Facility operation planning device and facility operation planning method

Abstract

A facility operation planning device (4) is characterized by comprising: a node condition generation unit (43) that generates a constraint condition for each node in a system in which a plurality of facilities and a plurality of nodes are combined; a variable range set generation unit (48) that generates a variable range serving as a candidate for limiting a decision variable, corrects the variable range serving as the candidate on the basis of the constraint condition for the node generated by the node condition generation unit (43), and stores the corrected variable range in a variable range set; and a variable range limiting unit (47) that uses the variable range set storing the variable range corrected by the variable range set generation unit (48) to limit the variable range of the decision variable before optimization of the system.

Description

Equipment operation planning device and equipment operation planning method

[0001] The present invention relates to an equipment operation planning device and an equipment operation planning method.

[0002] In the operational planning of energy systems composed of numerous energy facilities, measures are being taken to support customers' investment decisions regarding power facilities by generating multiple investment scenarios for their power facilities through time-series balance simulations of power supply and demand.

[0003] Assume that between energy equipment, a physical quantity, such as electrical energy, is exchanged for each planned time slot, for example, at a 30-minute granularity. The equipment operation planning problem can be described as an optimization problem where the input and output physical quantities of each piece of equipment in each time slot are variables.

[0004] Patent Document 1 describes an invention of an operation planning system that generates a plan by generating a combination of operation units to operate during the planning period such that the required total output value is met, and by determining the output setting value of the operation units included in the combination.

[0005] Japanese Patent Publication No. 2019-211989

[0006] Patent Document 1 does not address the combination of equipment operations, i.e., the limitations on the range of variables, when physical quantities are exchanged between multiple pieces of equipment. To speed up equipment operation planning, one could consider a method of pre-limiting the range of some variables in the optimization problem using machine learning or the like. However, depending on the limited range, it may not be possible to satisfy the constraints regarding the exchange of physical quantities between equipment, potentially resulting in an unfeasible operation plan. Constraints regarding the exchange of physical quantities between equipment include, for example, the conservation of energy.

[0007] Therefore, the present invention aims to accelerate the planning of operation plans for energy systems that include energy equipment.

[0008] To solve the aforementioned problems, the equipment operation planning device of the present invention is a system that combines multiple equipment and multiple nodes, including a conversion type equipment that takes one or more types of physical quantities as input and outputs one or more types of physical quantities as input, and uses predetermined types of output physical quantities output from the equipment to the nodes and predetermined types of input physical quantities input from the nodes to the equipment as decision variables, and generates constraint conditions for the nodes by relating the sum of the predetermined types of output physical quantities output from one or more equipment to the nodes to the sum of the predetermined types of input physical quantities input from the nodes to one or more equipment using equality or inequality signs; generates candidate ranges that restrict the decision variables, modifies the candidate ranges based on the constraint conditions for the nodes generated by the node condition generation unit and stores them in a range set; and uses the range set that stores the ranges modified by the range set generation unit to restrict the ranges of the decision variables before optimizing the system.

[0009] In other words, the equipment operation planning method of the present invention is characterized by comprising the steps of: a system combining multiple equipment and multiple nodes, including a conversion type equipment that takes one or more types of physical quantities as input and outputs one or more types of physical quantities as input, wherein predetermined types of output physical quantities output from the equipment to the nodes at each time and predetermined types of input physical quantities input from the nodes to the equipment are used as decision variables, a node condition generation unit generates constraint conditions for the nodes by relating the sum of predetermined types of output physical quantities output from one or more equipment to the nodes to the sum of predetermined types of input physical quantities input from the nodes to one or more equipment using equality or inequality signs; a range set generation unit generates candidate ranges that restrict the decision variables, the range set generation unit modifies the candidate ranges based on the constraint conditions for the nodes generated by the node condition generation unit and stores them in a range set; and a range limiting unit uses the range set modified by the range set generation unit to limit the range of the decision variables before optimizing the system. Other means will be described in the embodiments for carrying out the invention.

[0010] According to the present invention, it becomes possible to accelerate the planning of operation plans for energy systems that include energy equipment.

[0011] This is a diagram of the equipment operation planning system according to the first embodiment. This is a diagram of the equipment operation planning device. This is a diagram of the energy system. This is a graph showing input and output physical quantities as piecewise linear. This is a flowchart of the operation planning process. This is a flowchart of the range set generation process. This is a flowchart of the range set modification process. This is a diagram of the main components of the energy system. This is a diagram showing the range of each input and output physical quantity. This is the operation planning setting screen. This is a diagram of the equipment operation planning system according to the second embodiment. This is a flowchart of the operation planning process. This is a flowchart of the range set generation process.

[0012] Hereafter, embodiments for carrying out the present invention will be described in detail with reference to the figures. In this embodiment, the range of input and output physical quantities predicted by machine learning is repeatedly modified so that it satisfies the constraints on the exchange of physical quantities. Therefore, it is checked for each time slot whether the range of input and output physical quantities predicted by machine learning satisfies each constraint on the exchange of physical quantities. If any of the constraints on the exchange of physical quantities are violated, the range of input and output physical quantities is modified. Modifying the range of input and output physical quantities with respect to one constraint may affect other constraints, so the modification process of the range of input and output physical quantities is repeated by cycling through each constraint.

