Microgrid demand response identification and optimal scheduling method and system considering satisfaction degree

Abstract

The application provides a micro-grid demand response identification and optimal scheduling method and system considering satisfaction, an integrated classification identification center is constructed by using a machine learning classifier based on a random forest model, user time-of-use electricity price response behavior is quantified through a demand elasticity coefficient matrix, demand response of the micro-grid is accurately identified, the economy of the collaborative scheduling of multiple micro-grids is improved, and the function of balancing user satisfaction and the economy of the micro-grid by coordinating power balance and minimizing operation cost is realized. The application solves the data anomaly problem through a master-slave management framework and a machine learning classifier, effectively handles the interference of data anomaly on user behavior feature extraction, improves the accuracy of the scheduling strategy, and realizes the collaborative optimization of multiple micro-grids.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of demand response identification, and particularly relates to a micro-grid demand response identification and optimal scheduling method and system considering satisfaction. BACKGROUND

[0002] In the process of deepening the development of smart grids, micro-grids (MG) are the core carriers for integrating distributed energy and user-side resources, and their efficient operation is crucial for improving the flexibility and supply-demand balance of power grids. However, current micro-grids face significant theoretical challenges in customer-side demand response (DR) management, multi-micro-grid collaborative scheduling, and user experience optimization.

[0003] Firstly, micro-grid operation relies on massive user power consumption data, but factors such as communication failures, equipment errors, or network security threats often result in data missing, label errors, or time sequence disorder. Traditional identification methods based on rule matching or single algorithms cannot effectively handle the interference of data anomalies on user behavior feature extraction, leading to ambiguous DR participation state judgment and affecting the accuracy of scheduling strategies.

[0004] Secondly, the game equilibrium problem of multi-micro-grid collaborative scheduling restricts the overall efficiency of the system. Existing collaborative mechanisms mostly use centralized optimization frameworks, aiming to maintain system balance by sacrificing individual economic benefits of micro-grids. This asymmetric collaborative mode ignores the decision-making autonomy of micro-grids as independent market subjects, resulting in insufficient endogenous motivation for their participation in collaboration, especially in renewable energy fluctuation scenarios, which easily leads to conflicts of interest and power imbalance among micro-grids. The lack of quantitative evaluation of user satisfaction and integration of scheduling objectives is another theoretical gap. Traditional micro-grid scheduling takes economic or reliability as a single optimization target, lacking quantitative consideration of users' subjective experience in participating in DR. The evaluation of user satisfaction has not formed a theoretical system deeply coupled with the physical model of micro-grids, resulting in scheduling strategies that cannot respond to the differentiated needs of user groups. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a micro-grid demand response identification and optimal scheduling method and system considering satisfaction, which balances user satisfaction and the economic efficiency of micro-grids by coordinating power balance and minimizing operating costs.

[0006] The technical solution adopted by the present application to solve the above technical problems is: a micro-grid demand response identification and optimal scheduling method considering satisfaction, comprising the following steps:

[0007] S1: initialization, inputting information of the micro-grid;

[0008] S2: Constructing master-slave management architecture and economic optimization model of microgrid according to input information, respectively for coordinating power balance and minimizing operation cost;

[0009] S3: Using machine learning classifier based on random forest model to construct integrated classification recognition center called by master-slave management architecture, quantifying user time-of-use electricity price response behavior through demand elasticity coefficient matrix, and identifying customer side demand response participation label;

[0010] S4: Introducing electricity purchase cost evaluation model and power supply power evaluation model through customer side demand response participation label, dynamically adjusting scheduling strategy based on user actual cost and power deviation, quantifying satisfaction and balancing economy and power supply reliability.

[0011] According to the above scheme, in step S1, the information of the microgrid includes microgrid operation data, customer side demand response data, machine learning input features and interactive collaborative scheduling data.

