JP2019091131A - Facility operation support device, and method and system thereof - Google Patents

Abstract

To provide a facility operation support device for supporting to select a facility which can possibly achieve restriction of a necessary amount of power, as a demand response object.SOLUTION: A learning unit 211 of a facility operation support device acquires demand response implementation related data and weather related data in a learning period, inputs and calculates the demand response implementation related data and the weather related data in the learning period, to a calculation model of each facility, and outputs the calculation result in the learning time. An identification unit 212 receives activation of the demand response, acquires the demand response implementation related data and the weather related data, inputs and calculates the demand response implementation related data and the weather related data in the activation, to the calculation model of each facility, outputs the calculation result in the demand response activation, and calculates the priority of the facility used as the demand response object, on the basis of the calculation result in the learning time and the calculation result in the activation.SELECTED DRAWING: Figure 2

Description

  The present invention relates to facility operation support, and more particularly, to a facility operation support technique for executing demand response.

  With regard to the demand forecasting system, there is a prior art disclosed in Patent Document 1. This prior art uses a collection unit for collecting demand result data for each of a plurality of communities, a database for recording demand result data for each of a plurality of communities, and a demand for a designated forecast date of a designated community. There is disclosed a demand forecasting system comprising: a forecasting unit which forecastes based on demand performance data for each of a plurality of communities to be recorded.

JP, 2014-183656, A

  When demand response is triggered, it is necessary to select a community that can handle the requested amount of demand suppression. The demand prediction technology of Patent Document 1 can be applied to the selection of communities that can handle demand response. However, in the prediction technology in Patent Document 1, since only weather information that is an external factor is used as an input, it may not be possible to suppress the necessary amount of power in the following situation.

  For example, when demand response is activated, circumstances occur that the air conditioning temperature of the air conditioner can not be lowered, and even if the power of the facility can not be suppressed, the air conditioner is selected for the demand response, It may not be possible to achieve the required amount of power reduction.

  An object of the present invention is to provide a facility operation support apparatus, a facility operation support method, and a facility operation support system for supporting selection of a facility having a high possibility of achieving reduction of a necessary amount of power so that demand response can be selected. It is to provide.

  A preferred example of the present invention acquires demand response execution related data and weather related data in a learning period, and inputs the demand response execution related data in the learning period and the weather related data in an operation model of each facility A learning unit that calculates and outputs an operation result at the time of learning, and an activation of a demand response, acquires the demand response execution related data and the weather related data, and calculates an operation model of each facility when the activation is performed. The demand response execution related data and the weather related data are input and calculated, and the calculation result at the time of demand response activation is output, and the demand is calculated based on the calculation result at the time of learning and the calculation result at the time of the activation. An equipment operation support device including an identification unit that calculates priority of equipment to be a response target.

  Further, another preferable example of the present invention receives an instruction of start of learning and a learning period received from the operation screen, acquires demand response execution related data and weather related data in the learning period, and calculates an operation model of each facility To calculate the demand response execution related data in the learning period and the weather related data, output the calculation result at the time of learning, and accept the demand response activation, and the demand response execution related data and the weather related data The data is acquired, the demand response execution related data at the time of the activation and the weather related data are input to the operation model of each facility and the operation is performed, and the operation result at the time of the demand response activation is output. Based on the calculation result at the time of operation and the calculation result at the time of the activation, the priority of the equipment to be the demand response target is calculated Receives an instruction from the operation screen, a facility operation support method for outputting the calculated the priority to the operation screen.

  In addition, another preferable example of the present invention is a power company server that requests activation of demand response, an aggregator server that holds demand response execution related data, a weather server that holds weather related data, and an object of demand response Facility operation support system having a facility operation support server for calculating the priority of the required facilities, wherein the facility operation support server acquires the demand response execution related data from the aggregator server, and the weather from the weather server A learning unit that acquires related data, inputs the demand response execution related data and the weather related data to a calculation model targeted for demand response, calculates the calculation result, and outputs the calculation result at the time of learning; When received, the demand from the aggregator server The sponse execution related data is acquired, the meteorological related data is acquired from the meteorological server, and the demand response execution related data and the meteorological related data at the time of activation are input to the calculation model of the target of the demand response And an identification unit that outputs an operation result at the time of demand response activation, and calculates a priority of equipment to be a demand response target based on the operation result at the time of learning and the operation result at the time of the activation. It is an operation support system.

