WO2016185919A1 - Système de prédiction de besoin en énergie et procédé de prédiction de besoin en énergie - Google Patents

Système de prédiction de besoin en énergie et procédé de prédiction de besoin en énergie Download PDF

Info

Publication number
WO2016185919A1
WO2016185919A1 PCT/JP2016/063696 JP2016063696W WO2016185919A1 WO 2016185919 A1 WO2016185919 A1 WO 2016185919A1 JP 2016063696 W JP2016063696 W JP 2016063696W WO 2016185919 A1 WO2016185919 A1 WO 2016185919A1
Authority
WO
WIPO (PCT)
Prior art keywords
demand
information
time
data
granularity
Prior art date
Application number
PCT/JP2016/063696
Other languages
English (en)
Japanese (ja)
Inventor
将人 内海
渡辺　徹
悠 池本
羊子 崎久保
郁雄 茂森
広晃 小川
Original Assignee
株式会社日立製作所
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from JP2015211260A external-priority patent/JP6408451B2/ja
Application filed by 株式会社日立製作所 filed Critical 株式会社日立製作所
Priority to EP16796313.1A priority Critical patent/EP3300205A4/fr
Priority to US15/574,187 priority patent/US20180128863A1/en
Publication of WO2016185919A1 publication Critical patent/WO2016185919A1/fr

