Resource allocation system, resource allocation method, and program

Abstract

The purpose of the present disclosure is to optimize the allocation of a resource that is the power storage amount of a storage battery.　To this end, the present disclosure is a resource allocation system that allocates a resource, which is the power storage amount, of a storage battery. The resource allocation system includes: a resource allocation ratio selection unit that selects, on the basis of environment information, a prescribed distribution ratio of the power storage amount for a first individual charge / discharge plan generation device that generates a charge / discharge plan related to a first individual application and a second individual charge / discharge plan generation device that generates a charge / discharge plan related to a second individual application; and a power storage amount assignment unit that calculates, on the basis of the prescribed distribution ratio, a first power storage amount to be assigned to the first individual application and a second power storage amount to be assigned to the second individual application, and notifies the first individual charge / discharge plan generation device of the first power storage amount, and notifies the second individual charge / discharge plan generation device of the second power storage amount.

Description

Resource allocation system, resource allocation method, and program

[0001] The present disclosure relates to a technique for optimizing allocation of resources, which are electric power stored in a storage battery.

[0002] In recent years, there has been a trend to utilize energy storage resources such as storage batteries to adjust the gap between electricity demand and supply (supply and demand adjustment).For example, as a conventional technique, Non-Patent Document 1 describes the use of historical data from the electricity wholesale trading market to estimate the expected annual return and risk when allocating electricity to each commodity (resource allocation destination) and to optimize the portfolio based on the Capital Asset Pricing Model (CAPM), which is one of the modern portfolio theories in financial engineering.

[0003] Kazufumi Yuasa, Miki Ueshima, Tadatoshi Babasaki, Ichiro Omura, "Multi-objective Operation Strategy of Energy Storage Systems Based on Capital Asset Pricing Model," IEICE Transactions on Communications B J104-B (5), 2021-05-01, pp. 405-413

[0004] However, with conventional technology, portfolios are determined after calculating expected annual returns and risks, so it is not possible to optimally allocate (allocate) resources, i.e., the amount of energy stored in a storage battery, to each individual application, taking into account factors such as seasonal fluctuations and weather conditions. This creates the problem of opportunity losses occurring if too much energy is allocated to an individual application.

[0005] The present disclosure has been made in consideration of the above points, and aims to optimize allocation of resources, which are the amount of stored electricity in storage batteries.

[0006] In order to solve the above problem, the present disclosure provides a resource allocation system that allocates resources, which are the stored energy amounts of storage batteries, and includes: a resource allocation ratio selection unit that selects a predetermined allocation ratio of stored energy amounts based on environmental information for a first individual charge / discharge plan generation device that generates a charge / discharge plan for a first individual use and a second individual charge / discharge plan generation device that generates a charge / discharge plan for a second individual use; and a stored energy amount allocation unit that calculates a first stored energy amount to be allocated to the first individual use and a second stored energy amount to be allocated to the second individual use based on the predetermined allocation ratio, notifies the first individual charge / discharge plan generation device of the first stored energy amount, and notifies the second individual charge / discharge plan generation device of the second stored energy amount.

[0007] As described above, the present invention has the effect of optimizing the allocation of resources, which are the amounts of stored electricity in storage batteries.

[0008] FIG. 1 is a configuration diagram of a power system according to an embodiment. FIG. 2 is an electrical hardware configuration diagram of each system or device. FIG. 3 is a diagram showing an example of the data structure of a resource allocation table. FIG. 4 is a diagram showing initial values ​​of weights in the resource allocation table. FIG. 5 is a diagram showing an example of allocation ratios in the resource allocation table. FIG. 6 is a diagram showing the concept of an avoidable loss amount. FIG. 7 is a sequence diagram showing processing executed by a resource allocation system and an individual charging and discharging plan generation device.

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0010] [System Configuration of the Embodiment] First, the overall configuration of the power system of the present embodiment will be described with reference to Fig. 1. Fig. 1 is a configuration diagram of the power system according to the embodiment.

[0011] 1, a power system 10 of this embodiment is configured by a resource allocation system 20, a storage battery system 60, and a plurality of individual charge and discharge plan generation devices 90a, 90b, and 90c. The resource allocation system 20, the storage battery system 60, and the plurality of individual charge and discharge plan generation devices 90a, 90b, and 90c can communicate with each other via a communication network such as the Internet or a LAN (Local Area Network). The communication network may be connected wirelessly or via a wired connection.

[0012] Note that a single business operator may manage and operate the entire power system 10, or multiple business operators may manage and operate the resource allocation system 20, the storage battery system 60, the individual charge and discharge plan generation device 90a, the individual charge and discharge plan generation device 90b, and the individual charge and discharge plan generation device 90c, respectively. In this case, the business operator of the resource allocation system 20 is an aggregator (a specified wholesale supplier) or the like. The business operators of the storage battery system are electric power companies, power generation businesses, or the like. The business operators of the individual charge and discharge plan generation devices 90a, 90b, and 90c are retail electricity businesses, or the like.