  According to the present invention, it is possible to obtain a facility operation support apparatus, a facility operation support method, and a facility operation support system for supporting the selection of a facility having a high possibility of achieving suppression of a necessary amount of power as a demand response target. Can.

FIG. 1 is a diagram showing a network configuration of a facility operation support system according to an embodiment. The figure which shows the functional block of a facility operation assistance system. The figure which shows the processing sequence between servers. The figure which shows the operation flow of a facility operation assistance system. The figure which shows a weather information table and a demand response implementation information table. A figure explaining processing of a machine learning operation model. The figure which shows a calculation output result table. The figure which shows the processing flow at the time of learning. The figure which shows the processing flow at the time of demand response implementation. The figure which shows the processing sequence at the time of learning of an installation operation assistance server. The figure which shows the process sequence at the time of the demand response implementation of a plant operation assistance server. A figure showing an operation screen of a facilities operation support system. The other figure which shows the operation screen of an installation operation assistance system.

  Hereinafter, examples will be described in detail using the drawings.

  FIG. 1 shows a network diagram of a facility operation support system in the first embodiment. Reference numeral 101 denotes a facility operation support server. A learning unit that executes machine learning to configure the facility operation support system is built in the facility operation support server 101. Data is extracted from the weather server 102 and the aggregator server 103 via the network 104 as input in the facility operation support system. The weather server 102 acquires information from a server that manages weather information and holds the weather information.

  The aggregator server 103 acquires and holds demand response execution information from a server that receives a demand response trigger request. Reference numeral 105 denotes a power company server. Send a demand response activation request via the network 104. The functional units of the facility operation support server 101, the weather server 102, the aggregator server 103, and the power company server 105 are realized by executing software by a CPU (processor) of a computer that is a hardware resource. The aggregator server 103 is a server that collects suppression power (negawatt), which is a reduction in the power demand from equipment and the like.

  FIG. 2 shows a functional configuration of the facility operation support server in the facility operation support system. First, since the facility operation support system applies machine learning, it is necessary to carry out system learning using a large amount of input data. As input data used here, information at the time of past demand response implementation is used. A transmission / reception unit 205 is a processing unit that acquires data from the weather server 102 and the aggregator server 103 via the network 104. The storage unit 206 holds data acquired from each server as a table. Specifically, the weather information table 207, the demand response implementation information table 208, and the operation output result table 209 are held.

  The weather information table 207 stores weather-related data such as the temperature of the demand response implementation day, the climate, and the like. The demand response implementation information table 208 stores demand response implementation related data such as the demand response start time of the demand response implementation date, the demand response duration time, and the like. In the calculation output result table 209, processing results of variable parameters and calculation formulas in the calculation formula of machine learning are stored. The details of these data tables will be described in FIGS. 5 and 7.

  The arithmetic processing unit 210 includes a learning unit 211 and an identification unit 212. The learning unit 211 causes the arithmetic processing unit 210 to learn data of the data table included in the storage unit 206. By learning, the arithmetic processing unit 210 can find out processing methods for various input data.

  Second, we will explain how to use the learned facility operation support system. First, upon receiving a demand response activation request from the electric power company server 105, the transmission / reception unit 205 transmits data related to the demand response execution of the aggregator server 103 and data related to the weather of the weather server 102 via the network 104. get. Next, the acquired actual data is input to the arithmetic expression obtained by reading the variable parameter of the arithmetic output result table 209 in the storage unit 206, and the processing is performed by the identification unit 212. Then, the equipment for which the demand response is to be implemented is identified from the processing result.