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks

Definitions

  • the present invention relates to an energy demand prediction system and an energy demand prediction method.
  • Electricity companies that sell electricity to customers based on the concluded electricity supply contract, and that generate electricity with their own generators, or directly procure from other electricity companies or through exchanges In general, it is required that the amount of electricity procured and the amount of sales be matched at every settlement time. Therefore, it is important for the electric power company to accurately predict the future value of the total demand (contract demand) of the customers with whom the company has contracted (contract demand).
  • Patent Document 1 discloses that past time-series power demand data is separated into long-period fluctuation demand data and short-period fluctuation demand data, and is converted into long-period fluctuation demand data. Based on the short-term fluctuation demand data, the short-term demand forecast value is calculated for the short-period fluctuation demand based on the short-period fluctuation demand data. It is disclosed. According to such a method disclosed in Patent Document 1, a demand prediction value of total power can be calculated by adding both.
  • loads are divided into commercial existing power, commercial existing heat, commercial new power, commercial new heat, industrial existing power, industrial existing heat, industrial new power, and industrial new heat.
  • Generating a curve is disclosed. According to the method disclosed in Patent Document 2, a load curve can be generated from a small amount of basic data.
  • customer information data is classified based on attributes such as a house type and a possessed appliance type, usage information is randomly extracted from a database based on the classification result, and each piece of information is extracted from the extracted usage information.
  • a method for creating predicted data by obtaining the maximum usage amount of an attribute and obtaining a frequency distribution in which the maximum usage amount appears is disclosed.
  • Patent Documents 2 and 3 more accurate demand prediction is realized by reflecting different demand characteristics for each type of demand.
  • the characteristics of demand types such as existing and newly installed for business use and industrial use are time-invariant.
  • Patent Document 3 there is no mechanism for determining the number of demand types most suitable for prediction and the granularity (daily, weekly, etc.). In other words, these technologies have a problem that the number of demand types accompanying change and the granularity thereof cannot be accurately grasped.
  • An object of the present invention is to provide an energy demand forecasting system and an energy demand forecasting capable of forecasting demand while changing.
  • a demand pattern generation basis for extracting energy demand performance information of a preset past period as basic data for demand pattern generation.
  • the energy for each granularity of the time in the combination A classification granularity adjustment processing unit for generating demand pattern generation data composed of one or a plurality of unit data each representing demand, and each unit data constituting the demand pattern generation data for each of the demand pattern generation data
  • a segmentation processing unit that classifies a subset into a small number of subsets and extracts a demand pattern that represents a representative energy consumption trend for each subset, and for each of the demand pattern generation data
  • each of the demand pattern generation data Profiling processing unit that generates a combination of the demand pattern and the attribute information of the customer common to the demand
  • the energy demand prediction method for predicting energy demand in order to solve this problem, in the present invention, a first step of extracting energy demand performance information for a preset past period as basic data for demand pattern generation. And for each combination of contract demand granularity and time granularity set in advance one or more sets based on the extracted basic data for demand pattern generation, the energy demand for each time granularity in the combination is represented, respectively.
  • a third step of calculating an energy demand at a preset past date and time as a predicted value for evaluation, and the demand type for each of the demand types Based on the evaluation predicted value and the energy demand record information, the contract is made so that the error between the estimated value or predicted value of the energy demand at the past date and time and the actual observed value at the date and time is minimized.
  • 1 is an overall configuration diagram of an energy demand management system.
  • 1 is an overall configuration diagram of an energy demand prediction system. It is a functional block diagram of each apparatus which comprises an energy demand prediction system.
  • It is a conceptual diagram of the customer information to display.
  • It is a conceptual diagram of the sales contract forecast actual information to display.
  • It is a conceptual diagram of the demand type information to display.
  • It is a conceptual diagram of the demand type information to display.
  • It is a conceptual diagram of the demand type information to display.
  • It is a flowchart which shows the procedure of an energy demand prediction process.
  • It is a flowchart which shows the procedure of a segmentation process.
  • It is a flowchart which shows the procedure of the estimated value calculation process for evaluation.
  • It is a conceptual diagram which shows the effect by the technique to provide.
  • FIG. 1 shows the overall configuration of an energy demand management system 1 according to this embodiment.
  • the energy demand management system 1 calculates a predicted value for demand on an arbitrary date and time based on the actual demand information, and generates and controls an operation plan for an operable generator based on the calculated demand predicted value.
  • the communication paths 111 and 112 are communication paths that connect various devices and terminals constituting the energy demand management system 1 so that they can communicate with each other.
  • the communication paths 111 and 112 include a LAN (Local Area Network).
  • the electric power company 2 is a business company composed of a supply and demand manager 3, a sales manager 4, a transaction manager 5, and a facility manager 6.
  • the supply and demand manager 3 predicts the future demand for each 30-minute clearing time unit based on the sales plan owned by the company and the future sales plan, and procure power so that the predicted demand can be satisfied.
  • a department or person in charge for managing the quantity a forecast calculation device 31 for calculating a forecast value of demand, a demand type management device 30 for generating and managing a type of demand used to calculate the demand forecast value, An information input / output terminal 32 for exchanging data with these devices is provided.
  • the sales manager 4 is a department or person responsible for planning a long-term or short-term electricity sales plan, signing a new contract with a customer, and managing an existing power supply contract.
  • a sales management device 40 is provided for managing information on consumers who have signed sales plans and electricity supply contracts.
  • the transaction manager 5 is a department or a person in charge who plans and executes transactions for procuring electricity through direct contracts with other electric utilities or through exchanges.
  • a transaction management device 50 is provided for managing information on a contract for purchasing electricity that has already been contracted, and exchanging telegrams relating to transactions with other electric utilities and exchanges.
  • the facility manager 6 is a department or person in charge of planning and executing an operation plan for power generation facilities owned by the company or power generation facilities outside the company that can be incorporated into the company's own electricity procurement plan. Management of information, planning of operation plan of power generation equipment, equipment management device 61 for transmitting a control signal for execution, and receiving control signals from the equipment management device 61 and actually executing control of power generation equipment A control device 62 is provided.
  • the grid operator 7 is a business operator who manages power transmission / distribution system facilities over a wide area, measures the actual demand of each local customer, and stores the measured values. Is provided with a system information management device 70.
  • Trade market operator 8 is a business operator that comprehensively manages information and procedures necessary for conducting power transactions with a plurality of electric power companies.
  • a market operation management device 80 is provided for performing a process of adding orders received from a business operator.
  • the public information provider 9 is a company that provides past history information and future forecast information on weather such as temperature, humidity, and solar radiation, and distributes public information for delivering past history information and forecast information on weather.
  • a device 90 is provided.
  • the customer 10 is an individual or a corporation having a load facility and a power generation facility, and the electric power company 2 or the system operator 7 has a tendency of demand and power generation such as owned facilities and facilities, type of business, number of people in the room, location, etc.
  • An information input / output terminal 100 for transmitting information that can be affected, and a measuring device 101 for measuring the actual amount of demand and power generation are provided.
  • FIG. 2 shows an energy demand prediction system 12 according to the present embodiment that constitutes a part of the energy demand management system 1.
  • the energy demand prediction system 12 according to the present embodiment includes a demand type management device 30, a prediction calculation device 31, and a sales management device 40.
  • the demand type management device 30 uses the demand type information based on the demand record information 4006A and the customer information 4007A held by the sales management device 40, the weather past information 9001A and the industrial dynamic past information 9003A received from the public information distribution device 90. 3011 is generated, and the generated demand type information 3011A is transmitted to the prediction calculation device 31.