[0013] Furthermore, each of the individual charge and discharge plan generating devices 90a, 90b, and 90c has a charge and discharge plan generating engine for an individual application. Although three individual charge and discharge plan generating devices 90a, 90b, and 90c are shown in FIG. 1, the number may be two, or four or more. Furthermore, hereinafter, the individual charge and discharge plan generating devices 90a, 90b, and 90c will be collectively referred to as the "individual charge and discharge plan generating device 90."

[0014] The resource allocation system 20 is a system that optimally allocates the amount of electricity (resources) stored in the storage battery B to each individual charge / discharge plan generation device 90 by reinforcement learning.

[0015] The storage battery system 60 is constructed by a storage battery B, a storage battery information collection device 70, and a charge / discharge plan management device 80.

[0016] The storage battery B may be a single battery or multiple batteries. Although FIG. 1 shows a single storage battery B, multiple storage batteries may also be used. The storage battery B stores surplus electricity generated by solar panels or other devices at electric power companies or private homes. The electricity stored in the storage battery B may be electricity produced by renewable energy sources such as sunlight, or may be electricity produced by non-renewable energy sources such as petroleum.

[0017] The battery information collecting device 70 collects information related to the current amount of stored power in the storage battery B from the storage battery B. The charge / discharge plan management device 80 allocates and supplies power to be used to each of the individual uses u1, u2, and u3 based on the charge / discharge plan information acquired from each individual charge / discharge plan generating device 90a and information such as the remaining power of the storage battery B acquired from the storage battery B. In this embodiment, the individual uses u1, u2, and u3 have different charge / discharge cycles, but they may have the same charge / discharge cycle. While FIG. 1 shows three individual uses u1, u2, and u3 as an example, this is not limiting. There may be more than one individual use, and the number may be two, four, or more. Hereinafter, the individual uses u1, u2b, and u3 will be collectively referred to as "individual use u."

[0018] The individual use u indicates any one of a demand response (DR) use in the capacity market, a spot market (power sales) use in the wholesale electricity market, a tertiary adjustment reserve use in the demand adjustment market, and a backup use for information and communication technology (ICT) services, etc. For example, the individual uses u1, u2, and u3 may be the same use or may be different uses.

[0019] The individual charge and discharge plan generation device 90 corresponds one-to-one to each individual application u, and generates a charge and discharge plan for each individual application u using existing technology. The individual charge and discharge plan generation device 90 also receives feedback of information such as a buying and selling history, a penalty amount, a reward, and a remaining amount of stored electricity as an operation result from the individual application u corresponding to the individual charge and discharge plan generation device 90 itself. In FIG. 1 , the individual charge and discharge plan generation device 90a corresponds to the individual application u1, the individual charge and discharge plan generation device 90b corresponds to the individual application u2, and the individual charge and discharge plan generation device 90c corresponds to the individual application u3.

[0020] The resource allocation system 20, the battery system 60, the battery information collection device 70, the charge and discharge plan management device 80, and the individual charge and discharge plan generation device 90 are each configured by one or more computers as shown in FIG. 2.

[0021] [Hardware Configuration] Next, the electrical hardware configuration of the resource allocation system 20 will be described using Fig. 2. The resource allocation system 20 can be realized, for example, by causing a computer to execute a program. Note that the storage battery information collection device 70, the charge / discharge plan management device 80, and the individual charge / discharge plan generation device 90 have the same configuration as those of the resource allocation system 20, and therefore descriptions thereof will be omitted.

[0022] The resource allocation system 20 can be realized by using hardware resources such as a CPU and memory built into a computer to execute a program corresponding to the processing performed by the device. The program can be recorded on a computer-readable recording medium (such as a portable memory) and stored or distributed. The program can also be provided over a network by email or the like.

[0023] As shown in FIG. 2, the resource allocation system 20 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a CPU 104, an interface device 105, a display device 106, an input device 107, an output device 108, and the like, all of which are interconnected by a bus BS.

[0024] The program that realizes the processing on the computer is provided by a recording medium 101, such as a CD-ROM or a memory card. When the recording medium 101 storing the program is set in the drive device 100, the program is installed from the recording medium 101 to the auxiliary storage device 102 via the drive device 100. However, the program does not necessarily have to be installed from the recording medium 101, but may be downloaded from another computer via a communication network. The auxiliary storage device 102 stores the installed program as well as necessary files, data, etc.

[0025] The memory device 103 reads and stores the program from the auxiliary storage device 102 when an instruction to start the program is received. The CPU 104 realizes the functions related to the device in accordance with the program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a communication network, etc. The display device 106 displays a GUI (Graphical User Interface) etc. according to the program. The input device 107 is composed of a keyboard, mouse, buttons, touch panel, etc., and is used to input various operation instructions. The output device 108 outputs the results of calculations.

[0026] [Outline of Functional Configuration of Resource Allocation System] Returning to FIG. 1, an outline of the functional configuration of the resource allocation system 20 will be described.