  The identification result can be confirmed by operating the online business terminal 214 with the keyboard 213 or the like. In addition, when it is expressed as "learning" in the present embodiment, "the facility operation support system to which machine learning is applied inputs the various past performance data and repetitively executes the operation, whereby the optimum processing method is performed. To mean "deriving". In addition, when expressed as “identification,” “The facility operation support system that derived the optimal processing method by learning compares the output result derived by learning with the output result derived by inputting the actual data at the time of demand response activation. Shall mean "to do."

  FIG. 3 shows the processing sequence of each processing entity. The main processing units of the facility support system according to the embodiment are the power company server 105, the aggregator server 103, the facility operation support server 101, and the weather server 102. Then, the processing phase is classified into two: learning processing and identification processing. In the learning process, a use case of learning to be performed as preparation for demand response implementation is described. The facility operation support server 101 requests the aggregator server 103 for input data (past result data related to demand response execution), and acquires it (S301). The facility operation support server 101 requests the weather server 102 for input data (weather related historical data), and acquires weather related historical data (S302). The learning unit 211 of the facility operation support server 101 performs learning by using the past performance data acquired in the system as an input (S303). The identification process describes a use case of identification performed by a demand response.

  The power company server 105 sends a demand response activation request to the aggregator server 103 (S304). The aggregator server 103 sends a demand response activation request to the facility operation support server 101 (S305). The facility operation support server 101 requests the aggregator server 103 for input data (data at the time of activation related to demand response execution) and acquires the data (S306). The facility operation support server 101 requests the weather server 102 for input data (data on activation of the weather) and acquires the data (S307).

  The identifying unit 212 of the facility operation support server 101 performs identification by inputting the acquired data at the time of activation to the system (S308). The identification unit 212 displays the identified result (S309). The identification unit 212 determines a demand response implementation target facility (S310). In S310, the person may make a decision by operating the facility operation support server 101 on the operation screen, or the facility operation support server 101 may automatically make a determination using a predetermined standard.

  FIG. 4 shows an operation flow of the facility operation support system, and the operation flow is divided into learning processing and identification processing as in FIG. S401 and S402 correspond to learning processing, and S403 to S408 correspond to identification processing. Each step will be described hereinafter. The learning unit 211 of the facility operation support server 101 acquires input data from the aggregator server 103 and the weather server 102 (S401). The learning unit 211 of the facility operation support server 101 learns the acquired input data (S402).

  The identifying unit 212 of the facility operation support server 101 detects demand response activation (S403). The identification unit 212 of the facility operation support server 101 requests input data from each server (S404). The identifying unit 212 of the facility operation support server 101 acquires input data from each server (S405). The acquired input data is input to the identification unit 212 of the facility operation support server, and the input data is identified (S406).

  The facility operation support server 101 causes the user to confirm the identification result using the identification unit 212 (S407). Specifically, the priority of the implementation target of the demand response is displayed on the display unit. The facility operation support server 101 requests the target device for a demand response (S408). As in S310, in S408, the facility operation support server 101 may be operated on the operation screen to make a human decision, or the facility operation support server 101 may automatically make a decision using a predetermined standard. Although FIG. 4 shows the case where the learning process and the identification process do not overlap, they may be performed in overlapping periods.

  FIG. 5 is a diagram showing the details of the weather information table 207 and the demand response implementation information table 208. As shown in FIG. Each value stored in these tables is updated at the timing of demand response activation. The weather information table 207 is a table for holding weather-related data, and includes a date 501 column, a time 502 column, a weather 503 column, and a temperature 504 column.

  For example, focusing on the line # 1, it means that the weather at 14:00:00 on April 1, 2017 is rain, and the temperature at that time is 10 ° C. The demand response execution information table 208 is a table for holding the contents of the activation request at the time of the demand response execution. It also holds information on equipment that can be targeted for demand response. The demand response execution information table 208 includes the equipment 505 column, the date 506 column, the demand response activation time 507 column, the demand response duration time 508 column, the required suppression power 509 column, and the demand response activation result 510 column Configured

  For example, focusing on the # 1 row, there is a demand response activation request to carry out a 30 kwh demand response continuously from 14:00:00 on April 1, 2012, and there is no problem with A mud water pump It means that it can correspond.