[0012] According to the above scheme, in step S2, the master-slave management architecture of the microgrid includes master microgrid, slave microgrid and upper grid;

[0013] The master microgrid is a power output party, including distributed power supply and electric / thermal energy storage system; the master microgrid is used to generate low-price power sales plan through optimal scheduling;

[0014] The slave microgrid is a power input party, used to adjust electricity purchase strategy according to the power sales plan, realizing economic collaboration of multi-microgrid and upper grid;

[0015] The upper grid is a system-level energy hub, used to provide peak-valley time-of-use electricity price signal, guide electricity consumption behavior of the microgrid, and collaboratively schedule to balance and stably operate the master microgrid and the slave microgrid.

[0016] Further, in step S2, the collaborative scheduling of the master microgrid and the slave microgrid follows a closed-loop process of plan generation, cross-network transaction and grid collaboration, and the specific steps are:

[0017] S21: The master microgrid generates a power sales plan containing multiple time periods based on analysis of next-day renewable energy prediction, load curve and energy storage state, and shares the plan in real time to the slave microgrid through the master-slave management platform;

[0018] S22: The slave microgrid matches the power sales plan of the master microgrid according to the demand response label of customers, preferentially purchases low-price power for adjustable load; for customers not participating in demand response, the power supply stability is ensured through energy storage system or upper grid;

[0019] S23: If the main micro-grid is insufficient, the slave micro-grid purchases electricity from the upper grid to make up for the gap, and the upper grid dynamically adjusts the electricity price according to the system load state; the main micro-grid, the slave micro-grid and the upper grid regularly synchronize the power data to correct the prediction error and adjust the transaction plan of the subsequent period in real time, so as to ensure the power balance and economic optimization between micro-grids.

[0020] According to the above scheme, in the step S2, the economic optimization model is used to optimally configure the internal resources of the micro-grid through mathematical modeling, and the comprehensive cost of the micro-grid is minimized under the conditions of satisfying the power balance constraint, the energy storage state of charge constraint and the equipment operation constraint.

[0021] Further, in the step S2, the comprehensive cost is the sum of the total fuel cost, the power interaction cost, the system maintenance cost and the controllable unit start-stop cost, minus the heat energy sale income.

[0022] The total fuel cost includes the gas turbine operation cost and the fuel cell operation cost; the gas turbine operation cost is the unit cost of natural gas divided by the unit low price, and then multiplied by the total heat consumed by the gas turbine; the fuel cell operation cost is the unit cost of natural gas divided by the unit low price, and then multiplied by the total heat consumed by the fuel cell.

[0023] The power interaction cost is the sum of the average of the purchase price and the sale price multiplied by the interaction power between the micro-grid and the upper grid, and half of the difference between the purchase price and the sale price multiplied by the absolute value of the interaction power between the micro-grid and the upper grid.

[0024] The system maintenance cost is the sum of the product of the maintenance cost of each customer of each unit in the micro-grid and the power.

[0025] The controllable unit start-stop cost is the sum of the start-up costs when the adjacent two start-stop states of the controllable unit are inconsistent.

[0026] The heat energy sale income is the product of the unit price of heat energy sales and the heat load power.

[0027] Further, in the step S2,

[0028] The power balance constraint includes:

[0029] The sum of the power of each customer of each unit in the micro-grid minus the charging power of the energy storage system, plus the discharging power of the energy storage system, the interaction power between the micro-grid and the upper grid, is equal to the sum of the load power of the micro-grid and the electric power of the electric boiler.

[0030] The heat generated by the gas turbine plus the heat generated by the electric boiler minus the absorption power of the heat storage system plus the release power of the heat storage system is equal to the heat load power.

[0031] The heat generated by the gas turbine is equal to the ratio of the portion of the power of the gas turbine that is spent on heat loss and the power generation efficiency;

[0032] The energy storage state of charge constraint includes:

[0033] The product of the maximum charging rate of the electrical energy storage system and the capacity of the electrical energy storage system is taken as the lower limit of the range of the power of the electrical energy storage system, and the product of the maximum discharging rate of the electrical energy storage system and the capacity of the electrical energy storage system is taken as the upper limit of the range of the power of the electrical energy storage system; the product of the maximum charging rate of the thermal energy storage system and the capacity of the thermal energy storage system is taken as the lower limit of the range of the power of the thermal energy storage system, and the product of the maximum discharging rate of the thermal energy storage system and the capacity of the thermal energy storage system is taken as the upper limit of the range of the power of the thermal energy storage system;