[0027] As shown in Fig. 1, the resource allocation system 20 has an initial setting unit 30, an inference unit 40, and a learning unit 50. Each of these units has a function realized by an instruction from the CPU 104 in Fig. 2 based on a program. A resource allocation DB (Data Base) 45 is also constructed in the storage unit, which is the auxiliary storage device 102 or the memory device 103. In the resource allocation table DB 45, a resource allocation table T is constructed as a Q table for Q learning, which is an example of reinforcement learning.

[0028] The initial setting unit 30 performs processing when the resource allocation system 20 starts operation. For this purpose, the initial setting unit 30 includes a resource allocation table definition unit 31 and a weight setting unit 32.

[0029] The resource allocation table definition unit 31 receives setting information from an administrator of the resource allocation system 20, etc., and also acquires environmental information from the Japan Meteorological Agency, etc., storage capacity information indicating the capacity of storage battery B from the storage battery system 60, and information on each individual application from the individual charge / discharge plan generation device 90, and defines row information (a set of state vectors (an example of state observation information)) and column information (a set of allocation ratio vectors (also referred to as "allocation ratios")) in the resource allocation table DB 45, which will be described later. The "setting information" includes the minimum unit for allocating storage battery B to individual application u, a "capacity reservation contract capacity (also referred to as "contract capacity")" indicating the capacity that must be reserved for storage battery B according to a prior contract, and a "backup maintenance time" indicating the time that power must be maintained as a backup. The setting information is input manually by the administrator of the resource allocation system or automatically.

[0030] The storage capacity of storage battery B is an example of "storage battery information" as information related to the handled resource. The contract capacity and backup maintenance time are examples of information related to individual applications.

[0031] The weight setting unit 32 sets initial weights (ease of selection of allocation ratios) for the elements of the resource allocation table T (values ​​specified by row information and column information) and stores them in the resource allocation table T.

[0032] The inference unit 40 performs processing on or before the target date of resource allocation, and infers the amount of storage energy to be allocated to each individual use u. To this end, the inference unit 40 includes an input information receiving unit 41, a resource allocation ratio selecting unit 42, and a storage energy allocating unit 43.

[0033] The input information receiving unit 41 collects information on the environmental information, the amount of stored power that can be handled (remaining stored power), and individual uses on the target day for resource allocation, maps it to a state vector to be considered by the resource allocation system 20, and identifies the corresponding state vector. For example, if the weather is good on the target day for resource allocation and a large amount of power can be handled, not only can a large amount of power be discharged but also a large amount of power can be charged on the target day, and as a result, the amount of power handled on the target day may exceed the storable capacity, for example, 120%.

[0034] The resource allocation ratio selection unit 42 selects an allocation ratio of the resource, which is the amount of stored power, in accordance with the weights written in the resource allocation table T, based on the state vector specified by the input information reception unit 41 .

[0035] The storage capacity allocation unit 43 calculates the storage capacity to be allocated to each individual use u based on the allocation ratio selected by the resource allocation ratio selection unit 42, and notifies each individual charge / discharge plan generation device 90 of the allocated storage capacity (electricity amount).

[0036] The learning unit 50 performs machine learning using information on the operation results of resource allocation. Therefore, the learning unit 50 performs processing after the target date of resource allocation (i.e., the date on which the result of resource allocation becomes known). For this purpose, the learning unit 50 includes a charge / discharge result collection unit 51, a comprehensive evaluation calculation unit 52, and a resource allocation table update unit 53.

[0037] The charge / discharge result collecting unit 51 collects, from each individual charge / discharge plan generating device 90, information on the operation results of the power storage amount allocated to each individual use u by the power storage amount allocating unit 43.

[0038] The overall evaluation calculation unit 52 calculates, for each individual charge / discharge plan generation device 90, a penalty amount corresponding to the remaining amount of stored power that is ultimately not consumed even though the individual charge / discharge plan generation device 90 has allocated power, based on the information on the operation results collected by the charge / discharge result collection unit 51. Furthermore, the overall evaluation calculation unit 52 calculates a total remuneration value (overall evaluation value) for the target day of resource allocation by subtracting the penalty amount from the remuneration (amount).

[0039] The resource allocation table update unit 53 updates the weights in the resource allocation table T for the state vector specified on the target date for resource allocation and the selected allocation ratio vector based on the equations described below.

[0040] [Details of Functional Configuration of Resource Allocation System] Next, the functional configuration of the resource allocation system will be described in detail.

[0041] <Resource Allocation Table Definition Unit> The resource allocation table definition unit 31 acquires information on the environment, storage battery information, and individual uses targeted by the resource allocation system 20, and defines row information (a set of state vectors) and column information (a set of allocation ratio vectors) of the resource allocation table. Here, examples of the environment information, storage battery information, and information on individual uses are shown below. Note that all of this information is information on the target date for resource allocation (including forecast information).

[0042] ) (daily maximum or minimum value), backup maintenance time Next, the process of generating each vector by the resource allocation table definition unit 31 will be explained.