  FIG. 6 is a diagram for explaining the processing of the machine learning calculation model. The mechanism of a neural network is applied to the operation model of the present embodiment. First, a neural network is a model that mimics the mechanism of the human brain's neural network, and by providing a computer with a learning ability, it is an approach for solving various problems. In many cases, this method is mainly applied to multi-dimensional data such as images and statistics, and it is often applied to linear separation difficult problems, and is applied as pattern recognition such as image recognition based on feature quantities.

  In this embodiment, the pattern recognition ability of the neural network is applied. That is, the system is configured to determine which facility's past information is similar to the information at the time of demand response activation. We used a hierarchical network called neural network perceptron. The perceptron is composed of three layers, and is referred to as an input layer which takes in input data, an intermediate layer which performs weight calculation relating data from the input data side by side, and an output layer which outputs output data. The configuration of the learning system 601 will be described.

  In step 601, data is input for each facility, calculation is performed, and a learning output result used for comparison of pattern recognition is derived. The equipment illustrating the internal configuration of the system is A mud pump. There are x1 to x8 as reception channels, and data 602 is input. 602 are eight input data [date, time, weather, air temperature, activation time, duration, suppression power, activation result].

  Explaining based on the table data in FIG. 5, x1 is date 501, x2 is time 502, x3 is weather 503, x4 is air temperature 504, x5 is demand response activation time 507, x6 is demand response continuation time 508, x7 is necessary suppression The powers 509 and x8 are stored in the demand response activation result 510 respectively. The input data is stored in the input layer, and the calculation is performed by the calculation formula of the middle layer. Thereafter, the process moves to the output layer, and o1 to o8 are output as output results. A connection weight (weight) is given to the line connecting the units, and a good output result can be obtained by changing the weight according to the output result. This change in weight is the basis of learning. By increasing the number of times of learning, more optimal weights can be obtained. Although the system of the facility A is at 601, similar learning is performed for each of the facilities 602 and 603.

  FIG. 7 is a diagram showing details of the operation output result table 209. As shown in FIG. The operation output result table 209 is a table in which variable parameters (hereinafter referred to as combined load) in the machine learning equation and processing results of the equation are stored, and the equipment 701 column and the combined load W (12 to 878) 702 It is comprised by the row | line | column 705 row | line and the learning output result O (1-8) 706 row | line | column. For example, focusing on the # 1 job, the combined load and model output results in the operation model where table information of A mud pump was learned are the values of [1.21, 0.82, 0.02 to 0.21] and [1242.32 to 20.22] respectively. It means that there is. This information is updated at the time of learning.

  FIG. 8 is a process flow executed by the learning unit 211 at the time of learning. First, learning of the system is done monthly. The learning unit 211 of the facility operation support system activates the period of the past performance data to be learned and the number of times of learning, which are the start arguments, by inputting the learning unit 211 of the facility operation support system (S801). The past performance data to be learned is the actual value of the demand response implemented in the past (specifically, "implementation date", "weather", "air temperature", "invocation time", "control power", "invocation result" ). The number of times of learning is the number of times of updating the learning argument by machine learning operation using past actual result data as an input. The number of times of learning and the learning period can be designated by the user from the operation screen 1201 of FIG.

  By the process of S801, it becomes possible to start learning of the facility operation support system. The learning unit 211 of the facility operation support system acquires demand response execution related data for the number of days specified in the start argument for each facility from the demand response implementation information table 208 (S802). Then, the facility operation support system acquires weather related data from the weather information table as in S802 (S803). Applying these treatments to the sewerage business, the equipment in S802 is the equipment of the sewerage company, and refers to the owned equipment such as the mud water pump and air conditioner, and the demand response related data is the demand response It becomes the actual value at the point in time (specifically, "Act date", "Activation time", "Control power", "Actual result").