[0034] The product of the upper limit and the lower limit of the state of charge and the capacity of the electrical energy storage system is taken as the upper limit and the lower limit of the range of the capacity of the electrical energy storage system, respectively;

[0035] The initial capacity and the final capacity of the electrical energy storage system are equal;

[0036] The capacity of the electrical energy storage system is the remaining part of the capacity of the electrical energy storage system at the previous time after self-discharge, plus the product of the charging power and the charging efficiency of the electrical energy storage system in a unit of time minus the quotient of the discharging power and the discharging efficiency; the capacity of the thermal energy storage system at the current time is the remaining part of the capacity of the thermal energy storage system at the previous time after heat loss, plus the product of the heat absorption power and the heat absorption efficiency of the thermal energy storage system in a unit of time minus the quotient of the heat release power and the heat release efficiency;

[0037] The equipment operation constraint includes:

[0038] The upper limit power and the lower limit power of a certain controllable unit are set;

[0039] The upper and lower limit ranges of the power difference of a certain controllable unit at adjacent times are set as the rising slope rate and the falling slope rate.

[0040] According to the above scheme, in the step S3, the specific steps are:

[0041] S31: Build a data chain, including a collection layer and a preprocessing layer;

[0042] The collection layer is used to collect basic data, price signals and environmental data in real time;

[0043] The basic data includes a customer ID, a time period and real-time active power; the customer ID is a unique identifier of a user, used to distinguish different electricity consumers; the time period is an interval with a fixed time length, used to provide a time scale for capturing the time variation of electricity consumption behavior; the real-time active power is used to reflect the power consumption of the customer in each time period;

[0044] The price signal includes an upper grid time-of-use price and a demand response incentive price; the upper grid time-of-use price is used to define different price periods of peak, flat and valley; the demand response incentive price is formulated from the microgrid;

[0045] The environmental data includes temperature and light intensity; the temperature affects the electricity demand of the customer; the light intensity determines the output of photovoltaic power generation;

[0046] The preprocessing layer is used to repair missing values of data by using linear interpolation method, and to identify abnormal data by using quartile range algorithm to filter out noise points in the data;

[0047] S32: Adopting a random forest model to classify the demand response participation state, taking the price response feature as the classification feature, and using the demand elasticity coefficient matrix to describe the demand response, so as to identify the demand response on the customer side;

[0048] The demand elasticity coefficient matrix is obtained by statistical analysis of the customer-side electricity consumption under the historical time-of-use price; further, the power of each time period after the price adjustment of the time-of-use price, and the probability density function of the adjustment power of the participating customers are obtained.

[0049] According to the above scheme, in step S4, the electricity purchase cost evaluation model is: the proportion of the difference between the actual situation and the ideal situation of a certain user participating in demand response to the electricity fee of the user participating in demand response in the ideal situation, subtracted from the initial satisfaction degree of the user;

[0050] The power supply power evaluation model is: the proportion of the difference between the actual situation and the ideal situation of a certain user not participating in demand response to the power demand of the user not participating in demand response in the ideal situation, subtracted from the initial satisfaction degree of the user;

[0051] The proportion is divided by a unit interval parameter to eliminate evaluation bias.

[0052] A microgrid demand response identification and optimal scheduling system considering satisfaction degree,

[0053] An initialization submodule is used for initialization, and information of the microgrid is input;

[0054] A modeling submodule is used for constructing a master-slave management architecture and an economic optimization model of the microgrid according to the input information, which are respectively used for coordinating power balance and minimizing operation cost;

[0055] The demand identification sub-module is configured to utilize a machine learning classifier based on a random forest model to construct an integrated classification identification center for calling of a master-slave management architecture, quantify user time-of-use electricity price response behavior through a demand elasticity coefficient matrix, and identify a customer-side demand response participation label;

[0056] The evaluation sub-module is configured to introduce a power purchase cost evaluation model and a power supply power evaluation model through the customer-side demand response participation label, dynamically adjust a scheduling strategy based on actual cost and power deviation of the user, quantify satisfaction, and balance economy and power supply reliability.