  • the demand type information 3011 held by the demand type management device 30 includes, for example, an identifier for identifying a demand type for each customer (hereinafter referred to as a demand type ID), attribute information for explaining each demand type, and each demand type. Contains information on demand patterns.
  • the prediction calculation device 31 includes demand type information 3011 generated by the demand type management device 30, sales contract forecast information 4008 held by the sales management device 40, weather past information 9001 ⁇ / b> A and industrial dynamic past held by the public information distribution device 90. Based on the information 9003A, a demand at an arbitrary past date and time set in advance is calculated as a predicted value for evaluation.
  • the prediction calculation device 31 extracts a set of demand type information 3011 that minimizes the difference between the evaluation predicted value and the past observed value, and extracts the demand type information, the weather forecast information 9002A, and the industrial movement information 9004A. Based on the above, the demand forecast value of the designated future date and time is calculated, and the demand forecast information 4009A held by the sales management device 40 is generated. In addition, the prediction calculation device 31 transmits the demand prediction value calculated in this way to the facility management device 61 and the transaction management device 50.
  • the sales management device 40 holds demand record information 4006A, customer information 4007A, sales contract forecast information 4008A, and demand forecast information 4009A.
  • the demand record information 4006A is information generated based on the past demand record information 4006A of a contracted customer or a contract target customer acquired from the measuring device 101 or the grid information management device 70.
  • the actual value of electricity demand (energy demand) every 30 minutes in the past several years for each consumer 10 is included.
  • the customer information 4007A is information provided from the customer 10 when applying for an electric power supply contract from the customer 10, and includes, for example, a location, a business type, a building type, a floor area, a management company, the number of people in the room, It includes information indicating consumer attributes such as contract power capacity, start date / time and end date / time of electricity supply.
  • the sales contract forecast information 4008 is information created by the sales management device 40 or the sales manager 4, for example, the power of the acquired contract for each demand type in weekly or monthly units over an arbitrary period in the past and future. Includes information such as planned and actual capacity and number of houses.
  • the demand prediction information 4009A is information generated by the prediction calculation device 31, for example, the time unit granularity is from 30 minutes to the year, the time range is from the next several hours to the next several years, and the target granularity is one contract. Contains information on demand forecasts from demand units to total contract demand units.
  • FIG. 3 shows a specific configuration of each device constituting the energy demand prediction system 12.
  • the demand type management device 30 is constituted by an information processing device such as a personal computer, a server computer, or a handheld computer, for example, and a CPU (Central Processing Unit) 3001 that controls the operation of the demand type management device 30 in an integrated manner, an input device 3002, An output device 3003, a communication device 3004, and a storage device 3005 are provided.
  • a CPU Central Processing Unit
  • the input device 3002 is composed of a keyboard or a mouse
  • the output device 3003 is composed of a display or a printer.
  • the communication device 3004 includes a NIC (Network Interface Card) for connecting to a wireless LAN or a wired LAN.
  • the storage device 3005 includes a storage medium such as a RAM (Random Access Memory) and a ROM (Read Only Memory).
  • the storage device 3005 stores various computer programs such as a demand pattern generation basic data extraction unit 3006, a predicted value evaluation data generation unit 3007, a classification granularity adjustment processing unit 3008, a segmentation processing unit 3009, and a profiling processing unit 3010. .
  • the basic data extraction unit 3006 for demand pattern generation obtains actual information (hereinafter referred to as demand pattern generation basis) for a preset past period of one or more contract demands set in advance from the demand actual information 4006A (FIG. 2).
  • This program has a function of extracting data).
  • the predicted value evaluation data generation unit 3007 is a program having a function of generating evaluation data for calculating a difference from the predicted value for evaluation generated by the prediction calculation device 31 from the demand record information 4006A.
  • the classification granularity adjustment processing unit 3008 is configured to set the granularity and time of contract demand (consumer unit) set in advance one or more sets based on the basic data for demand pattern generation extracted by the basic data extraction unit 3006 for demand pattern generation. It is a program having a function of generating one or more data sets for generating a demand pattern according to combination information with the granularity.
  • contract demand granularity refers to a specific contract demand unit such as one contract demand unit, total contract demand unit, or an arbitrary contract demand unit between the two.
  • time granularity refers to a specific unit of time such as several hours, one day, several days, one week, one month, or one year. In the following, each of the granularity of contract demand and the granularity of time is set one by one, and the generated data for generating a demand pattern is called one data set or demand pattern generating data.
  • the segmentation processing unit 3009 classifies contract demands having similar trends in periodic fluctuations in power consumption in each data set based on each of the data sets generated by the classification granularity adjustment processing unit 3008, and represents a representative from each.
  • This program has a function of generating a typical demand pattern.
  • the profiling processing unit 3010 is based on the demand pattern generated by the segmentation processing unit 3009 and attribute information that can cause demand patterns such as customer information 4007A, past weather information 9001A, and industrial dynamic past information 9003A.
  • a program having a function of extracting attribute information common to each demand pattern and generating a demand type.
  • the storage device 3005 stores a database such as a demand type information storage unit 3011.
  • the demand type information storage unit 3011 holds demand type information 3011A.
  • the demand type information 3011A is information on the demand type generated by the profiling processing unit 3010.
  • the predictive calculation device 31 is configured by an information processing device such as a personal computer, a server computer, or a handheld computer, and controls the operation of the predictive calculation device 31 in an integrated manner, an input device 3102, an output device 3103, and a communication device. 3104 and a storage device 3105.
  • an information processing device such as a personal computer, a server computer, or a handheld computer
  • the storage device 3105 includes a storage medium such as a RAM and a ROM, and includes various computer programs such as an evaluation predicted value calculation processing unit 3106, an evaluation calculation unit 3107, and a final predicted value calculation processing unit 3108, and preset weighting factors. 3109 is stored.
  • the estimated value calculation processing unit 3106 for evaluation includes information on one or more sets of demand types generated by the demand type management apparatus 30, and each demand type such as sales contract forecast information 4008A, weather past information 9001A, and industrial dynamic past information 9003A. Based on the information of the common attribute information, the demand record information 4006A managed by the sales management device 40, and the preset weight coefficient information 3109A, the demand in the past evaluation target date and time is set as the evaluation predicted value. It is a program having a function to calculate.
  • the evaluation calculation unit 3107 compares each of the evaluation predicted values calculated by the evaluation predicted value calculation processing unit 3106 with the evaluation data generated by the demand type management device 30, and the demand type set having the smallest difference. Is a program having a function of extracting. Details of the demand type set will be described later.
  • the final predicted value calculation processing unit 3108 includes the demand type set extracted by the evaluation calculation unit 3107 and the future of attribute information common to each demand type such as sales contract prediction actual information 4008A, weather forecast information 9002A, and industrial movement information 9004A.
  • This is a program having a function of calculating a demand forecast value at an arbitrary future date and time set in advance based on information including a forecast value, a forecast value, and an assumed value.
  • weighting factor information 3109A is information including various weighting factors used when the demand for the past evaluation target date and time set in advance by the evaluation predicted value calculation processing unit 3106 is calculated as the evaluation predicted value as described above. is there. Details of the weight coefficient information 3109A will be described later.
  • the sales management device 40 is composed of an information processing device such as a personal computer, a server computer, or a handheld computer, and controls the operation of the sales management device 40 in an integrated manner, an input device 4002, an output device 4003, and a communication device. 4004 and a storage device 4005.
  • the input device 4002 is composed of a keyboard or a mouse
  • the output device 4003 is composed of a display or a printer.
  • the communication device 4004 includes a NIC for connecting to a wireless LAN or a wired LAN.