  Then, the weather related data in S803 is the actual value at the time when the demand response is performed (specifically, “execution date and time”, “weather”, “air temperature”). The processing of S802 and S803 makes it possible to acquire input data. Next, the learning unit 211 of the facility operation support system performs calculation by inputting the acquired data as input data to the computing model for each facility (S804). If these treatments are applied to the sewerage business, the input data is the actual value at the time of the demand response implementation (specifically, “implementation date”, “weather”, “air temperature”, “invocation time”, “control power”) The equipment is a sewerage company's equipment, and refers to owned equipment such as a mud pump and an air conditioner. The operation model is an operation unit in the processing unit of the facility operation support system incorporating the machine learning.

  By the process of S804, it is possible to input (input data) to the facility operation support system to perform calculation (learning). The learning unit 211 of the facility operation support system sets the correction value of the coupling load obtained by the calculation result of S804 in the model (S805). If these processes are applied to the sewerage business, the operation result is the output result of the input data processed by the operation unit of the facility operation support system, and the combined load is the degree of association between the input data and the operation model. It is a variable parameter of the arithmetic expression in.

By the process of S 805, it is possible to update the arithmetic expression of the arithmetic model with the set of connection weights.
The learning unit 211 of the facility operation support system repeatedly calculates the flow of S804 and S805 many times until the number of calculations becomes larger than N (S806). This N can be arbitrarily changed by the user. As a result, an operation model for each facility is constructed.

  FIG. 9 illustrates the process flow when the operation model is implemented. First, the processing at the time of execution is performed each time a demand response is triggered. The identification unit 212 of the facility operation support system updates the operation information table when the demand response is received (S901). If this process is applied to the sewerage business, the operation information data obtained when the demand response is received is the actual value (specifically, “execution date and time”, “activation time”, and “suppression power”).

  By the process of S901, it becomes possible to start the priority selection process of the facility operation support system. The identifying unit 212 of the facility operation support system acquires, from the demand response implementation information table 208, demand response implementation related data at the implementation date and time of the demand response (S902).

  The identifying unit 212 of the facility operation support system acquires, from the weather information table, weather-related data at the implementation date and time of the demand response (S903). If these processes are applied to the sewerage project, the implementation information data acquired from the implementation date and time of the demand response will be the actual value (specifically, “demand response activation time”, “weather”, “air temperature”, “demand response continuation” "Time", "necessary power control"). It becomes possible to acquire input data by the processing of S902 and S903. In S902 and S903, the demand response implementation related data and weather related data at the date and time of implementing the demand response are acquired, but may be when the demand response is activated. Furthermore, the learning unit 211 may use data up to the time of activation of the demand response including the demand response execution related data and weather related data close to the activation of the demand response not used for learning. It is desirable for the identification unit to select data to be calculated, so that the identification unit can calculate changes in the state of equipment that can not be grasped in a period in which learning is not possible.

  The identification unit 212 of the facility operation support system inputs the data obtained in S902 and S903 as an input into the calculation model with the combined load set in S805, and performs calculation (S904). If this process is applied to the sewerage business, the operation model of S805 is constructed for each sewerage business facility and is input to each operation model. By the process of S904, it becomes possible to acquire the output result of the operation model in the input data for each facility. The same connection weight as at the time of learning is set, and the operation model at the time of implementation can be an operation model similar to the operation model at the time of learning.

  The identification unit 212 of the facility operation support system calculates the standard deviation (variation degree) between the operation model output result for each facility in the input data at the time of learning and the output result of the operation model for each facility in the input data at the time of triggering the demand response. Do. The calculation result is displayed on the assumption that the operation model that results in the smaller standard deviation is the facility for which the demand response is to be performed, which can be the target for the higher priority demand response (S905). According to the calculation result displayed on this screen, the user can stop the equipment and can cope with the demand response.