[0013] Figure 1 is a diagram showing the configuration of the equipment operation planning system 400 according to the first embodiment. The equipment operation planning system 400 consists of an equipment operation planning device 4, a constant data management and distribution device 1, a predictive model learning and management device 2, an operation and monitoring terminal 3, and an equipment control device 5, and plans the operation of the equipment of the controlled system 6.

[0014] The equipment operation planning device 4 is configured as a functional unit that includes a conversion-type equipment condition generation unit 41, a storage-type equipment condition generation unit 42, a node condition generation unit 43, a constant setting unit 44, an objective function generation unit 45, an optimal plan planning unit 46, a variable range restriction unit 47, and a variable range set generation unit 48. The equipment operation planning device 4 further includes an optimal plan management unit 401 and a constant data management unit 402 as information management units, and is configured to include an equipment operation planning program 403. This equipment operation planning device 4 controls the equipment included in the controlled system 6 via the equipment control device 5.

[0015] The controlled system 6 includes supply-type equipment, conversion-type equipment, and storage-type equipment. This controlled system 6 includes a node that receives output physical quantities from the equipment. Input physical quantities are input to the equipment from this node. The supply-type equipment outputs one or more physical quantities. The conversion-type equipment takes one or more types of physical quantities as input and outputs one or more types of physical quantities. The storage-type equipment takes one or more types of physical quantities as input, outputs physical quantities of the same type, and stores the difference.

[0016] The conversion-type equipment condition generation unit 41 generates constraint conditions for conversion-type equipment included in the controlled system 6. The storage-type equipment condition generation unit 42 generates constraint conditions for storage-type equipment included in the controlled system 6.

[0017] The node condition generation unit 43 generates constraint conditions for the node by relating the sum of predetermined types of output physical quantities output from one or more pieces of equipment to the node with the sum of predetermined types of input physical quantities input from this node to one or more pieces of equipment using an equals sign or an inequality sign. The constant setting unit 44 sets various setting values ​​related to the controlled system 6 as constants in the constant data management unit 402.

[0018] The objective function generation unit 45 generates an objective function for evaluating the equipment operation plan of the controlled system 6. The optimal plan formulation unit 46 formulates an optimal plan for operating the equipment of the controlled system 6 based on the objective function generated by the objective function generation unit 45 and stores it in the optimal plan management unit 401. The optimal plan formulation unit 46 then optimizes the controlled system 6. The domain restriction unit 47 restricts the domain of the decision variables before optimizing the controlled system 6 using a domain set that stores the domains modified by the domain set generation unit 48. The domain restriction unit 47 trains a machine learning model on the relationship between the previously restricted domains of decision variables and the resulting plan for each node.

[0019] The domain set generation unit 48 uses predetermined types of output physical quantities output from equipment to nodes and predetermined types of input physical quantities input from nodes to equipment as decision variables. The domain set generation unit 48 then generates candidate domains that restrict these decision variables, modifies the candidate domains based on the node constraints generated by the node condition generation unit 43, and stores them in the domain set.

[0020] The domain set generation unit 48 divides the input physical quantity and expresses the output physical quantity as a linear function of each division. The divisions whose predicted probability is greater than or equal to a predetermined probability are designated as candidate domains to restrict the decision variable. The domain set generation unit 48 expresses the relationship between the input physical quantity and the output physical quantity as a piecewise linear function and defines the domains based on the predicted probability of each division, thereby simplifying complex nonlinear relationships. Furthermore, by defining the domains based on the predicted probabilities, the domains can be appropriately selected according to the predicted probabilities. When the constraints on a node do not hold, the domain set generation unit 48 modifies itself to broaden the domains that restrict the decision variable by lowering the predetermined probability compared with the predicted probability of the division.

[0021] The domain set generation unit 48 determines the priority of each decision variable when modifying the domain constraints of each decision variable related to a node, based on the prediction results of the machine learning model. The domain set generation unit 48 may prepare multiple node traversal orders when modifying the domain set and perform the modification of the domain set in parallel processing for each traversal order.

[0022] When the optimal planning unit 46 determines that optimization of the controlled target system 6 cannot be performed, the variable domain set generation unit 48 may correct the variable domain set, and the optimal planning unit 46 may perform optimization of the controlled target system 6 again.

[0023] The optimal plan management unit 401 stores the optimal plan formulated by the optimal planning unit 46. The constant data management unit 402 manages various setting values related to the controlled target system 6 as constants.

[0024] The facility operation planning program 403 is a program executed by a CPU (Central Processing Unit) described later. When the CPU executes the facility operation planning program 403, each functional unit of the facility operation planning apparatus 4 is implemented.

[0025] The facility control apparatus 5 comprises a control unit 51, a communication unit 52, and a control data management unit 53. The control unit 51 performs overall control of the facility control apparatus 5 to control the facilities of the controlled target system 6. The communication unit 52 outputs a control signal to the apparatus based on control data received from the facility operation planning apparatus 4. The communication unit 52 further collects data from the apparatus and transfers the collected data to the facility operation planning apparatus 4.