  • the storage device 4005 is composed of a storage medium such as a RAM and a ROM, and stores various computer programs such as a data input / output unit 4010 and a sales contract forecast information generation unit 4011.
  • the data input / output unit 4010 registers, searches, updates, and deletes databases such as the demand record information storage unit 4006, the customer information storage unit 4007, the sales contract forecast information storage unit 4008, and the demand forecast information storage unit 4009. It is a program having a function of performing an operation.
  • the sales contract forecast information generation unit 4011 has a function of creating sales contract forecast information 4008A based on the demand type information 3011A generated by the demand type management apparatus 30 and the customer information 4007A held by the sales management apparatus 40. It is a program that has.
  • the storage device 4005 also stores databases such as a demand record information storage unit 4006, a customer information storage unit 4007, a sales contract forecast information storage unit 4008, and a demand forecast information storage unit 4009.
  • the demand result information storage unit 4006 is a database in which information received and acquired from the measuring device 101 and the system information management device 70 is stored, and includes various kinds of record information on demand for past years of contract demand and contract target demand. Information is stored.
  • the granularity of the performance information is, for example, 30 minutes.
  • the customer information storage unit 4007 is a database in which information acquired based on an application from the customer 10 at the time of application or conclusion of an electricity supply contract is stored, and the location, type of business, building type, floor of the customer 10 Various types of information including attributes of the customer 10 such as area, management company, and number of people in the room are stored.
  • the sales contract forecast information storage unit 4008 is a database that stores information on schedules and results of electricity sales plans generated by the sales contract forecast information generation unit 4011 and the sales manager 4, and stores past and future information for each demand type. Various information including the number of contracts over the period and the planned and actual measured values of contract power capacity are stored.
  • the demand prediction information storage unit 4009 is a database that stores information on predicted values of future demand generated by the prediction calculation device 31, and stores predicted values of demand every 30 minutes from a preset start date to end date. Various information including it is stored.
  • FIG. 4 shows a conceptual diagram of customer information 4007A.
  • the customer information 4007A is information created by the sales manager 4 based on the application information from the customer 10 at the time of applying for or concluding the electricity supply contract from the customer 10 (FIG. 1).
  • the customer information 4007A is used when the profiling processing unit 3010 generates demand type information.
  • the customer information 4007A includes a customer ID column 4007A1, a location column 4007A2, an industry column 4007A3, a building type column 4007A4, a total floor area column 4007A5, a management company column 4007A6, a number of people column 4007A7, a contract capacity column 4007A8, and a supply.
  • the table configuration includes a start date column 4007A9 and a supply end date column 4007A10.
  • the customer ID column 4007A1 stores an identifier (customer ID) unique to each customer 10 assigned to each customer 10, and the location column 4007A2 and the industry column 4007A3 respectively correspond to the corresponding customer. Ten locations and business types are stored.
  • the building type column 4007A4, the floor area column 4007A5 column, the management company column 4007A6, and the number of people column 4007A7 store the building type, floor area, facility management company, and normal occupancy number of the corresponding customer 10, respectively.
  • the contract capacity column 4007A8, the supply start date column 4007A9, and the supply end date column 4007A10 store the contract power capacity, the start date of the electric supply, and the end date of the electric supply, respectively.
  • the customer with the customer ID “C001” has the location “Ota-ku, Tokyo”, the type of business “retail”, the type of building “office”, and the floor area “100m2”, facility management company “Company A”, normal occupancy number “12”, contract power capacity “30kW”, supply start date “March 25, 2015”, supply end date Is shown to be “undecided”.
  • FIG. 5 shows a conceptual diagram of sales contract forecast information 4008A.
  • the sales contract forecast information 4008A is information created by the sales management device 40 or the sales manager 4 based on the demand type information generated by the demand type management device 30 and the customer information 4007A. Includes planned and actual values of power capacity and number of houses for each type of demand.
  • the sales contract forecast information 4008A includes, as shown in FIG. 5, a demand type ID column 4008A1 and a period set in advance by the sales manager 4 provided corresponding to each demand type ID column 4008A1.
  • the table configuration includes a plan value column 4008A2 and an actual value column 4008A3 of contracts for each demand type.
  • the granularity and range of the period indicated by 4008A2 are preset by the sales manager 4.
  • the demand type ID column 4008A1 stores an identifier specific to each demand type described later, and the plan value column 4008A2 of the contract for each demand type for each period has a corresponding period. Stores the planned value of the contracted power capacity planned for the corresponding demand type. Also, the actual value of contract power capacity for the demand type corresponding to the corresponding period is stored in the actual value column 4008A3 of the contract for each demand type for each period.
  • the actual value of the contract type power capacity of the demand type “S1-DT001” is “2000kW” for the first week of FY2003, “3000kW” for the second week, and thereafter For this reason, it has been shown that the results have not yet been aggregated ("null").
  • the planned value of the contract power capacity for this demand type is shown by the sales manager 4 that the contract power capacity is “4000KW” for the third week of FY2015 and “5000kW” for the fourth week of FY2015. Yes.
  • FIGS. 6 to 9 show conceptual diagrams of the demand type information 3011A.
  • the demand type information 3011A is information generated by the supply and demand manager 3 and the profiling processing unit 3010, and includes information described below.
  • FIG. 6 is a conceptual diagram showing one piece of demand type information 3011A preset by the supply and demand manager 3.
  • This demand type information 3011A is information including the granularity of contract demand and the granularity of time of the demand type to be generated.
  • the demand type information 3011A has a table configuration including a demand type set ID column 3011A1, a contract demand granularity column 3011A2, and a time granularity column 3011A3.
  • the contract demand granularity column 3011A2 stores the contract demand granularity
  • the time granularity column 3011A3 stores the time granularity.
  • a combination of the contract demand granularity stored in the corresponding contract demand granularity column 3011A2 and the time granularity stored in the corresponding time granularity column 3011A3 (hereinafter referred to as a demand type set).
  • the identifier (demand type set ID) unique to the demand type set given to the call is stored.
  • the demand type set “S1” is a demand type representing a combination of the granularity of contract demand “total contract demand unit” and the granularity of time “8760 hours” equivalent to one year. Shown to be a set.
  • FIG. 7 is a conceptual diagram showing one of the demand type information 3011A generated by the profiling processing unit 3010.
  • This demand type information 3011A is information including the relationship between the customer 10 and the demand type.
  • the demand type information 3011A has a table configuration including a customer ID column 3011A4 and a plurality of demand type ID columns 3011A5 provided for each demand type set.
  • the customer ID column 3011A4 stores a customer ID assigned to each customer 10, and each demand type ID column 3011A5 corresponds to a corresponding period in the corresponding demand type set of the corresponding customer 10.
  • the identifier (demand type ID) representing the demand type at is stored.
  • the customer 10 named “C001” has a demand type set “S1”, the demand type for the first period is “S1-DT014”, and the demand type for the second period. Is “S1-DT043”, and the demand type of the last period “T1” is “S1-DT022”.
  • the demand type set is “SS”
  • the demand types up to the period 1, the period 2, and the last period “TS” are different from those in the case where the demand type set is “S1”. It has been shown.
  • the number of periods until the last period “T1” when the demand type set is “S1” and the number of institutions until the last period “TS” when the demand type set is “S2” are not necessarily the same.
  • the number of each period is determined based on the time granularity column 3011A3 of the demand type information 3011A with respect to the period of the demand pattern generation basic data extracted by the demand pattern generation basic data extraction unit 3006.
  • FIG. 8 is a conceptual diagram showing one of the demand type information 3011A generated by the profiling processing unit 3010.
  • This demand type information 3011A is information including common attributes of the demand types.
  • the demand type information 3011A has a table configuration including a demand type set ID column 3011A6, a demand type ID column 3011A7, and a plurality of attribute information columns 3011A8.
  • the demand type set ID column 3011A6 stores the demand type set ID assigned to each demand type set.
  • the demand type ID column 3011A7 stores the demand of one demand type generated from the corresponding demand type set.
  • the type ID is stored. For this reason, in this demand type information 3011A, the same number of rows as the demand types that can be generated from the demand type set are provided for one demand type set ID, and different demands are stored in the demand type ID column 3011A7 of each row.