[0043] (Row Information (State Vector Set) Generation Process) The resource allocation table definition unit 31 provides rules for converting input data and features to be handled by the resource allocation system 20 based on the environmental information, storage battery information, and information on individual uses. The following is an example of a conversion rule. Information on time: The resource allocation table definition unit 31 converts to features based on the date d of the target day for resource allocation.

[0044] The seasonal information may be, for example, 12 patterns from January to December, or 4 patterns from spring, summer, autumn and winter.

[0045] The day of the week information may have seven patterns, for example, Monday to Sunday. Alternatively, two patterns may be used, for example, weekdays and holidays. Weather information: The resource allocation table definition unit 31 converts the weather forecast for the target day of resource allocation into a feature.

[0046] For example, there are three weather patterns (atmospheric conditions): sunny, rainy, and cloudy. Eleven patterns ranging from 0% to 1000% may be used based on the probability of precipitation. Other information with continuous values: The resource allocation table definition unit 31 converts this information into feature quantities based on various forecast data for the target day of resource allocation.

[0047] The maximum or minimum temperature forecast for a target day, the total area power generation or demand forecast for a target day, etc. are discrete values ​​with intervals characterized by a minimum value, a maximum value, or an increment width that can be arbitrarily specified. The value range or increment width may be changed based on statistics.

[0048] Furthermore, if the temporal granularity of the information provided as input information (data) differs from the resource allocation decision cycle, the resource allocation table definition unit 31 may convert the respective granularities by applying statistical processing, etc. For example, if the resource allocation decision cycle is one day and the input data is 30 minutes for 24 hours (48 points), the maximum or minimum value of the 48 points, or the average value of the 48 points, is used as a feature. Time-series features (increasing trend, decreasing trend, convex shape, concave shape, etc.) extracted from the 30-minute 24-hour data may also be used.

[0049] Furthermore, the resource allocation table definition unit 31 combines one or more feature quantities extracted by the above processing to generate a state vector sd When the resource allocation table definition unit 31 identifies the extracted N types of feature quantities with index i, the state vector of date d is expressed as follows: Furthermore, a set consisting of all possible values ​​of the state vector is defined as a state vector set.

[0050] State vector: s d ∈{(x d,1 ,x d,2 ,…,x d,N )|x d,i is the feature of item i} (Example of row information (state vector set) generation process) When day of the week information and total power generation amount in the area for one day (minimum value: 0 kWh, maximum value: 1 million kWh, increment: 100,000 kWh, 10 intervals of 0 kWh range, 10 kWh range, 20 kWh range, ..., 90 kWh range) are used as feature amounts, state vector set = {"(day of the week, total power generation amount in the area for one day)"} = {(Monday, "0 kWh range"), (Monday, "10 kWh range"), (Monday, "20 kWh range"), ..., (Sunday, "80 kWh range"), (Sunday, "90 kWh range")} (Column information (allocation ratio vector set) generation process) The resource allocation table definition unit 31 generates an allocation ratio vector set based on the number of individual uses u. Each individual use is represented by i = {1, 2, ..., N}. If the allocation ratio is a discrete value with M+1 levels, the probability vector (p1, p2, ..., p N ) is an allocation ratio vector, which is denoted as a for convenience. Also, the set of all possible values ​​of the allocation ratio vector is the allocation ratio vector set.

[0051] ・Probability vector:

[0052] ・Allocation ratio vector: a∈{(p1, p2, ..., p N )|p i is the allocation ratio of use i} (Example of column information (allocation ratio vector set) generation process) When three individual uses are set as resource allocation destinations and the allocation ratio is a four-level discrete value Allocation ratio vector set: {"(allocation ratio of individual use u1, allocation ratio of individual use u2, allocation ratio of individual use u3)"} =

[0053] (Resource Allocation Table Generation Process) The resource allocation table definition unit 31 defines the resource allocation table T by treating the state vector set and allocation ratio vector set generated by the above process as row (vertical) information and column (horizontal) information, respectively. For example, the resource allocation table definition unit 31 expresses an element of the resource allocation table T in the state vector s and allocation ratio vector a as T(s, a).

[0054] (Example of Resource Allocation Table Generation Process) Based on the above-described example of row information (state vector set) generation process and example of column information (allocation ratio vector set) generation process, the resource allocation table generated is as shown in Fig. 3. Fig. 3 is a diagram showing an example of the data structure of the resource allocation table.

[0055] <Weight Setting Unit> The weight setting unit 32 sets initial weights (ease of selection of allocation ratios) for elements (values ​​specified by row information and column information) of the resource allocation table T, and stores them in the resource allocation table DB 45.

[0056] (Weight Generation Process) The value of an element of the resource allocation table T(s, a) in the state vector s and the allocation ratio vector a is called the weight w(s, a). Here, the initial weight for all elements is set to 0 uniformly. An example is shown in FIG. 4. FIG. 4 is a diagram showing the initial values ​​of weights in the resource allocation table. The initial date d used to identify the resource allocation table T is set to 0. The generated resource allocation table may be edited if the user of the resource allocation system 20 wishes to arbitrarily change the initial weight. When the resource allocation system 20 suspends and resumes allocation operations, it may utilize a resource allocation table obtained from a previous operation (an updated resource allocation table).