[0026] The constant data management and distribution apparatus 1 comprises a demand forecast data management unit 11, a weather forecast data management unit 12, a facility data management unit 13, an acquisition unit 14, and a distribution unit 15. The constant data management and distribution apparatus 1 acquires various types of forecast data from the outside via the acquisition unit 14, and distributes the various types of forecast data via the distribution unit 15.

[0027] The constant data management and distribution apparatus 1 receives demand forecast data from a demand forecast data distribution server 71 via the acquisition unit 14 through a network 73, and stores the demand forecast data in the demand forecast data management unit 11. Then, the distribution unit 15 distributes the demand forecast data to the facility operation planning apparatus 4.

[0028] The constant data management and distribution device 1 receives weather forecast data from the weather forecast data distribution server 72 via the network 73 using the acquisition unit 14 and stores it in the weather forecast data management unit 12. The distribution unit 15 then distributes this weather forecast data to the equipment operation planning device 4.

[0029] The constant data management and distribution device 1 distributes the equipment data stored in the equipment data management unit 13 to the equipment operation planning device 4 via the distribution unit 15.

[0030] The predictive model learning and management device 2 comprises a learning unit 21 and a predictive model management unit 22. The learning unit 21 causes a machine learning model, which predicts whether or not the range of the decision variables for each node can be restricted, to learn whether or not the range of the decision variables for each node can be restricted. Based on instructions from the range restriction unit 47, the learning unit 21 causes the machine learning model to learn the relationship between the range of decision variables that have been restricted in the past and the planning results for each node's decision variables. The predictive model management unit 22 manages the machine learning model learned by the learning unit 21.

[0031] The operation and monitoring terminal 3 comprises an operation unit 31 and a display unit 32. The operation unit 31 is, for example, a keyboard, mouse, or touch panel, and accepts user input. The display unit 32 is, for example, a display, and shows text, images, or graphics to the user.

[0032] Figure 2 is a hardware configuration diagram of the equipment operation planning device 4. The equipment operation planning device 4 includes a CPU 411, memory 412, secondary storage device 413, input / output interface 414, and network interface 415.

[0033] The CPU 411 is one or more central processing units that provide overall control for the equipment operation planning device 4. The memory 412 is a temporary storage device, such as a volatile storage device or a non-volatile storage device. The secondary storage device 413 is a large-capacity storage device such as a hard disk or an SSD (Solid State Drive).

[0034] The input / output interface 414 is an interface for input signals from the keyboard and mouse and output signals to the display. The network interface 415 is an interface for connecting to a network and exchanging information with other devices.

[0035] In the memory 412, each functional module of the equipment operation planning device 4 is loaded as a computer program and executed by the CPU 411.

[0036] Furthermore, the equipment operation plan planning program 403 is stored in the secondary storage device 413.

[0037] The equipment operation planning device 4 may be implemented as a virtual machine instead of a physical machine. The equipment operation planning device 4 connects to the equipment or the device that controls the equipment via a network. The equipment operation planning device 4 may also be implemented in a cloud environment. In that case, the equipment operation planning device 4 is connected to the environment in which the equipment is configured via a network.

[0038] Figure 3 is a diagram of the configuration of the simulation model 8 of the energy system. The simulation model 8 includes a power supply network consisting of a power purchase facility 801, a solar power generation facility 802, a storage battery 803, and a storage battery power conditioner 807, which supplies power to the power demand facility 805. In the diagram, the solar power generation power conditioner 806 is abbreviated as "PV PCS".

[0039] The power supply network in simulation model 8 includes nodes 809, 810, and 811. The equipment operation planning device 4 pre-defines the power demand for each time period through a prior simulation.

[0040] Of these facilities, the electricity purchase equipment 801 and the solar power generation equipment 802 are supply-type facilities that output one or more physical quantities. The solar power generation power conditioner 806 and the battery power conditioner 807 are conversion-type facilities that take one or more types of physical quantities as input and output one or more types of physical quantities. The battery 803 is a storage-type facility that takes one or more types of physical quantities as input, outputs the same type of physical quantity, and stores the difference.

[0041] Node 810 combines the electricity purchased from the power purchase equipment 801, the power generated by the solar power conditioner 806, and the discharged power output by the battery power conditioner 807. Node 810 further supplies power to the power demand equipment 805 and to the battery power conditioner 807.

[0042] The electricity purchase equipment 801 supplies purchased electricity and outputs electricity purchase costs and CO2 emissions. The solar power generation equipment 802 supplies DC power to the solar power generation power conditioner 806 via node 809. The solar power generation power conditioner 806 converts the DC power to AC power and supplies this AC power to node 810.

[0043] The battery power conditioner 807 converts the AC power supplied from node 810 into DC power and supplies it to the battery 803 via node 811. The DC power supplied from the battery 803 is then supplied to the battery power conditioner 807. The battery power conditioner 807 converts the DC power into AC power and supplies it to node 810.

[0044] In the pre-simulation, the user manually or through the system defines the simulation data for the target planning period for the simulation model 8. Then, they create planning conditions by combining the simulation model 8 and the simulation data, and run the simulation.