  • the type demand type ID is stored.
  • each attribute information column 3011A8 stores the contents of attribute information common to contract demands belonging to the corresponding demand type generated from the corresponding demand type set.
  • the demand type “S1-DT001” generated from the demand type set “S1” has the operating company “Company A” as attribute information common to contract demands belonging to this demand type. It is shown that N pieces of attribute information are extracted, such as “floor area is“ 90 m 2 or more ”, industry is“ manufacturing ”, and temperature is“ 15 ° C. or less ”.
  • N pieces of attribute information are extracted, such as “floor area is“ 90 m 2 or more ”, industry is“ manufacturing ”, and temperature is“ 15 ° C. or less ”.
  • the demand type “S1-DT002” generated from the demand type set “S1” only two attribute information of “Operating company B” and “50 people or more” was extracted. It is shown. In this case, since the difference from other demand types can be explained, the attribute information after the third is set as “null”. Note that “null” is set in the same manner when the attribute information is insufficient.
  • FIG. 9 is a conceptual diagram showing one of the demand type information 3011A generated by the profiling processing unit 3010.
  • This demand type information 3011A includes a typical demand pattern for each demand type.
  • the demand type information 3011A has a table configuration including a demand type set ID column 3011A6, a demand type ID column 3011A7, and a plurality of time columns 3011A9.
  • the demand type set ID column 3011A6 stores the demand type set ID assigned to each demand type set.
  • the demand type ID column 3011A7 stores the demand of one demand type generated from the corresponding demand type set.
  • the type ID is stored. For this reason, in this demand type information 3011A, the same number of rows as the demand types that can be generated from the demand type set are provided for one demand type set ID, and different demands are stored in the demand type ID column 3011A7 of each row.
  • the type demand type ID is stored.
  • each time column 3011A9 stores a value representing a typical demand pattern in the corresponding time of the corresponding demand type of the corresponding demand type set.
  • the demand type “S1-DT001” in the demand type set “S1” has the time 1 “0.66”, the time 2 “0.28”, and the time 3 “-0.088”. It is shown that this is a demand pattern connected to the series.
  • the time here means, for example, a unit of 30 minutes.
  • the demand pattern shown in FIG. 9 shows the demand pattern obtained as a result of normalizing the original data so that the average becomes 0 and the standard deviation becomes 1 at the time of the data generation processing in the classification granularity adjustment processing unit 3008.
  • the classification granularity adjustment processing unit 3008 may proceed without normalization, and in this case, the unit of the demand pattern value shown in FIG. 9 is the data recorded in the demand record information 4006A. It is the same as the unit.
  • FIG. 10 shows a processing procedure for energy demand prediction processing. This process is started when the demand type management apparatus 30 receives an input operation from the demand / supply manager 3, and the process from step S101 to step S105 is executed by the demand type management apparatus 30 to predict The processing from step S106 to step S108 is executed by the arithmetic unit 31.
  • processing is executed based on various computer programs stored in the CPU 3001 and the storage device 3005 of the demand type management apparatus 30, and various computers stored in the CPU 3101 and the storage device 3105 of the prediction arithmetic device 31. Processing is executed based on the program.
  • the processing entity will be described as various computer programs included in the demand type management device 30 and the prediction calculation device 31.
  • the basic data extraction unit for demand pattern generation 3006 extracts the actual demand information 4006A for the preset period of the customer 10 as the basic data for demand pattern generation from the actual demand information 4006A (S101). .
  • the predicted value evaluation data generation unit 3007 obtains the demand record information 4006A of all the consumers 10 of the past evaluation target date and time set in advance from the demand record information 4006A, and adds them up, for example, every 30 minutes. Thus, evaluation value evaluation data is generated (S102).
  • the demand pattern generation basic data extraction unit 3006 extracts the demand pattern for each combination of the contract demand granularity and the time granularity set in advance by the classification granularity adjustment processing unit 3008 by the supply and demand manager 3.
  • the generation basic data is shaped, and demand pattern generation data corresponding to the set number of combinations is created (S103).
  • the basic data extraction unit for demand pattern generation 3006 extracts data for the past year, and the combination of the granularity of contract demand and the granularity of time is determined.
  • total contract demand and “24 hours” as shown in the third line of FIG. 6, the total contract demand is obtained by adding the demands of all consumers 10 minutes of the basic data for generating the demand pattern.
  • demand pattern generation unit data By dividing the total contract demand in units of time, it is shaped into demand pattern generation data consisting of a total of 365 data (hereinafter referred to as demand pattern generation unit data) with 24 hours as one.
  • demand pattern generation unit data demand pattern generation data consisting of a total of 365 data (hereinafter referred to as demand pattern generation unit data) with 24 hours as one.
  • demand pattern generation unit data Such shaping processing is similarly repeated for other combinations shown in FIG. 6 to generate demand pattern generation data for each combination shown in FIG.
  • the segmentation processing unit 3009 is a portion of the demand pattern generation unit data having similar trends in periodic fluctuations in demand.
  • the data is classified into sets, and a demand pattern for each representative subset (hereinafter referred to as a demand pattern) is generated based on the classification result (S104).
  • a demand pattern for each representative subset (hereinafter referred to as a demand pattern) is generated based on the classification result (S104).
  • the specific processing content of step S104 will be described later with reference to FIG.
  • the profiling processing unit 3010 based on the representative demand pattern generated by the segmentation processing unit 3009, common attribute information from the attribute information group of the customer 10 that generates each of the demand patterns,
  • the demand type information 3011A is generated by extracting from the customer information 4007A, the weather past information 9001A, and the industry dynamic past information 9003A, and the generated demand type information 3011A is transmitted to the prediction arithmetic device 31, and the demand type information storage unit 3011 (S105).
  • the profiling processing unit 3010 generates a CART based on the relationship information between the representative demand pattern ID generated by the segmentation processing unit 3009 and the customer ID associated with the contract demand belonging to this demand pattern.
  • the demand type information 3011A shown in FIGS. 7, 8, and 9 is generated using a decision tree creation algorithm such as ID3 and random forest.
  • the evaluation predicted value calculation processing unit 3106 for each of the sets of demand type information 3011A generated by the profiling processing unit 3010, demand type information 3011A, sales contract forecast information 4008A, customer information 4007A, past weather information 9001A, Based on the industrial dynamic past information 9003A, the demand on the past date and time preset by the supply and demand manager 3 is calculated as a predicted value for evaluation (S106). A specific processing procedure will be described later with reference to FIG.
  • steps S104 to S106 are executed for each of the demand pattern generation data generated by the classification granularity adjustment processing unit 3008.
  • the evaluation calculation unit 3107 calculates the evaluation predicted value for each set of demand type information 3011A generated by the evaluation predicted value calculation processing unit 3106 and the evaluation actual value generated by the predicted value evaluation data generating unit 3007. The difference is calculated, a set of demand type information 3011A that minimizes the difference is extracted, and the demand type information 3011A is updated (S107).
  • the final predicted value calculation processing unit 3108 manages the supply and demand based on the demand type information 3011A, the sales contract prediction actual information 4008A, the weather forecast information 9002A, and the industrial movement information 9004A extracted by the evaluation predicted value calculation processing unit 3106.
  • the demand forecast value of the prediction target date and time preset by the person 3 is calculated and registered in the demand forecast information 4009A (S108).
  • the specific processing procedure for calculating the demand forecast value is as follows: sales contract forecast information 4008A, weather forecast information 9002A, and industry dynamics information 9004A input as past information in the processing procedure of the evaluation forecast value calculation processing unit 3106 in step S106. Is input as sales contract forecast actual information 4008A, weather forecast information 9002A, and industrial dynamic past information 9003A, which are information on future prediction target date and time, and can be realized by the same processing procedure.
  • FIG. 11 shows specific processing contents of the segmentation processing unit 3009 in step S104 of the energy demand prediction processing described above with reference to FIG. This process is executed by the segmentation processing unit 3009 and the CPU 3001 when the classification granularity adjustment processing unit 3008 is finished or triggered by the operation of the demand / supply manager 3.