[0057] (Row (Vertical) Information (State Vector Set) Addition Process) When new input data is utilized during operation of the resource allocation system 20, the weight setting unit 32 can expand the state vector set.

[0058] When machine learning progresses and history is accumulated, the weight setting unit 32 improves performance by further refining the state vector.

[0059] The state vector refinement is performed on information having continuous values. The weight setting unit 32 holds the number of observations of the corresponding state vector in a table of the same format as the resource allocation table T, and divides the state vectors that have been observed a number of times equal to or greater than a predetermined threshold. In this case, the weight setting unit 32 sets the weights before the row information is expanded for all values ​​of the added feature quantities, with respect to the weights in the resource allocation table T with expanded row information.

[0060] (Example of Row Information (State Vector Set) Addition Processing) When the feature quantity of item 3 is extended by new input data to the existing feature quantities of item 1 and item 2, the weight setting unit 32 sets the weights in the extended resource allocation table T as follows. Here, the existing weights in the state vector s=(x1, x2) and the allocation ratio vector a are set as w s,a When the state vector expanded from item 3 is s'=(x1, x2, x3), the new weight w s',a is given as follows:

[0061] For all possible values ​​of x3, w s',a = w s,a Let's say.

[0062] (Column Information (Allocation Ratio Vector Set) Addition Process) When a new individual use is added during operation of the resource allocation system 20, the weight setting unit 32 can extend the allocation ratio vector set.

[0063] When machine learning progresses and a history is accumulated, the weight setting unit 32 improves performance by further refining the allocation ratios. For example, the weight setting unit 32 maintains the number of times a corresponding allocation ratio has been selected in a table of the same format as the resource allocation table T, and divides allocation ratio vectors that have been selected a number of times equal to or greater than a predetermined threshold. In this case, the weight setting unit 32 sets the weights in the resource allocation table T with expanded column information to be consistent with the weights before the column information was expanded.

[0064] (Example of column information (allocation ratio vector set) addition process) When a new individual use u3 is added to the existing individual uses u1 and u2, the weight setting unit 32 sets the weights in the expanded resource allocation table T as follows. Here, the existing weights in the state vector s and the allocation ratio vector a=(p1, p2) are set as w s,a When the distribution ratio vector obtained by expanding item 3 is a' = (p1, p2, p3), the new weight w s,a' is given as follows:

[0065] When p3 = 0, w s,a' =w s,a Let's say.

[0066] When p3≠0, w s,a' =C, where C is an arbitrary value predetermined by the user of the resource allocation system 20.

[0067] <Input Information Receiving Unit> The input information receiving unit 41 receives, as input, environmental information and storage battery information (amount of stored power) on the target date for resource allocation, and identifies the corresponding state vector.

[0068] (Input Information Reception Process) The input information reception unit 41 refers to the resource allocation table definition unit for input information. It is assumed that the input information is provided in a CSV (comma separated values) file or the like each time the storage capacity allocation unit 43 calculates resource allocation. Separate CSV files may be provided for each type of input information, or an API (Application Programming Interface) or the like may be constructed.

[0069] (State Vector Identification Process) Similar to the resource allocation table definition unit 31, the input information reception unit 41 identifies a state vector from the input information (data) based on the conversion rules given in the environmental information and storage battery information.

[0070] (Example of state vector identification processing) When using day of the week information and total area power generation for one day (minimum value: 0 kWh, maximum value: 1 million kWh, increment: 100,000 kWh, 10 intervals: 0 kWh, 10 kWh, 20 kWh, ..., 90 kWh) as feature quantities, when the input information "total area power generation for Saturday, July 1, 2023 = 640,000 kWh" is given, the state vector becomes "(Saturday, 60 kWh)".

[0071] ・When the input information is "Total power generation in the area on Sunday, July 2, 2023 = 880,000 kWh," the state vector becomes "(Sunday, 80 kWh range)."

[0072] ・When the input information is "Total area power generation on Monday, July 3, 2023 = 720,000 kWh," the state vector becomes "(Monday, 70 kWh range)."

[0073] <Resource Allocation Ratio Selector> The resource allocation ratio selector 42 selects an allocation ratio in accordance with the weights listed in the resource allocation table T based on the state vector specified by the input information receiver 41 .

[0074] (Distribution Ratio Selection Process) The resource distribution ratio selection unit 42 selects the state vector s d Based on this, the resource allocation ratio vector a is calculated according to the following formula: d (However, a d is one of the distribution ratio vector sets), where N represents the number of individual uses u to which the storage capacity is allocated, and ∈ is a parameter given separately.

[0075] (Example of allocation ratio selection process) Fig. 5 is a diagram showing an example of allocation ratios in a resource allocation table. When the state vector is specified as "(Sunday, 80 kWh range)", the weights of the resource allocation table T are as follows:

[0076] From the resource allocation table shown in Figure 5, when a is the following value, maxT(s d , a)=30.