[0045] From the power purchase equipment 801 to node 810, the amount of purchased electricity x １ The output is as follows: From the solar power conditioner 806 to node 810, the generated power x ２ The following is output. From node 810 to power demand equipment 805, the amount of electricity demanded c １ The following is output. From node 810 to the battery power conditioner 807, the amount of charge x ３is output. Between the sum of input electric energy, which is the input physical quantity of node 810, and the sum of output electric energy, which is the output physical quantity, the law of conservation of electric energy holds. That is, the following formula (1) holds. x １ + x ２ = c １ + x ３ ... (1)

[0046] This simulation model 8 is used to calculate the optimal solution for the facility operation planning problem with a 30-minute granularity for 24 hours, totaling 48 time slots. Here, the objective function is the total power purchase cost [yen] for 48 time slots in the planning target period.

[0047] The facility operation planning apparatus 4 formulates a facility operation plan that minimizes the power purchase cost. In order to speed up the planning, the range of input and output physical quantities, which are the input-output physical quantities, is limited before the optimization process while considering the law of conservation of electric energy at the three nodes.

[0048] FIG. 4 is a graph showing input-output physical quantities in piecewise linear form. For each conversion facility, the relationship between the input physical quantity and the output physical quantity is given as a piecewise linear relationship. Here, the number of segments is three, and each segment has an independent linear relationship. The vertical axis of the upper graph indicates the output physical quantity, and the horizontal axis indicates the input physical quantity. The vertical axis of the lower graph indicates the prediction probability, and the horizontal axis indicates the input physical quantity. The prediction probability for segment #1 is 0.9, the prediction probability for segment #2 is 0.8, and the prediction probability for segment #3 is 0.2.

[0049] The segment number in the general formula of the piecewise linear relationship is l, and l is a natural number. In the facility operation plan, a binary segment variable Z that indicates which segment is to be used ｇｌ (t) is the variable to be optimized. When the value of the segment variable is 1, the segment is included in the domain. When the value of the segment variable is 0, the segment is excluded from the domain.

[0050] By machine learning, for each time slot t, each facility g, and the relationship of each input-output physical quantity, the probability p that the segment variable Z ｇｌ (t) becomes 1 ｇ ｌ (t) is predicted in advance. The prediction probability p ｇｌ The interval in which (t) is less than a predetermined probability is estimated to be less likely to be used in the optimal solution. In this case, the piecewise variable Z ｇｌ The candidate values ​​for (t) are set to 0, and this range is not explored in the subsequent optimization process.

[0051] Predicted probability p ｇ ｌ The interval in which (t) is greater than or equal to a predetermined probability is estimated to be likely to be used in the optimal solution. In this case, the piecewise variable Z ｇｌ The candidate values ​​for (t) are set to 1, and this region is to be explored in the subsequent optimization process.

[0052] Figure 5 is a flowchart of the operation plan formulation process. First, the transformation-type equipment condition generation unit 41, the storage-type equipment condition generation unit 42, the node condition generation unit 43, the constant data management unit 402, the constant setting unit 44, and the objective function generation unit 45 each process to generate the optimization problem to be solved (step S10). Then, the domain set generation unit 48 acquires a machine learning model from the prediction model learning and management device 2 (step S11). The domain set generation unit 48 performs a domain set generation process using the machine learning model (step S12). This allows the optimal plan formulation unit 46 to speed up the formulation of the optimization plan. Then, the domain restriction unit 47 restricts the domain of input and output physical quantities based on the generated domain set (step S13). The domain restriction unit 47 determines whether or not the domain of the decision variable for each node can be restricted based on the prediction results of the machine learning model. The domain restriction unit 47 trains the machine learning model on the relationship between the domains of previously restricted decision variables and the resulting plans for each node. The optimal plan formulation unit 46 then formulates an optimal plan (step S14). The optimal plan formulation unit 46 determines whether or not an optimal plan has been formulated (step S15). If an optimal plan could not be formulated (No), the process proceeds to step S16, where the generation of another set of domains is instructed, and the process returns to step S11. If an optimal plan has been formulated (Yes), the process in Figure 5 ends. This allows the generation of another set of domains and the formulation of an optimal plan to be repeated until an optimal plan can be formulated appropriately.

[0053] Figure 6 is a flowchart of the domain set generation process shown in step S12 of Figure 5. Initially, the domain set generation unit 48 calculates the probability p that the value of the piecewise variable is 1 using a machine learning model. ｇ ｌ (t) is predicted (step S20). This makes it possible to predict which section l is most likely to be used in the linear relationship of each piece of equipment g.

[0054] The domain set generation unit 48 predicts the probability p ｇ ｌ Based on (t), the piecewise variable Z ｇｌ The values ​​of the range (t) (∈{0,1}) are calculated, and a range set is constructed by combining them (step S21). Then, the range set generation unit 48 generates n possible traversal orders for node c (step S22). This is because the modification of the range set may be successful depending on the traversal order of node c.

[0055] Next, the domain set generation unit 48 performs parallel processing to modify the domains included in the domain set so that the constraints of the input and output physical quantities are satisfied in each cycle order (steps S23a to S23n), and then terminates the process shown in Figure 6. The process of modifying the domains by changing the cycle order of node c can be processed independently in parallel, so it is possible to modify the domains faster than processing them sequentially.

[0056] Figure 7 is a flowchart of the domain set correction process shown in steps S23a to S23n of Figure 6. The domain set generation unit 48 is executed repeatedly for each time slot t of the planned target (step S30). Next, the domain set generation unit 48 is executed repeatedly for all nodes c (step S31). Note that in the loop processing from steps S31 to S35, the domain set generation unit 48 may, but is not limited to, prepare multiple node traversal orders and perform correction of the domain set for each traversal order in parallel processing. This makes it possible to select the optimal node traversal order.

[0057] The domain set generation unit 48 calculates the domain of the input and output physical quantities from the equipment to node c when each variable is restricted to the domain of the domain set (step S32). From the domains from the equipment to node c, the equipment involved in the constraint violation is extracted (step S33).

[0058] The variable domain set generation unit 48 modifies candidate partition variables that contribute to the output physical quantity output from the equipment to node c, or the input physical quantity input from node c to the equipment (step S34). Then, the variable domain set generation unit 48 determines whether or not processing has been performed for all nodes c (step S35). If there are any unprocessed nodes c, the process returns to step S31. The variable domain set generation unit 48 determines whether the modification of the candidate set has converged, or whether the specified γ modification cycles have been performed (step S36).

[0059] If the modification of the candidate set does not converge and the specified γ modifications have not been performed, the domain set generation unit 48 returns to step S31. If the modification of the candidate set converges or the specified γ modifications have been performed, the process proceeds to step S37.

[0060] The domain set generation unit 48 determines whether or not processing has been performed for all time slots t that are being planned. If processing has not been performed for any of the time slots t, the process returns to step S30. If processing has been performed for all time slots t, the process shown in Figure 7 is terminated.

[0061] Figure 8 is a diagram showing the main components of the energy system. The amount of electricity purchased x is transmitted from the power purchase equipment 801 to node 810. １ The output is: Purchased electricity amount x １ The range of the variable is between 0 and 10. From the solar power conditioner 806 to node 810, the generated power x ２ The output is: Amount of power generated x ２ The range of the value is between 0 and 10.

[0062] From node 810 to power demand equipment 805, the amount of electricity demanded d １ The output is: Demand amount d １ The range of this constant is only 40. From node 810 to battery power conditioner 807, the amount of charge x ３ The output is: Charged energy x ３The range of x is between 0 and 500. The law of conservation of energy holds between the sum of the input physical quantities of node 810 and the sum of the output physical quantities. That is, the following equation (2) holds: x １ +x ２ = d １ +x ３ … (2)

[0063] Figure 9 shows the range of each input / output physical quantity to node 810 when each variable is tentatively fixed to a candidate value. When the equipment is a power purchase equipment, the physical quantity is the amount of power purchased, the type is the variable, and the candidate range is x １ = 1. When the equipment is a solar power generation facility, the physical quantity is the amount of electricity generated, the type is a variable, and the candidate range is 0 ≤ x ２ The range is ≤ 10. When the equipment is a battery power conditioner, the physical quantity is the amount of charge, the type is a variable, and the range is 0 ≤ x. ３ The value is ≤ 500.

[0064] Figure 10 shows the setting screen 9 for operation planning. This setting screen 9 is displayed on the display unit 32 of the operation / monitoring terminal 3 shown in Figure 1. The setting screen 9 includes a time granularity text box 91, a number of planned time slots 92, a speed optimization application combo box 93, and an upper limit γ text box 94.

[0065] The user can set the time granularity of the simulation by entering a numerical value in the time granularity text box 91 using the operation unit 31. The user can set the number of time slots to be planned for the simulation by entering a numerical value in the number of planned time slots 92 using the operation unit 31. The user can select to speed up the simulation by limiting the range of some variables by selecting "Apply" in the speed-up application combo box 93 using the operation unit 31. If "Do not apply" is selected in the speed-up application combo box 93, the simulation will be executed without limiting the range of variables. The user can set the upper limit γ of the number of constraint check iterations using the operation unit 31.

[0066] Figure 11 is a diagram showing the configuration of the equipment operation planning system 400A according to the second embodiment. The equipment operation planning system 400A is composed of an equipment operation planning device 4A different from that of the first embodiment, a constant data management and distribution device 1, a predictive model learning and management device 2, an operation and monitoring terminal 3, and an equipment control device 5, and plans the operation of the equipment of the controlled system 6.

[0067] The equipment operation planning device 4A is configured as a functional unit that includes a conversion-type equipment condition generation unit 41, a storage-type equipment condition generation unit 42, a node condition generation unit 43, a constant setting unit 44, an objective function generation unit 45, an optimal plan planning unit 46, a variable range restriction unit 47, and a variable range set generation unit 48. The equipment operation planning device 4 further includes an information management unit that includes an optimal plan management unit 401, a constant data management unit 402, and a statistical information management unit 404, and is configured to include an equipment operation planning program 403. This equipment operation planning device 4A controls the equipment included in the controlled system 6 via the equipment control device 5.

[0068] The statistical information management unit 404 stores statistical information on the range restrictions of each node related to previously formulated operation plans, and statistical information on past operation plans formulated based on the set of range restrictions at that time. The range restriction unit 47 determines whether or not to implement range restrictions for each decision variable related to each node based on this past statistical information. The range restriction unit 47 restricts the range of the decision variable related to each node based on the relationship between the past range restrictions of the decision variable related to each node and the plan results at that time.