  • the segmentation processing unit 3009 generates a feature quantity representing a periodic fluctuation of demand for each piece of demand pattern generation unit data of the demand pattern generation data generated by the classification granularity adjustment processing unit 3008 (S111). Specifically, each demand pattern generation unit data is normalized so that the average is 0 and the standard deviation is 1, and then subjected to Fourier transform processing to extract frequency components, and this is generated for each demand pattern. This is the feature quantity of the unit data.
  • the segmentation processing unit 3009 sets one or more patterns for classifying the unit data for demand pattern generation (S112). Specifically, it is set to classify into two, to classify into three, or the like.
  • the segmentation processing unit 3009 selects one of the number of patterns to be classified set in step S112, and classifies the demand pattern generation unit data into the selected number of patterns based on the feature amount generated in step S111. (S113). Specifically, if the number of patterns to be classified is 2, the frequency component, which is the feature quantity of each demand pattern generation unit data, is input, and there is no supervision of neighborhood optimization such as k-means, EM algorithm, or spectral clustering.
  • the unsupervised clustering algorithm for discriminating surface optimization such as the unsupervised clustering algorithm or unsupervised SVM (Support Vector Machine), VQ algorithm, and SOM (Self-Organizing Maps) (Subset).
  • the segmentation processing unit 3009 calculates an evaluation index value for evaluating the number of patterns based on the classification result (S114). Specifically, the segmentation processing unit 3009 uses an index that measures the cohesiveness inside each classified data group such as the Akaike information criterion, and an index that measures the separability between each classified data group such as a margin. The evaluation index value for the number of patterns is calculated.
  • the segmentation processing unit 3009 repeats the above steps S113 and S114 for the number of patterns to be classified set in step S112.
  • the segmentation processing unit 3009 determines the number of patterns to be classified based on the evaluation index value for each number of patterns calculated in step S114, and extracts the result of classification based on the number of patterns (S115). Specifically, the segmentation processing unit 3009 classifies the number that minimizes the index that measures the cohesiveness inside each subset, the number that maximizes the index value that measures the separability between the subsets, and the like. The number of patterns is determined, and the result classified by the number of patterns is extracted from the calculation result in step S113.
  • the segmentation processing unit 3009 generates a representative demand pattern from the extracted classification result (S116). Specifically, the segmentation processing unit 3009 generates a representative demand pattern by calculating an average value of feature quantities of each classified demand pattern generation unit data and performing inverse Fourier transform on the calculated average value. To do.
  • FIG. 12 shows specific processing contents of the evaluation predicted value calculation processing unit 3106 in step S106 of the energy demand prediction processing described above with reference to FIG. This process is executed by the evaluation predicted value calculation processing unit 3106 and the CPU 3101 when the profiling processing unit 3010 ends or triggered by the operation of the demand and supply manager 3.
  • the evaluation predicted value calculation processing unit 3106 first has sales contract prediction actual information 4008A, weather past information 9001A, industry dynamic past information 9003A, demand record information 4006A, and the like of the evaluation target date and time preset by the supply and demand manager 3. Based on the above, all the demand types related to the evaluation target date and time are extracted from the demand type information 3011A generated by the profiling processing unit 3010 (S121).
  • the evaluation predicted value calculation processing unit 3106 first extracts from the sales contract forecast actual information 4008A all demand types whose actual value or planned value of the power contract capacity is not zero at the forecast target date and time. Then, the evaluation predicted value calculation processing unit 3106 sorts out the demand types that match the weather of the target date and time and the information of the industry dynamic information 9004A among the extracted demand types based on the attribute information column 3011A8 of the demand type information 3011A. Thus, the demand type ID related to the target date and time is extracted, and the demand pattern information associated with the extracted demand type ID is acquired from 3011A9.
  • the predicted value calculation processing unit 3106 for evaluation selects one of the extracted demand types, acquires the actual value and the planned value of the sales contract capacity of this demand type from the sales contract predictive information 4008A, and sets the target date and time.
  • the ratio between the planned value of contract power capacity and the actual value of contract power capacity at any past date and time is calculated, and the maximum value and minimum value of the demand on the prediction target date are calculated based on the calculated ratio and the actual demand information 4006A. (S122).
  • the evaluation predicted value calculation processing unit 3106 Acquire the actual value of "3000kW” as the demand type contracted power capacity. Further, when an arbitrary past day is “April 5, 2015” corresponding to the first week of FY2015, the actual value “2000 kW” is acquired as the contract power capacity of this demand type. Therefore, the ratio is Is calculated as
  • the evaluation predicted value calculation processing unit 3106 calculates the maximum value and the minimum value of the demand type “S1-DT001” “April 5, 2015” respectively by the following formulas (2) and (3): It calculates based on the regression equation shown in FIG.
  • y ⁇ and z ⁇ are the explained maximum and minimum demand values, respectively
  • x1, x2, x3, and x4 are explanatory variables, for example, x1 is the month, x2 is the average temperature, x3 is the actual value of the minimum demand on the previous day, and x4 is the minimum temperature.
  • a, b, c, d, e, f, g, and h are coefficients.
  • the evaluation predicted value calculation processing unit 3106 first determines the maximum and minimum values of demand, which are explained variables, from the past record of the demand type “S1-DT001” stored in the demand record information 4006A. , The explanatory variables month, average temperature, the actual value of the minimum demand on the previous day, and the minimum temperature are extracted. Next, the evaluation predicted value calculation processing unit 3106 estimates the values of the coefficients a to h by the least square method based on the extracted past results. At this time, the evaluation predicted value calculation processing unit 3106 multiplies the extracted past performance by a forgetting weight coefficient, thereby estimating a coefficient that places more importance on reproducibility with respect to the latest past performance value. Specifically, the evaluation predicted value calculation processing unit 3106 estimates the values of the coefficients a to d that minimize the following equation (4), for example, in the case of the regression equation of the maximum demand value.
  • N is the number of extracted past actual values
  • y n is the actual value of the extracted past maximum demand value
  • w n is the weight coefficient corresponding to the extracted n-th y n and y ⁇ n. It is.
  • the weighting factor is the reciprocal of the past number of days, for example, if the past actual value one day ago is 1/2, if it is one day ago, 1/2, if it is three days ago, it is 1/3.
  • the evaluation predicted value calculation processing unit 3106 calculates the predicted value of the maximum value and the minimum value of the demand target day by multiplying the calculated ratio with respect to each of the calculated maximum value and the minimum value. To do.
  • the evaluation predicted value calculation processing unit 3106 is configured so that the maximum and minimum predicted values for the target day calculated in step S122 match the maximum and minimum values for the target date of the demand pattern, respectively.
  • the entire demand pattern is adjusted (S123).
  • the evaluation predicted value calculation processing unit 3106 performs the above steps S122 and S123 for all the demand types extracted in step S121.
  • the evaluation predicted value calculation processing unit 3106 calculates the demand predicted value on the target day by adding all the adjusted demand patterns for each time zone (S124).
  • the facility manager 6 uses the facility management device 61 to generate an operation plan for an operable power generation facility, and the control device 62 Send to.
  • the control device 62 that has received the operation plan generates a specific control plan for the power generation facility and executes actual control.
  • the transaction manager 5 uses the transaction management device 50 to create a transaction plan related to the procurement of electric power with other electric utilities and the transaction market, and with respect to the market operation management device 80, the purchase order and the order cancellation. Send a message such as
  • FIG. 13 is a conceptual diagram for explaining the effect and principle of the present embodiment.
  • the graph 1300 shown in FIG. 13 shows the error of the demand forecast value when using the granularity of the time unit divided on the horizontal axis and the demand type information 3011A generated at the respective time unit granularity on the vertical axis.
  • the classification granularity is too small, as shown in 1301A, the periodic feature information of the demand performance information 4006A disappears, so that the prediction error becomes large.
  • the classification granularity is excessive, as shown in 1303A, since the information amount of the feature indicating a large cycle is larger than the information amount indicating the feature of a small cycle, the difference in the feature of the small cycle cannot be captured. The prediction error becomes large.
  • an optimum division granularity as shown in 1302A is determined, and therefore the error in the demand forecast value is minimized. Note that FIG. 13 refers only to the granularity of time units for simplicity of explanation, but the same applies to the units of consumers.