[0077] For example, if ∈=0.1, the resource allocation ratio selection unit 42 selects the following a with a probability of 0.91 as shown in FIG. d The allocation ratio is selected, and there is a probability of 0.01 that another allocation ratio will be selected.

[0078] <Storage Amount Allocation Unit> The storage amount allocation unit 43 calculates each storage amount to be allocated to each individual use u based on the allocation ratio selected by the resource allocation ratio selection unit 42, and notifies each storage amount to the individual charge / discharge plan generation device 90 corresponding to each individual use u.

[0079] (Allocated storage amount calculation process) The storage amount allocation unit 43 calculates the storage amount C that can be used on date d. d When the resource allocation ratio selection unit 42 selects the allocation ratio a d =(p d,1 ,p d,2 ,…,p d,N ), the following storage capacity is allocated to each individual use i={1,2,...,N}:

[0080] The storage allocation for individual use i = p d,i C d is.

[0081] (Example of calculation process for allocated storage capacity) The available storage capacity on date d is 300 kWh, and the following d The allocation ratio is selected as follows.

[0082] In this case, the power storage allocation unit 43 allocates 100 kWh to the individual use u1, 0 kWh to the individual use u2, and 200 kWh to the individual use u3.

[0083] (Process of notifying allocated power storage amount) The power storage amount allocating unit 43 notifies each individual charge / discharge plan generating device 90 corresponding to the individual use u of the allocated power storage amount. For example, the power storage amount allocating unit 43 notifies the individual charge / discharge plan generating device 90a corresponding to the individual use u1 to which the power amount (100 kWh) has been allocated of the power storage amount (100 kWh).

[0084] <Charge / Discharge Result Collection Unit> The charge / discharge result collection unit 51 collects, from each individual charge / discharge plan creation device 90, the operation results (remuneration, remaining amount of stored power) of the allocated amount of stored power.

[0085] (Charge / Discharge Result Collection Process) The charge / discharge result collection unit 51 collects the operation results (reward, remaining amount of stored electricity) of the amount of stored electricity on date d from each individual charge / discharge plan creation device 90. The reward for the operation results of the amount of stored electricity of the charge / discharge plan creation device i on date d is calculated as r d,i Furthermore, the remaining amount of stored power (remaining stored power) that is not consumed among the amount of stored power allocated to the charge / discharge plan generation device i on date d is expressed as follows:

[0086] It is assumed that the operation results of the amount of stored power are transferred in a CSV file or the like. A separate CSV file may be provided for each individual charge / discharge plan generation device 90, or an API or the like may be constructed. If the individual charge / discharge plan generation device 90 cannot be linked systematically, the user of the resource allocation system 20 sets the information manually.

[0087] (Definition of remuneration) (a) Remuneration in the capacity market: The daily rate of the contract amount for securing capacity. When DR is activated, the amount shall be the daily rate minus an economic penalty. (b) Remuneration in the wholesale electricity market: The amount obtained by subtracting a commission from the purchase price. (c) Remuneration in the demand adjustment market: The amount settled from the contract fee, penalty fee, etc. (d) Remuneration for backup use: An amount arbitrarily determined by the user of the resource allocation system 20. When used as a backup power source, if the power supply is cut off due to a shortage of allocated power storage, the amount shall be the amount arbitrarily determined by the user mentioned above minus the amount of damages caused by the shutdown of the battery storage system 60.

[0088] Next, the overall evaluation calculation unit 52 will be described.

[0089] <Comprehensive evaluation calculation unit> The comprehensive evaluation calculation unit 52 calculates a penalty corresponding to the remaining amount of stored power, and calculates a total remuneration value for the target day for the actual allocation of the amount of power (resources). Note that Fig. 6 is a diagram showing the concept of the amount of avoidable loss.

[0090] (Penalty calculation process for remaining power storage) Remaining power storage amount for individual use i on date d

[0091] The totals are shown below.

[0092] The set of individual uses i for which the remaining amount of stored energy is zero and the reward is a positive value is shown below.

[0093] At this time,

[0094] The definition of opportunity loss is shown below.

[0095] Similarly, the set of individual uses i for which the remaining energy storage is zero and the reward is a negative value is shown below.

[0096] At this time,

[0097] The definition of avoidable loss is as follows:

[0098] However, F d,i is the necessary and sufficient discharge amount for individual use i. For example, the necessary and sufficient discharge amount in the capacity market (DR compatible) is given by the contract capacity.

[0099] Based on the above calculations, the comprehensive evaluation calculation unit 52 calculates

[0100] The maximum opportunity loss amount for

[0101] The larger of the maximum avoidable loss amount for d,L Let's say.

[0102] (Total Reward Calculation Process) The overall evaluation calculation unit 52 calculates w1, ..., w N ,w L is a weight (importance for each use, a value between 0 and 1) arbitrarily set by the user of the resource allocation system 20, and the total reward value r is calculated by the following formula: d Calculate.