[0069] The domain restriction unit 47 restricts the domain of the decision variables before optimizing the controlled system 6, using a domain set that stores the domains modified by the domain set generation unit 48. Before the optimal planning unit 46 optimizes the system, the domain restriction unit 47 restricts the domain of the decision variables for each node based on the relationship between past domain restrictions for the decision variables for each node and the planning results at that time.

[0070] The domain set generation unit 48 then determines the priority of each decision variable when modifying the domain restrictions of each decision variable related to a node, based on the relationship between past domain restrictions of the decision variables related to the node and the planning results at that time. The domain set generation unit 48 then determines the priority of each decision variable when modifying the domain restrictions of each decision variable related to a node, based on past planning results.

[0071] Figure 12 is a flowchart of the operation plan formulation process. First, the optimization problem to be solved is generated by the transformation-type equipment condition generation unit 41, the storage-type equipment condition generation unit 42, the node condition generation unit 43, the constant data management unit 402, the constant setting unit 44, and the objective function generation unit 45 (step S40).

[0072] The domain set generation unit 48 performs a domain set generation process using past statistical information from the statistical information management unit 404 (step S41). This allows the optimal plan formulation unit 46 to speed up the formulation of an optimization plan. Then, the domain limiting unit 47 limits the domain of input and output physical quantities based on the generated domain set (step S42). The domain limiting unit 47 determines whether or not to implement domain limiting for each decision variable related to each node, based on past statistical information of the decision variables related to each node.

[0073] Furthermore, the optimal plan formulation unit 46 formulates an optimal plan (step S43). The optimal plan formulation unit 46 determines whether or not it was able to formulate an optimal plan (step S44). If it was not able to formulate an optimal plan (No), it proceeds to step S45, instructs the generation of a different set of variables, and returns to the process in step S41. If it was able to formulate an optimal plan (Yes), the process in Figure 12 is completed.

[0074] Figure 13 is a flowchart of the domain set generation process shown in step S41 of Figure 12. Initially, the domain set generation unit 48 calculates the probability p that the value of the piecewise variable is 1 based on past statistical information stored in the statistical information management unit 404. ｇ ｌ (t) is predicted (step S50). This makes it possible to predict which section l is most likely to be used in the linear relationship of each piece of equipment g.

[0075] The domain set generation unit 48 predicts the probability pｇ ｌ Based on (t), the piecewise variable Z ｇｌ The values ​​for the range (t) (∈{0,1}) are calculated, and a range set is constructed by combining them (step S51). Then, the range set generation unit 48 determines the priority of the decision variables of node c based on past statistical information and generates n possible rotation orders for node c (step S52). This is because the rotation order of node c may determine whether the range set is corrected successfully. By generating the rotation order of node c considering the priority of the decision variables of each node c, the range can be corrected starting with the decision variables with higher priority. Then, by ending the range correction process after correcting the decision variables with higher priority, the range set can be corrected quickly.

[0076] Next, the domain set generation unit 48 performs parallel processing to modify the domains included in the domain set so that the constraint conditions for the input and output physical quantities are satisfied in each cycle order (steps S53a to S53n), and then terminates the process shown in Figure 13. The process of modifying the domains by changing the cycle order of node c can be processed independently in parallel, so it is possible to modify the domains faster than processing them sequentially.

[0077] The configuration and effects of the original claims of this invention are described below.

[0078] [1] A system (controlled system 6) that combines multiple equipment and multiple nodes, including a conversion type equipment that takes one or more types of physical quantities as input and outputs one or more types of physical quantities, wherein predetermined types of output physical quantities output from the equipment to the nodes at each time and predetermined types of input physical quantities input from the nodes to the equipment are used as decision variables, and a node condition generation unit (43) generates constraint conditions for the nodes by relating the sum of the predetermined types of output physical quantities output from one or more equipment to the nodes to the sum of the predetermined types of input physical quantities input from the nodes to one or more equipment using an equals or inequality sign; a variable domain set generation unit (48) generates candidate variable domains that restrict the decision variables, modifies the candidate variable domains based on the constraint conditions for the nodes generated by the node condition generation unit (43), and stores them in a variable domain set; and a variable domain restriction unit (47) restricts the variable domains of the decision variables before optimizing the system using the variable domain set that stores the variable domains modified by the variable domain set generation unit (48). A device for planning equipment operation (4, 4A) characterized by being equipped with the following features.

[0079] This will enable faster planning of the operation of energy systems, including energy equipment.

[0080] [2] The equipment operation planning device (4, 4A) according to claim 1, characterized in that the domain set generation unit (48) divides the input physical quantity and expresses the output physical quantity as a linear function of each division, and the divisions in which the predicted probability of each division is equal to or greater than a predetermined probability are designated as candidate domains to restrict the decision variable.

[0081] This simplifies the relationship between input and output physical quantities, thereby reducing computational complexity. Furthermore, the range of variation can be selected using the predicted probability for each category.

[0082] [3] The equipment operation planning device (4, 4A) according to claim 2, characterized in that the domain set generation unit (48) modifies the domain that restricts the decision variable by lowering the predetermined probability when the constraint condition relating to the node does not hold.