  • one or more sets are set based on the past demand record information 4006A for each measured customer 10.
  • the generation of demand pattern generation data used to generate a demand pattern which is a typical energy consumption trend, is generated.
  • one or more demand patterns indicating typical demand trends are generated, and the attributes that exist in common among the attributes of the customers who generate the generated demand patterns are the customer information 4007A and the weather forecast information 9002A.
  • a demand type is generated, and based on the generated demand type, sales contract forecast information 4008A and information that can be an attribute such as weather past information 9001A and industrial dynamic past information 9003A are used to calculate the total contract demand of any target date and time.
  • a predicted value is calculated for each demand type set, a demand type set having a minimum error between the predicted value and an actual observed value is extracted, and a final demand predicted value is calculated using the extracted demand type set. Therefore, according to the energy demand management system 1, it is possible to make a demand prediction while the continuous characteristics of the total contract demand change in a short period.
  • the maximum demand can be estimated from the demand attribute, so that the procurement amount of electricity and the sales amount are matched at every settlement time, instead of a predetermined period in advance. It is possible to estimate the maximum demand and to apply in advance for long-term procurement of power and transmission planning.
  • the demand type is a unit time (for example, 30 minutes) of the contract customer in a predetermined period (for example, one month or one week).
  • the maximum demand which is the maximum demand, and information consisting of attributes.
  • the feature of the maximum demand value is divided into subsets using a clustering algorithm, and the demand pattern Is generated. Further, instead of the block 3108, a demand pattern is estimated from the attribute of demand, and thereby the maximum demand for a predetermined period is estimated.
  • the transaction management device 50 uses the estimated (predicted) value of the maximum demand to make a power procurement plan for the predetermined period, and sends a message to the market operation management device 80 to execute the plan.
  • the ratio of the contract power capacity at a certain past time point and the prediction target time point is multiplied by the actual demand value at the past time point.
  • the present invention is not limited to this.
  • the prediction of the maximum value and the minimum value of the demand at the time of the prediction using a multiple regression model, an autoregressive model, etc. The value may be multiplied by a ratio. Thereby, it becomes possible to switch to a model with higher demand prediction accuracy.
  • an actually observed value may be used as the maximum value or the minimum value at the prediction target time point.
  • the demand can be predicted with higher accuracy reflecting the actually observed values.
  • the contract power capacity is used for calculating the ratio, but the present invention is not limited to this, and the ratio may be calculated based on the number of contracts. Thereby, it is possible to predict the maximum value or the minimum value of demand according to the increase or decrease in the number of contracts.
  • the estimated value to calculate was demonstrated as a series of a 30-minute unit of a certain day, it is not restricted to this, For example, only the maximum value of a day, a week, a month, and a year is shown. It may be calculated and output. As a result, it is possible to output a predicted value at a desired time section.
  • the predicted value to be finally calculated has been described as the value of the total contract demand.
  • the present invention is not limited to this, and any contract demand may be performed alone. This makes it possible to predict demand only for a specific area or demand type group.
  • a series of processes related to generation of demand types is performed only once for the sake of simplicity.
  • the present invention is not limited to this. For example, daily, weekly, monthly It may be executed at a predetermined interval such as a unit, or may be executed at an arbitrary time by an operation of the supply and demand manager 3. As a result, it is possible to prevent deterioration in demand prediction accuracy due to changes in the size and manner of demand over time.
  • the energy demand management system 1 for the sake of simplicity of description, all combinations of contract demand and time granularity are defined in advance, and the configuration in which the combination having the smallest predicted value is selected from among the combinations.
  • the present invention is not limited to this, and for example, one of the above combinations is set first, and thereafter, an optimization process is performed so as to search for an optimal combination while repeating prediction value evaluation and resetting the combination. It may be a configuration. Thereby, it becomes possible to optimize the computational resources required by the energy demand management system 1 to a necessary and sufficient amount.
  • the energy demand management system 1 in this Embodiment it demonstrated as a structure of the Example which determines the combination of a contract demand and the granularity of time based only on the demand track record information 4006A observed in the past for the sake of simplicity.
  • the present invention is not limited to this, and for example, it may be configured as an online process that performs a series of processes related to the generation of a demand type at the same time as observing a demand result. This makes it possible to perform adaptive demand prediction that always follows changes in the size and manner of demand over time.
  • the basic data for generating the demand pattern may be filtered based on attribute information that is arbitrarily set in advance, such as a midsummer day or the type of device, and then used for subsequent processing.
  • attribute information such as a midsummer day or the type of device
  • the explanatory variables of the regression equation used when calculating the maximum value and the minimum value of the demand are respectively the month and the average temperature, and the actual value and the minimum value of the minimum value of the previous day.
  • temperature it is not limited to this.
  • the maximum, minimum, and average values of weather information such as temperature, humidity, solar radiation, sunshine duration, precipitation, snowfall, wind speed, atmospheric pressure, etc.
  • Customer information such as past values of demand, such as actual values of maximum values and minimum values, or the number of occupants in a customer building may be used as explanatory variables.
  • the regression equation used may be a linear equation or a nonlinear regression equation such as a linear equation, a second or higher order polynomial, a Fourier series, or a neural network, or may be an AR model, an ARMA model, an ARIMA model, a VAR model, or the like.
  • An autoregressive model may be used.
  • For autoregressive models in determining the target delay element and its number (order), calculate the autocorrelation or partial autocorrelation for the past data, or both, and for the delay that has no statistically significant correlation
  • the weight may be zero. Thereby, it is possible to avoid a decrease in prediction accuracy due to delayed data having no autocorrelation, and to reduce the processing load of parameter estimation of the autoregressive model.
  • the form and number of explanatory variables to be used the form of a regression equation
  • One of two or more types of regression models with different values may be selected on the basis of an information criterion such as AIC, and similarly, the calculation results of a plurality of regression models may be weighted and averaged based on the information criterion.
  • the maximum value and the minimum value of demand may be calculated. This makes it possible to automatically select the most suitable demand prediction model among changes in the size and manner of demand over time, and to perform adaptive prediction with respect to the change.
  • the forgetting weight coefficient used when estimating the coefficient of the regression equation for calculating the maximum value and the minimum value of the demand has been described as the reciprocal of the past number of days.
  • an arbitrary function such as an exponential function or a sigmoid function may be used, and a method of giving a smaller weight of the past actual value in a period in which the forecast target date and the climate are reversed may be performed.
  • the demand type information created by using a decision tree creation algorithm such as CART, ID3, random forest, etc. is used without any particular processing.
  • the present invention is not limited to this.
  • classification accuracy of demand pattern and outlier determination for attribute value range of branch condition of each branch Based on the attribute and attribute value range set in advance, one or more types are used without evaluating branches or leaves that fall below the certainty level defined by a preset threshold, or both.
  • the demand pattern information and the demand pattern may be substituted. Thereby, it becomes possible to maintain the correct answer rate of the demand pattern identification by the decision tree at a certain level.
  • the number of divisions when the demand pattern generation unit data is divided into subsets and the demand pattern is extracted is the minimum index for measuring the cohesiveness inside each subset.
  • the present invention is not limited to this, but is not limited to this, and it is not possible to identify errors in demand pattern identification using decision trees generated by the profiling processing unit.
  • the number of divisions may be determined so that this is minimized using the identification rate as an index, or the number of divisions may be determined so that this is minimized using the final prediction error as an index. This makes it possible to predict demand at a certain level even when there is not enough usable attribute information.
  • the display unit is omitted for the sake of simplicity of explanation.
  • the output result of each processing unit and the intermediate result of each processing unit are output to a display or a printer. You may output suitably through an apparatus.