[0103] <Resource allocation table update unit> The resource allocation table update unit 53 updates the weights for the state vector identified on the target date for actual resource allocation and the allocation ratio vector selected by the resource allocation ratio selection unit 41 based on the following update formula.

[0104] (Resource Allocation Table Update Processing) The resource allocation table update unit 53 updates the state vector s d , and the allocation ratio vector a selected by the resource allocation ratio selection unit 41 d , and the total reward value r d Based on this, the resource allocation table T is updated using the following update formula:

[0105] Here, Q(s d , a d ) is the state vector s d and the distribution ratio vector a d The parameter α∈{0,1} represents the learning rate; the closer α is to 1, the more importance is placed on the most recent past situation. The parameter γ∈{0,1} represents the discount rate; the closer γ is to 0, the more future value is discounted in the calculation.

[0106] The above update formula is modeled after Q-learning in reinforcement learning, but this update formula may be replaced with other update formulas in reinforcement learning (Sarsa, Monte Carlo method).

[0107] [Processing or Operation of the Embodiment] Next, processing or operation of the embodiment will be described with reference to Fig. 7. Fig. 7 is a sequence diagram showing processing executed by the resource allocation system and the individual charging and discharging plan generation device.

[0108] S11: The input information receiving unit 41 collects, for example, environmental information and the like relating to the target day for resource allocation, which is the following day, maps it to a state vector that the resource allocation system 20 should take into consideration, and identifies it as a situation vector.

[0109] S12: The resource allocation ratio selection unit 42 calculates a resource allocation ratio using an allocation ratio corresponding to the specified situation, based on the resource allocation table T (a correspondence table between state vectors and allocation ratio vectors) managed in the resource allocation table DB 45. Furthermore, the power storage amount allocation unit 43 allocates the power storage amount of the storage battery B to the charge / discharge plan generation device 90 corresponding to each individual use u, based on the resource allocation ratio calculated by the resource allocation ratio selection unit 42.

[0110] S13: The storage energy allocation unit 43 transmits the allocated storage energy to each individual charge and discharge plan generating device 90. As a result, each individual charge and discharge plan generating device 90 acquires the allocated storage energy.

[0111] S14: Each individual charge / discharge plan generation device 90 generates a charge / discharge plan based on the allocated amount of stored power. The contents of this plan are not made public. Then, each individual charge / discharge plan generation device 90 instructs the charge / discharge plan management device 80 to charge / discharge (especially discharge) according to the contents of the charge / discharge plan. As a result, the charge / discharge plan management device 80 carries out charging / discharging according to the contents of each charge / discharge plan for each individual use u.

[0112] S15: Each individual charge / discharge plan generation device 90 receives feedback of operation results including charge / discharge results from the individual use u corresponding to the device itself, and transmits the feedback to the charge / discharge result collection unit 51 of the resource allocation system 20. As a result, the charge / discharge result collection unit 51 collects operation results including charge / discharge results from each individual charge / discharge plan generation device 90.

[0113] S16: The comprehensive evaluation calculation unit 52 calculates, for each individual charge and discharge plan generation device 90, a penalty amount corresponding to the remaining amount of stored power that is not ultimately consumed even though the individual charge and discharge plan generation device 90 has allocated power.

[0114] S17: The overall evaluation calculation unit 52 calculates the total reward value (overall evaluation value) for the target day of resource allocation by subtracting the penalty amount from the reward (amount). Furthermore, the resource allocation table update unit 53 updates the weights in the resource allocation table T for the state vector specified for the target day of resource allocation and the selected allocation ratio vector.

[0115] 7 are repeated during operation, but can be interrupted midway. When operation is interrupted and then resumed, the resource allocation system 20 may utilize a resource allocation table (updated resource allocation table) obtained during previous operation, as described above.

[0116] [Major Effects of the Embodiment] As described above, the resource allocation system 20 of the present embodiment has the effect of optimizing the allocation of resources, which are the amounts of stored power in storage batteries, in consideration of environmental information and the like.

[0117] In addition, the resource allocation system 20 utilizes reinforcement learning to update the resource allocation table T, which is a Q table, thereby achieving the effect of quickly treating a newly established electricity trading market as one of the resource allocation destinations for stored energy capacity.

[0118] Furthermore, the resource allocation system 20 can cooperate with each individual charge / discharge plan generation device 90 .

[0119] In addition, the resource allocation system 20 can feed back actual operating results (transaction results) to the resource allocation plan at any time by feeding back the actual charging and discharging results for each individual use u via each individual charging and discharging plan generation device 90.