[0083] This allows for the selection of variable ranges in order of highest predicted probability, enabling the creation of appropriate driving plans.

[0084] [4] The equipment operation planning device (4, 4A) according to claim 1, further comprising an optimal planning unit (46) that optimizes the system based on the constraints relating to each node.

[0085] This allows for system optimization.

[0086] [5] The equipment operation planning device (4, 4A) according to claim 4, characterized in that the range limiting unit (47) determines whether or not the range of the decision variable for each node can be limited based on the prediction results of the machine learning model.

[0087] This allows us to limit the decision variables for each node and quickly formulate a system operation plan.

[0088] [6] The equipment operation planning device (4, 4A) according to claim 5, characterized in that the range limiting unit (47) causes the machine learning model to learn the relationship between the range of the decision variable that has been limited in the past and the planning result at that time for each of the decision variables relating to the node.

[0089] This makes it possible to appropriately train the machine learning model on whether or not to restrict the decision variables of each node.

[0090] [7] The equipment operation planning device (4, 4A) according to 4, characterized in that the range limiting unit (47) determines whether or not to implement range limiting for each of the decision variables relating to each node based on past statistical information of the decision variables relating to each node.

[0091] This allows us to limit the decision variables for each node and quickly formulate a system operation plan.

[0092] [8] The equipment operation planning device (4, 4A) according to claim 7, characterized in that the range limiting unit (47) limits the range of the decision variable for each node based on the relationship between past range limits of the decision variable for each node and the planning results at that time, before the optimal planning unit (46) optimizes the system.

[0093] This allows us to limit the decision variables for each node and quickly formulate a system operation plan.

[0094] [9] The equipment operation planning device (4, 4A) according to claim 4, characterized in that the domain set generation unit (48) determines the priority of the decision variables when modifying the domain restrictions of each decision variable relating to the node based on the relationship between past domain restrictions of the decision variables relating to the node and the planning results at that time.

[0095] This allows us to restrict the range of the decision variables for each node in order of priority.

[0096]

[10] The equipment operation planning device (4, 4A) according to claim 1, characterized in that the domain set generation unit (48) determines the priority of each of the decision variables when modifying the domain constraints of each decision variable relating to the node, based on the prediction results of a machine learning model.

[0097] This allows us to restrict the range of the decision variables for each node in order of priority.

[0098]

[11] The equipment operation planning device (4, 4A) according to claim 1, characterized in that the domain set generation unit (48) prepares multiple node traversal sequences for modifying the domain set, and performs the modification of the domain set in each of the traversal sequences in parallel processing.

[0099] This allows for more reliable modification of the range set.

[0100]

[12] If the optimal planning unit (46) determines that it is not possible to perform the optimization of the system, the variable domain set generation unit (48) modifies the variable domain set, and the optimal planning unit (46) performs the optimization of the system again, characterized in that the equipment operation planning device (4, 4A) according to 4.

[0101] This allows the range set to be repeatedly modified until system optimization is performed.

[0102]

[13] A system (controlled system 6) that combines multiple equipment and multiple nodes, including a conversion type equipment that takes one or more types of physical quantities as input and outputs one or more types of physical quantities, wherein a node condition generation unit (43) generates constraint conditions for the node by relating the relationship between the sum of the predetermined types of output physical quantities output from one or more equipment to the node and the sum of the predetermined types of input physical quantities input from the node to the equipment using an equals sign or an inequality sign; a variable domain set generation unit (48) generates candidate variable domains that restrict the decision variables, and based on the constraint conditions for the node generated by the node condition generation unit (43), the variable domain set generation unit (48) modifies the candidate variable domains and stores them in a variable domain set; A method for formulating an equipment operation plan, comprising the step of limiting the range of the decision variable using the range set modified by the range set generation unit (48) before the range limiting unit (47) optimizes the system.

[0103] This will enable faster planning of the operation of energy systems, including energy equipment.

[0104] 《Modifications》 The present invention is not limited to the embodiments described above, and includes various modifications. For example, the embodiments described above are described in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. It is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace parts of the configuration of each embodiment with other configurations.

[0105] Each of the above configurations, functions, processing units, and processing means may be partially or entirely implemented in hardware, such as an integrated circuit. Each of the above configurations and functions may also be implemented in software by a processor interpreting and executing a program that implements each function. Information such as programs, tables, and files that implement each function can be stored in a recording device such as memory, a hard disk, or an SSD (Solid State Drive), or on a recording medium such as a flash memory card or a DVD (Digital Versatile Disk).

[0106] In each embodiment, the control lines and information lines shown are those deemed necessary for explanation and do not necessarily represent all control lines and information lines in the actual product. In practice, it can be assumed that almost all components are interconnected.