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

L'invention aborde le problème de prédiction du besoin en des circonstances dans lesquelles des caractéristiques de besoins contractés totaux continus varient sur une courte période de temps. La solution selon l'invention consiste en ce que ce procédé de prédiction de besoin en énergie est configuré de telle sorte : que des ensembles de données de génération de motif de besoin pour chaque combinaison parmi une ou plusieurs combinaisons prédéfinies d'une granularité de besoin contracté et d'une granularité de temps sont générés sur la base d'informations de besoin en énergie réelles pour une période historique par défaut ; qu'un type de besoin est généré pour chaque ensemble de données de génération de motif de besoin ; qu'un besoin en énergie à une date et une heure historiques par défaut est calculé comme valeur prédite pour l'évaluation, pour chaque type de besoin ; qu'une granularité de besoin contracté et une granularité de temps sont déterminées sur la base de la valeur prédite pour l'évaluation pour chaque type de besoin et des informations de besoin en énergie réelles, de manière à minimiser une erreur entre une valeur estimée ou une valeur prédite de besoin en énergie à une date et une heure historiques, et la valeur observée réelle à ladite date et à ladite heure ; et que la valeur de besoin en énergie à une date et une heure définies arbitrairement est estimée ou prédite sur la base des résultats déterminés.
PCT/JP2016/063696 2015-05-21 2016-05-09 Système de prédiction de besoin en énergie et procédé de prédiction de besoin en énergie WO2016185919A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP16796313.1A EP3300205A4 (fr) 2015-05-21 2016-05-09 Système de prédiction de besoin en énergie et procédé de prédiction de besoin en énergie
US15/574,187 US20180128863A1 (en) 2015-05-21 2016-05-09 Energy Demand Predicting System and Energy Demand Predicting Method

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
JP2015-104036 2015-05-21
JP2015104036 2015-05-21
JP2015211260A JP6408451B2 (ja) 2015-05-21 2015-10-27 エネルギー需要予測システム及びエネルギー需要予測方法
JP2015-211260 2015-10-27

Publications (1)

Publication Number Publication Date
WO2016185919A1 true WO2016185919A1 (fr) 2016-11-24

Family

ID=57320203

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2016/063696 WO2016185919A1 (fr) 2015-05-21 2016-05-09 Système de prédiction de besoin en énergie et procédé de prédiction de besoin en énergie

Country Status (1)

Country Link
WO (1) WO2016185919A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111784046A (zh) * 2020-06-30 2020-10-16 中国人民解放军国防科技大学 一种预估风暴轴活动未来变化趋势的方法
CN112749473A (zh) * 2020-09-14 2021-05-04 万洲电气股份有限公司 基于实时在线分析和模拟离线分析的能效安全诊断系统

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH09285010A (ja) * 1996-04-11 1997-10-31 Toshiba Corp 電力需要予測支援装置
JP2014164393A (ja) * 2013-02-22 2014-09-08 Mitsubishi Electric Corp 電力予測装置、設備制御システム、電力予測方法、及びプログラム

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH09285010A (ja) * 1996-04-11 1997-10-31 Toshiba Corp 電力需要予測支援装置
JP2014164393A (ja) * 2013-02-22 2014-09-08 Mitsubishi Electric Corp 電力予測装置、設備制御システム、電力予測方法、及びプログラム

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3300205A4 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111784046A (zh) * 2020-06-30 2020-10-16 中国人民解放军国防科技大学 一种预估风暴轴活动未来变化趋势的方法
CN111784046B (zh) * 2020-06-30 2023-12-22 中国人民解放军国防科技大学 一种预估风暴轴活动未来变化趋势的方法
CN112749473A (zh) * 2020-09-14 2021-05-04 万洲电气股份有限公司 基于实时在线分析和模拟离线分析的能效安全诊断系统
CN112749473B (zh) * 2020-09-14 2023-11-28 万洲电气股份有限公司 基于实时在线分析和模拟离线分析的能效安全诊断系统

Similar Documents

Publication Publication Date Title
JP6895416B2 (ja) エネルギー需要予測システム及びエネルギー需要予測方法
JP7319757B2 (ja) データ処理システム及びデータ処理方法
JP6837949B2 (ja) 予測システム及び方法
JP6742894B2 (ja) データ予測システムおよびデータ予測方法
JP2018092439A5 (fr)
US11107094B2 (en) Prediction system and prediction method
JP6474782B2 (ja) データ分析システム及び方法
JP2018163515A (ja) 予測システム、及び、予測方法
Park et al. A framework for baseline load estimation in demand response: Data mining approach
US20190279239A1 (en) Power Trading Assistance Device and Market Price Prediction Information Generation Method
Welisch et al. Assessment of RES technology market values and the merit-order effect–an econometric multi-country analysis
JP2017142781A (ja) 確率的価格及び急変予測
Huang et al. A Bayesian Network model for predicting cooling load of commercial buildings
WO2017212880A1 (fr) Système de prédiction de données et procédé de prédiction de données
Voronin et al. A hybrid electricity price forecasting model for the Nordic electricity spot market
Arjmand et al. Meta-learning in multivariate load demand forecasting with exogenous meta-features
WO2016185919A1 (fr) Système de prédiction de besoin en énergie et procédé de prédiction de besoin en énergie
Jiang et al. An optimized forecasting approach based on grey theory and Cuckoo search algorithm: a case study for electricity consumption in New South Wales
Singla et al. Load forecasting using back propagation algorithm
Haq Machine learning for load profile data analytics and short-term load forecasting
Utsumi et al. Forecasting Electricity Demand with Dynamic Characteristics Based on Signal Analysis and Machine Learning
Gajowniczek et al. Seasonal peak demand classification with machine learning techniques
Sinitsyna et al. Identification of Electric Power Consumption Patterns in Relation to Standard Load Profiles
Inala et al. Optimal Load Prediction in a Smart Metering Network using Random Forest Algorithm
Zhao Research on E-Commerce Retail Demand Forecasting Based on SARIMA Model and K-means Clustering Algorithm

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16796313

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15574187

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE