sets y r (s')←y r Let (s'-1).

[0066] Below, y m (s) ∈ Y, among y r (s') only y r (s')←y r y set to (s'-1) m Change the set of (s)(m=1,···,M, s=1,···,t) to predict data Y r Let's assume that.

[0067] The optimal plan calculation unit 112 uses change prediction data Y r Using this, the optimal plan data representing the optimal driving plan for the specified period is calculated (step S303), similar to step S105 above. That is, the optimal plan calculation unit 112 uses the change prediction data Y instead of the prediction data Y. r Using this, the optimal plan data is calculated in the same manner as in step S105 above. Hereafter, this optimal plan data is Xr ={x k r Let (s)|k=1,···,K, s=1,···,t}.

[0068] The contribution calculation unit 116 calculates the evidence target x k' Planned value x for the same time and label as (s') k' r (s')∈X r Using the contribution Δ r The calculation is performed (step S304). For example, the contribution calculation unit 116 calculates the evidence target x k' Planned value x for (s') k' r Change in (s') |x k' (s')-x k' r (s')| Contribution Δ r It is calculated as follows. However, this is the contribution Δ r This is one example of a method for calculating the contribution Δ, and other methods may be used to determine the contribution Δ r The following may be calculated.

[0069] The contribution calculation unit 116 determines whether or not there are any unselected labels (step S305).

[0070] If it is determined in step S306 that there are unselected labels, the contribution calculation unit 116 selects the unselected label r∈{1,···,M}\R (step S306) and returns to step S302. Here, R is the set of numbers of the selected labels. As a result, steps S302 to S304 are repeatedly executed for all labels r∈{1,···,M}, and the contribution Δ1,···,Δ M This is calculated.

[0071] On the other hand, if it is determined that there are no unselected labels in step S306 above, the contribution calculation process is terminated.

[0072] <Variation> The following describes variations of the above embodiment. Note that multiple variations can be combined as appropriate, as long as they do not contradict each other.

[0073] • Variation 1 The prediction information that constitutes the prediction data calculated in step S103 of Figure 4 may include, for example, the upper limit of output, the lower limit of output, the output efficiency, etc. That is, the prediction data Y = {y m With respect to (s)|m=1,···,M, s=1,···,t}, the set of label numbers {1,···,M} may include label numbers representing the upper output limit, lower output limit, and output efficiency. This makes it possible to calculate optimal plan data that takes into account changes in step S105 of Figure 4, for example, even if the upper and lower output limits or output efficiency change according to weather information.

[0074] Note that in step S302 of Figure 8, if the label r is "output limit", then y r (s')←y r (s'-1) is also acceptable, or y r (s')←D is also acceptable. Here, D is a predetermined large real number (e.g., the maximum value that can be taken as a real number). Similarly, if the label r is the "lower limit of output" in step S302 of Figure 8, then y r (s')←y r (s'-1) is also acceptable, or y r (s') ← -D is also acceptable. y r (s')←D or y r By setting (s')←-D, the optimal plan data X in step S303 of Figure 8 is optimized while ignoring the upper or lower output limit represented by label r. r It is possible to calculate this.

[0075] Also, if the label r is "output efficiency" in step S302 of Figure 8, then y r (s')←y r (s'-1) is also acceptable, or y r You can set a predetermined function for (s'), or y r (s') can be removed.r By deleting (s'), the optimal planning data X r that performs optimization ignoring the output efficiency represented by the label r in step S303 of FIG. 8 can be calculated.

[0076] · Variant 2 In step S205 of FIG. 5, for example, the basis information represented by a causal diagram (Reference 2) may be created. A causal diagram is a directed graph in which the causal relationship between nodes is represented by edges. In this case, since time series data of the contribution degree Δ m is required, for example, for s', s'-1, s'-2, ···, etc., the contribution degree calculation process shown in FIG. 8 is repeatedly executed, and the time series data {Δ m (s'), Δ m (s'-1), Δ m (s'-2), ···} of the contribution degree Δ m needs to be obtained. Here, Δ m (s) is the contribution degree Δ m calculated when the contribution degree calculation process shown in FIG. 8 is executed with s'←s.