[0107] 400, 400A Equipment Operation Planning System 4, 4A Equipment Operation Planning Device 1 Constant Data Management and Distribution Device 2 Prediction Model Learning and Management Device 3 Operation and Monitoring Terminal 5 Equipment Control Device 6 Controlled System 41 Transformation-type Equipment Condition Generation Unit 42 Storage-type Equipment Condition Generation Unit 43 Node Condition Generation Unit 44 Constant Setting Unit 45 Objective Function Generation Unit 46 Optimal Planning Unit 47 Variable Range Restriction Unit 48 Variable Range Set Generation Unit 401 Optimal Planning Management Unit 402 Constant Data Management Unit 403 Equipment Operation Planning Program 404 Statistical Information Management Unit 51 Control Unit 52 Communication Unit 53 Control Data Management Unit 11 Demand Forecast Data Management Unit 12 Weather Forecast Data Management Unit 13 Equipment Data Management Unit 14 Acquisition Unit 15 Distribution Unit 71 Demand Forecast Data Distribution Server 73 Network 72 Weather forecast data distribution server 21 Learning unit 22 Forecast model management unit 31 Operation unit 32 Display unit 411 CPU 412 Memory 413 Secondary storage device 414 Input / output interface 415 Network interface 8 Simulation model 801 Power purchase equipment 802 Solar power generation equipment 803 Storage battery 807 Storage battery power conditioner 805 Power demand equipment 806 Solar power generation power conditioner 809 Node 810 Node 811 Node 9 Settings screen 91 Time granularity text box 92 Planned number of time slots 93 Speed ​​optimization application combo box 94 Upper limit γ text box

Claims

1. A system comprising multiple pieces of equipment and multiple nodes, including a conversion-type equipment that takes one or more types of physical quantities as input and outputs one or more types of physical quantities as output, wherein predetermined types of output physical quantities output from the equipment to the nodes at each time and predetermined types of input physical quantities input from the nodes to the equipment are used as decision variables, and a node condition generation unit generates constraint conditions for the nodes by relating the sum of the predetermined types of output physical quantities output from one or more pieces of equipment to the nodes to the sum of the predetermined types of input physical quantities input from the nodes to one or more pieces of equipment using an equals or inequality sign; a variable domain set generation unit generates candidate variable domains that restrict the decision variables, modifies the candidate variable domains based on the constraint conditions for the nodes generated by the node condition generation unit, and stores them in a variable domain set; and a variable domain limiting unit uses the variable domain set that stores the modified variable domains to restrict the variable domains of the decision variables before optimizing the system.

2. The equipment operation planning device according to claim 1, characterized in that the variable domain set generation unit divides the input physical quantity and expresses the output physical quantity as a linear function of each division, and the divisions in which the predicted probability of each division is equal to or greater than a predetermined probability are designated as candidate variable domains to restrict the decision variable.

3. The equipment operation planning device according to claim 2, characterized in that the variable domain set generation unit modifies the variable domain that restricts the decision variable by lowering the predetermined probability when the constraint condition relating to the node does not hold.

4. The equipment operation planning device according to claim 1, further comprising an optimal planning unit that optimizes the system based on the constraints relating to each node.

5. The equipment operation planning device according to claim 4, characterized in that the range limiting unit determines whether or not the range of the decision variable for each node can be limited based on the prediction results of the machine learning model.

6. The equipment operation planning device according to claim 5, characterized in that the range limiting unit causes the machine learning model to learn the relationship between the range of a decision variable that has been limited in the past and the planning result at that time for each of the decision variables relating to the node.

7. The equipment operation planning device according to claim 4, characterized in that the range limiting unit determines whether or not to implement range limiting for each of the decision variables relating to each node, based on past statistical information of the decision variables relating to each node.

8. The equipment operation planning device according to claim 7, characterized in that the range limiting unit limits the range of the decision variable for each node based on the relationship between past range limits of the decision variable for each node and the planning results at that time, before the optimal planning unit performs the optimization of the system.

9. The equipment operation planning device according to claim 4, characterized in that the domain set generation unit determines the priority of the decision variables when modifying the domain restrictions of each decision variable relating to the node, based on the relationship between past domain restrictions of the decision variables relating to the node and the planning results at that time.

10. The equipment operation planning device according to claim 1, characterized in that the domain set generation unit determines the priority of each decision variable when modifying the domain constraints of each decision variable relating to the node, based on the prediction results of a machine learning model.

11. The equipment operation planning device according to claim 1, characterized in that the domain set generation unit prepares multiple node traversal sequences for modifying the domain set, and performs the modification of the domain set in each of the traversal sequences in parallel processing.

12. If the optimal planning unit determines that it is unable to perform the optimization of the system, the variable domain set generation unit modifies the variable domain set, and the optimal planning unit performs the optimization of the system again, characterized in that the equipment operation planning device according to claim 4.

13. A method for planning equipment operation in a system that combines multiple equipment and multiple nodes, including a conversion type equipment that takes one or more types of physical quantities as input and outputs one or more types of physical quantities as output, wherein predetermined types of output physical quantities output from the equipment to the nodes at each time and predetermined types of input physical quantities input from the nodes to the equipment are used as decision variables, and a node condition generation unit generates constraint conditions for the nodes by relating the sum of predetermined types of output physical quantities output from one or more equipment to the nodes to the sum of predetermined types of input physical quantities input from the nodes to one or more equipment using an equals or inequality sign; a range set generation unit generates candidate ranges that restrict the decision variables, and based on the constraint conditions for the nodes generated by the node condition generation unit, the range set generation unit modifies the candidate ranges and stores them in a range set; and a range limiting unit uses the range set modified by the range set generation unit to limit the range of the decision variables before optimizing the system.

Images