[0120] [Supplementary Information] The present disclosure is not limited to the above-described embodiments and may have the following configurations or processes (operations). (1) The resource allocation system 20 can be realized by a computer and a program, but the program can also be recorded on a (non-transitory) recording medium or provided via a communication network. (2) The CPU 104 as a processor may be single or multiple. (3) The resource allocation table T described above is an example of resource allocation information. In other words, the resource allocation table T is merely one form, and is not necessarily limited to the form of a table as long as it can manage both the state vector (state observation information) and the allocation ratio vector (allocation ratio). Similarly, the resource allocation table definition unit 31 is an example of a resource allocation information definition unit, the resource allocation table DB 45 is an example of a resource allocation information DB, and the resource allocation table update unit 53 is an example of a resource allocation information update unit.

[0121] REFERENCE SIGNS LIST 10 Power system 20 Resource allocation system 30 Initial setting unit 31 Resource allocation table definition unit 32 Weight setting unit 40 Inference unit 41 Input information reception unit 42 Resource allocation ratio selection unit 43 Power storage amount allocation unit 45 Resource allocation table DB 50 Learning unit 51 Charging / discharging result collection unit 52 Overall evaluation calculation unit 53 Resource allocation table update unit 60 Rechargeable battery system 70 Storage battery information collection device 80 Charging / discharging plan management device 90 Individual charging / discharging plan generation device 90a Individual charging / discharging plan generation device 90b Individual charging / discharging plan generation device 90c Individual charging / discharging plan generation device T Resource allocation table u1 Individual use u2 Individual use u3 Individual use

Claims

1. A resource allocation system for allocating a resource which is the amount of power stored in a storage battery, comprising: a resource allocation ratio selection unit that selects a predetermined allocation ratio of the amount of power stored based on environmental information, for a first individual charge / discharge plan generation device that generates a charge / discharge plan for a first individual use and a second individual charge / discharge plan generation device that generates a charge / discharge plan for a second individual use; and a power storage amount allocation unit that calculates a first amount of power stored to be allocated to the first individual use and a second amount of power stored to be allocated to the second individual use based on the predetermined allocation ratio, notifies the first individual charge / discharge plan generation device of the first amount of power stored, and notifies the second individual charge / discharge plan generation device of the second amount of power stored.

2. The resource allocation system according to claim 1, wherein the resource allocation ratio selection unit selects the predetermined allocation ratio based on the environmental information and battery information including the amount of power stored that can be handled by the storage battery.

3. The resource allocation system according to claim 1 or 2, further comprising: a storage unit that stores resource allocation information configured by a set of state observation information defined from each of a plurality of the environmental information and a set of allocation ratios defined based on each of the first individual use and the second individual use, with weights for reinforcement learning set; and an input information reception unit that inputs predetermined environmental information for a target day of resource allocation and identifies predetermined state observation information related to the predetermined environmental information. The resource allocation ratio selection unit selects the predetermined allocation ratio according to the weights set in the resource allocation information based on the predetermined state observation information.

4. The resource allocation system according to claim 2, A storage unit that stores resource allocation information configured by weights for reinforcement learning, a set of state observation information defined from each of the plurality of pieces of environmental information and the battery information, and a set of allocation ratios defined based on each of the first individual use and the second individual use; An input information receiving unit that inputs predetermined environmental information and predetermined battery information for the target day of resource allocation and specifies predetermined state observation information related to the predetermined environmental information and the predetermined battery information; The resource allocation ratio selection unit selects the predetermined allocation ratio according to the weight set in the resource allocation information based on the predetermined state observation information. A resource allocation system.

5. The resource allocation system according to claim 3, A charge / discharge result collection unit that collects a first reward and a first remaining battery amount from the first individual charge / discharge plan generation device as the charge / discharge result of the first individual use; An overall evaluation calculation unit that calculates a first total reward value for the target day of resource allocation based on a first penalty amount corresponding to the first reward and the first remaining battery amount; A resource allocation information update unit that updates the weight set in the resource allocation information based on the predetermined state observation information, the predetermined allocation ratio, and the first total reward value. A resource allocation system having the above.

6. The charge / discharge result collection unit collects a second reward and a second remaining battery amount from the second individual charge / discharge plan generation device as the charge / discharge result of the second individual use, The overall evaluation calculation unit calculates a second total reward value for the target day of resource allocation based on a second penalty amount corresponding to the second reward and the second remaining battery amount, The resource allocation information update unit updates the weight set in the resource allocation information based on the predetermined state vector, the predetermined allocation ratio, and the second total reward value. The resource allocation system according to claim 5.

7. A resource allocation method executed by a resource allocation system that allocates resources, which are the power storage amounts of storage batteries, wherein the resource allocation system performs a resource allocation ratio selection process of selecting a predetermined allocation ratio of the power storage amount based on environmental information for a first individual charge / discharge plan generation device that generates a charge / discharge plan for a first individual use and a second individual charge / discharge plan generation device that generates a charge / discharge plan for a second individual use, and based on the predetermined allocation ratio, calculates a first power storage amount to be allocated to the first individual use and calculates a second power storage amount to be allocated to the second individual use, notifies the first power storage amount to the first individual charge / discharge plan generation device, and notifies the second power storage amount to the second individual charge / discharge plan generation device. A resource allocation method that executes the above.

8. A program for causing a computer to execute the method according to claim 7.

Images