  For example, when a situation where the air conditioning temperature of the air conditioner can not be lowered suddenly after the demand response is triggered, according to the present embodiment, the past computation model output result and the computation model output result after the demand response is triggered And the degree of variation between According to the present embodiment, it is possible to avoid stopping by selecting such an air conditioning facility as a facility for which the demand response is to be implemented. In addition, since the degree of variation is high for equipment that has been temporarily stopped at the time of demand response activation, equipment that has been stopped is selected as the facility for which demand response is to be implemented, and required suppression power is required after activation of demand response. The problem of not being able to be obtained can be avoided.

  Instead of taking the standard deviation of the output result of the computation model, the output result of the computation model for each facility in the input data at the time of demand response activation (for example, O1 etc.) in the input data at learning The absolute value of the difference between etc. is calculated for each of the outputs O1 to O8. Then, it may be assumed that the equipment for which the demand response is to be implemented has a high priority in the order of the equipment for which the sum of the differences is small.

  As shown in FIG. 8 and FIG. 9, the introduction of machine learning makes it possible to repeatedly learn a wide variety of data and to explicitly derive insights from the system, thus requiring human judgment. The location is unnecessary. Therefore, by calculating and accumulating the past results (feature data) of facilities for the sewerage enterprise and machine learning the accumulated data as input data, the most suitable implementation target for the demand response to be triggered in the future It is possible to predict the equipment.

  FIG. 10 illustrates a processing sequence at the time of learning of the facility operation support server. The main processing units of the facility operation support server according to the embodiment are the online business terminal 214, the learning unit 211, and the storage unit 206. First, the online business terminal 214 inputs an input of a learning target period (period of past data for machine learning) and an input number of times of learning (number of updates of variable parameter in machine learning) (S1003). By pressing the learning start button, the online business terminal 214 delivers the learning target period and the number of times of learning to the learning unit 211, and starts the learning process (S1004).

  For the table information 1001 in which the learning unit 211 in the storage unit 206 has "A mud water pump" indicating one of the facilities as a key, the date 506 in FIG. Information is called (S1005). The table information 1001 having "A mud water pump" as a key is composed of a weather information table 207 and a demand response execution information table 208. The date, time, weather, air temperature, demand response activation time, demand response duration, required suppression power, and demand response activation result regarding the mud water pump of the equipment row 505 of FIG. 5 are extracted from the demand response execution related data and the weather related data It is the table information 1001 which made "A mud water pump" the key. In S1005, the learning unit 211 inputs each item of the table information 1001 called from the storage unit 206 to the operation model of machine learning (S1006).

  The learning unit 211 carries out the calculation processing of the calculation model, and updates the calculation output result and the combined load to the table information 1001 having “A muddy water pump” as a key (S1007). As for the calculation output result, the learning output results O (1 to 8) 706 to 707 of the calculation output result table 209 shown in FIG. 7 are updated. As for the connection load, the connection loads W (12 to 878) 702 to 705 of the calculation output result table 209 shown in FIG. 7 are updated. Then, the learning unit 211 calls information in the learning target period specified by the user or the like on the table information 1002 whose key is the equipment “B air conditioner” in the storage unit 206, and the date 506 in FIG. S1008).

  In S1008, the learning unit 211 inputs the table information 1002 called from the storage unit 206 to the operation model of machine learning (S1009). The learning unit 211 carries out the calculation processing of the calculation model, and updates the calculation output result and the connection weight to the table information 1002 in the storage unit 206 with “B air conditioner” indicating one of the facilities as a key S1010). As with “A mud pump” and “B air conditioner”, calculation processing is performed on “C substation machine”, “D control pump” and all facilities that are subject to demand response, and the value of the calculation output result table 209 Update Then, the calculation is repeated for the designated number of times of learning (S1011).

  FIG. 11 illustrates a processing sequence when the demand response of the facility operation support server is performed. The processing entities of the facility operation support server according to the embodiment are the aggregator server 103, the identification unit 212, and the storage unit 206. The aggregator server 103 detects a demand response activation request (S1103). When the aggregator server 103 detects the activation request, the aggregator server 103 requests the identification unit 212 to start the identification process (S1104). In the identification process, each process is performed on an operation model in which the parameter of each facility is set.

  For the storage unit 206, the identification unit 212 sets the combined load W (12 to 878) 702 to 705 and the learning output result O (1 to 8) 706 to the equipment in the operation output result table 209 as "A muddy water pump". 707 is called (S1105). The identification unit 212 sets the combined load called from the operation output result table 209 to the operation model of the A mud water pump of the identification unit 212 (S1106). The identification unit 212 calls information from the weather information table 207 to the storage unit 206 in order to obtain weather information at the time of demand response activation (S1107). The identification unit 212 inputs the value of each item related to the weather, which is called from the information from the weather information table 207 of the storage unit 206, to the operation model of “A mud water pump” (S1108).

  The identification unit 212 calls the performance information of the demand response implementation information table 208 to the storage unit 206 in order to acquire the performance information of the demand response when the demand response is activated (S1109). The identification unit 212 inputs the information of each item of the demand response result which is called from the demand response execution information table 208 of the storage unit 206 into the operation model of “A mud water pump” (S1110). The identification unit 212 inputs the information from the weather information table 207 and the data of the information from the demand response execution information table 208 to the operation model in which the variable parameter such as the combined load of the A mud pump is set, and executes the operation ( S1111).

  The identifying unit 212 outputs the learning output result O (1 to 8) of S1111 at the time of learning in the “muddy water pump” model and the learning output result O (1 to 8) of the operation output result table 209 at the time of execution of the demand response. The standard deviation value (variation degree) with 706 to 707 is calculated (S1112). The above is calculation of the performance value in the calculation model whose installation is A mud pump. Similarly, for the operation model of all the facilities, the standard deviation value is calculated by using the demand response performance information as an input. Then, after calculating the standard deviation values of all the facilities, the identification unit 212 arranges the standard deviation values of the respective facilities in ascending order, and finds the minimum value (S1113). The fact that the standard deviation value is the minimum value means that the standard deviation value is similar to the operation model with less variation, which is the minimum value. The second smallest value means that it is similar to the next. That is, the equipment with the highest priority for implementing demand response is the equipment with the second highest. The priority is displayed on the screen for visualization.

  12 and 13 show operation screens in the facility operation support system. First, reference numeral 1201 denotes a Home screen for selecting whether the screen is a learning screen or a priority screen. The system learning situation has not been completed, the number of times of learning, and the learning period are also "-". Therefore, it is necessary to change the screen to the next screen by pressing the button "To learning screen". A screen 1202 is a screen transitioned from the Home screen 1201 and is a learning screen. By defining the learning period and the number of times of learning on this screen, it is possible to specify the learning period when learning and the number of times of learning. After designation, it is necessary to press the learning start button to carry out learning.

  A screen 1203 illustrated in FIG. 13 is a screen transitioned by pressing a learning button on the learning screen 1202. It is possible to confirm that the system learning situation is completed, the number of times of learning, and the learning period are also specified. In addition, since the message "the demand response has been received" can be confirmed, it is necessary to confirm the priority of the equipment for the demand response on the priority screen. Reference numeral 1204 denotes a priority screen on which facilities capable of demand response at the time of demand response activation are arranged in order of priority from the top.

  According to this embodiment, since the relevancy between input data can be derived automatically, it is possible to obtain the necessary result in the system without requiring human judgment.

101: Equipment operation support server 102: Weather server 103: Aggregator server 104: Network 105: Power operator server 206: Storage unit 207: Weather information table 208: Demand response execution information table 209: Calculation output result table 210: Calculation processing unit 211: Learning unit 212: Identification unit

Claims (9)

Demand response execution related data and weather related data in the learning period are acquired, and the demand response execution related data in the learning period and the weather related data are input to the operation model of each facility to calculate the operation at the time of learning A learning unit that outputs the result;
In response to the activation of demand response, the demand response execution related data and the weather related data are acquired, and the demand response execution related data and the weather related data at the time of the activation are input to the operation model of each facility It has an identification unit which calculates and outputs the calculation result at the time of the activation, and calculates the priority of the equipment to be the demand response target based on the calculation result at the time of the learning and the calculation result at the time of the activation. An equipment operation support apparatus characterized by In the facility operation support device according to claim 1,
The identification unit
Facility operation support characterized by calculating the degree of variation between the calculation result at the time of the learning and the calculation result at the time of the activation, and setting the equipment having a small degree of variation as the target of the demand response having a high priority. apparatus. In the facility operation support device according to claim 2,
The connection weight is the degree of relevance of input data to the calculation model,
The learning unit is
The connection load obtained by calculating the demand response execution related data of the designated period and the weather related data by inputting into the calculation model of the respective facilities and calculating them is set in the calculation model;
The identification unit
A facility operation support apparatus, which executes an operation at the time of activation based on the operation model in which the connection load is set. In the facility operation support device according to claim 3,
The demand response implementation related data includes a demand response activation time, a required suppression power, and a demand response duration time, the weather related data includes the weather and the air temperature, and the identification unit is the calculation model of each facility. An equipment operation support apparatus, which calculates a demand response execution related data and the weather related data by inputting them and outputs a calculation result at the time of the activation. In the facility operation support device according to claim 2,
The calculation result at the time of learning and the calculation result at the time of the activation each have a plurality of calculation results,
The identification unit
Equipment operation support characterized by calculating a standard deviation between the calculation result at the time of learning and the calculation result at the time of the activation, and calculating the equipment with the lower calculated standard deviation as the equipment with higher priority. apparatus. In the facility operation support device according to claim 3,
The learning unit is
Calculating and calculating the demand response execution related data and the weather related data in the learning period received from the operation screen;
The identification unit
An equipment operation support apparatus, which acquires and calculates the demand response execution related data and the weather related data at the time of activation of the demand response. In the facility operation support device according to claim 1,
Storing the demand response execution related data and the weather related data in a storage unit;
The demand response implementation related data includes a demand response activation time and suppression power for each facility of the sewerage business,
The calculation result at the time of the learning and the calculation result at the time of the activation each have the calculation result for each of the facilities,
The equipment operation support apparatus, wherein the identification unit outputs the equipment for which the demand response is targeted for high priority. Receive instructions from the operation screen to start learning and learning period,
Acquiring demand response execution related data and weather related data in the learning period,
The demand response execution related data of the learning period and the weather related data are input to a calculation model of each facility and calculated;
Output the calculation result at the time of learning,
Accept trigger of demand response,
Acquiring the demand response execution related data and the weather related data;
The demand response execution related data at the time of the activation and the weather related data are input to a calculation model of each facility and calculated.
Output the operation result when the demand response is triggered,
Based on the calculation result at the time of the learning and the calculation result at the time of the activation, the priority of the equipment to be the demand response target is calculated;
A facility operation support method comprising: receiving an instruction from the operation screen; and outputting the calculated priority to the operation screen. An electric utility server that requests demand response activation, an aggregator server that holds demand response execution related data, a weather server that holds weather related data, and facility operation that calculates the priority of the equipment that is the target of the demand response A facility operation support system having a support server,
The facility operation support server is
Acquiring the demand response implementation related data from the aggregator server and acquiring the weather related data from the weather server;
The demand response execution related data and the weather related data are input to a demand response target calculation model, and calculation is performed,
A learning unit that outputs an operation result during learning;
When the demand response is triggered, the demand response execution related data is acquired from the aggregator server, the weather related data is acquired from the weather server, and the calculation model of the target of the demand response is the above-mentioned in the case of the invocation. Demand response execution related data and the weather related data are input and calculated, and the calculation result at the time of demand response activation is output, and the demand response is calculated based on the calculation result at the time of learning and the calculation result at the time of the activation. An equipment operation support system comprising: an identification unit that calculates a priority of a target installation.

Images