[0077] At this time, an example of the basis information represented by a causal diagram is shown in FIG. 9. The basis information shown in FIG. 9 is a causal diagram when two basis presentation targets x k1' (s') and x k2' (s') are selected. In the basis information shown in FIG. 9, two basis presentation targets x k1' (s') and x k2' (s') are respectively represented by nodes 201 and 202, and the time series data {y m (s)|s = 1, ···, s'} of each prediction information is represented by nodes 301 to 304. In addition, a cause is assigned to the start point of the edge connecting each node, and a result is assigned to the end point. As a result, the user can grasp the cause for the basis presentation target by the causal diagram.

[0078] <Summary> As described above, the rationale presentation device 10 according to one embodiment can present the user (e.g., operator of equipment such as a plant) with rationale information for one or more planned values ​​included in the optimal operation plan obtained using the optimization method. Therefore, by using the rationale presentation device 10 according to one embodiment, the user can trust the operation plan and achieve optimal operation of the target equipment (e.g., operation that minimizes costs and CO2 emissions).

[0079] The present invention is not limited to the embodiments specifically disclosed above, and various modifications, changes, and combinations with known technologies are possible as long as they do not deviate from the spirit described in the claims.

[0080] [References] Reference 1: Optimization Benchmark Problems for Industrial Applications - Optimization Benchmark Problems for Energy Plant Operation Planning (Ver. 1), Internet<URL:https: / / takashi0825.github.io / ssl_webpage / BP-IA / ja / P1-1.html> Reference 2: Aapo Hyvarinen, Kun Zhang, Shohei Shimizu, Patrik O. Hoyer. Estimation of a Structural Vector Autoregression Model Using Non-Gaussianity. Journal of Machine Learning Research, 11: 1709-1731, 2010. [Explanation of Symbols]

[0081] 10. Evidence Presentation Device 101 Input Device 102 Display device 103 External I / F 103a Recording medium 104 Communication I / F 105 RAM 106 ROM 107 Auxiliary storage 108 processors 109 Bus 110 Collection Department 111 Prediction Section 112 Optimal Plan Calculation Unit 113 Acquisition Department 114 Information Organizing Department 115 Target Selection Section 116 Contribution Calculation Unit 117. Section for Presenting Evidence 118 Databases

Claims

1. A selection unit that selects the planned values ​​to be included in the operation plan of the target equipment for which justification should be provided, A calculation unit that calculates, for each of the multiple time-series data used in calculating the aforementioned driving plan, the change in the value of the subject of justification when at least the value at the same time as the subject of justification among the values ​​included in the time-series data is taken as the value at the previous time, A presentation unit that presents evidence information representing the basis for the evidence target using the change in the aforementioned value of the evidence target, An information processing device having

2. The aforementioned display unit is, The information processing device according to claim 1, which presents evidence information that includes a label representing the name of the time-series data with the largest change in the value of the evidence subject to presentation.

3. The aforementioned display unit is, The information processing device according to claim 2, which presents the label, the magnitude of the change in value when the time series data having the label is moved from the same time as the subject of the justification to the time immediately preceding it, and the sign of the change in value when the time series data having the label is moved from the same time as the subject of the justification to the time immediately preceding it, expressed in natural language.

4. The aforementioned display unit is, The information processing device according to claim 1, which presents the evidence information represented by a causal diagram showing the causal relationship between the subject of evidence presentation and the plurality of time-series data.

5. The information processing device according to any one of claims 1 to 4, wherein the operation plan is an optimal operation plan calculated by an optimization method using the plurality of time-series data.

6. A selection procedure for selecting the planned values ​​to be included in the operating plan of the target equipment and for which justification should be provided, A calculation procedure for each of the multiple time-series data used in calculating the aforementioned driving plan, which calculates the change in the value of the subject of justification when at least the value at the same time as the subject of justification among the values ​​included in the time-series data is taken as the value at the previous time, A presentation procedure for presenting supporting information that represents the basis for the aforementioned target of evidence presentation, using changes in the value of the target of evidence presentation, A method of information processing performed by a computer.

7. A selection procedure for selecting the planned values ​​to be included in the operating plan of the target equipment and for which justification should be provided, A calculation procedure for each of the multiple time-series data used in calculating the aforementioned driving plan, which calculates the change in the value of the subject of justification when at least the value at the same time as the subject of justification among the values ​​included in the time-series data is taken as the value at the previous time, A presentation procedure for presenting supporting information that represents the basis for the aforementioned target of evidence presentation, using changes in the value of the target of evidence presentation, A program that causes a computer to execute something.

Images