[0057] The present application has the following beneficial effects:

[0058] 1. The microgrid demand response identification and optimal scheduling method and system considering satisfaction of the present application utilize a machine learning classifier based on a random forest model to construct an integrated classification identification center, quantify user time-of-use electricity price response behavior through a demand elasticity coefficient matrix, accurately identify demand response of the microgrid, improve economy of collaborative scheduling of multiple microgrids, and realize functions of balancing user satisfaction and economy of the microgrid through coordinated power balance and minimized operation cost.

[0059] 2. The present application solves data anomaly problems through a master-slave management framework and a machine learning classifier, effectively handles interference of data anomaly on user behavior feature extraction, improves accuracy of the scheduling strategy, and realizes collaborative optimization of multiple microgrids.

[0060] Of course, implementing any product of the present application does not necessarily need to achieve all the advantages described above at the same time. BRIEF DESCRIPTION OF DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0062] Figure 1 is a flow chart of an embodiment of the present application.

[0063] Figure 2 is a customer electricity demand information diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solutions and advantages of the present application more clear, the following will further describe the present application in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0065] Embodiment 1

[0066] Referring to Figure 1 The specific steps of the micro-grid demand response identification and optimal scheduling method considering satisfaction are as follows:

[0067] S1: initialization, inputting information of the micro-grid;

[0068] S2: constructing a master-slave management architecture and an economic optimization model of the micro-grid according to the input information, which are respectively used for coordinating power balance and minimizing operation cost;

[0069] S3: constructing an integrated classification identification center called by the master-slave management architecture by using a machine learning classifier based on a random forest model, quantifying user time-of-use price response behavior through a demand elasticity coefficient matrix, and identifying a customer-side demand response participation label;

[0070] S4: introducing a power purchase cost evaluation model and a power supply power evaluation model through the customer-side demand response participation label, dynamically adjusting the scheduling strategy based on user actual cost and power deviation, quantifying satisfaction and balancing economic efficiency and power supply reliability.

[0071] Further, in step S1, the information of the micro-grid includes micro-grid operation data, customer-side demand response data, machine learning input features, and interactive collaborative scheduling data.

[0072] In step S2, the master-slave management architecture of the micro-grid includes a master micro-grid, a slave micro-grid, and an upper-level grid;

[0073] The master micro-grid is a power output side, including distributed power sources and electric / thermal energy storage systems; the master micro-grid is used to generate a low-price power sales plan through optimal scheduling;

[0074] The slave micro-grid is a power input side, used to adjust the power purchase strategy according to the power sales plan, and realize economic collaboration of the multi-micro-grid and the upper-level grid;

[0075] The upper-level grid is a system-level energy hub, used to provide a peak-valley time-of-use price signal, guide the power consumption behavior of the micro-grid, and collaboratively schedule the power balance and stable operation of the master micro-grid and the slave micro-grid.

[0076] Further, in step S2, the collaborative scheduling of the master micro-grid and the slave micro-grid follows a closed-loop process of plan generation, cross-grid transaction, and grid collaboration, and the specific steps are as follows:

[0077] S21: the master micro-grid generates a power sales plan including multiple time periods based on analysis of the next day's renewable energy prediction, load curve, and energy storage state, and shares it in real time to the slave micro-grid through the master-slave management platform;

[0078] S22: According to the demand response label of the customer, the master micro-grid matches the power sales plan, and preferentially purchases low-price power for the adjustable load; for the customer who does not participate in the demand response, the power supply stability is ensured by the energy storage system or the upper power grid;

[0079] S23: If the master micro-grid power supply is insufficient, the slave micro-grid purchases power from the upper power grid to make up for the gap, and the upper power grid dynamically adjusts the electricity price according to the system load state; the master micro-grid, the slave micro-grid and the upper power grid regularly synchronize the power data, correct the prediction error, and real-time adjust the transaction plan of the subsequent period, to ensure the power balance and economic optimization between the micro-grids.

[0080] In step S2, the economic optimization model is used to optimally configure the internal resources of the micro-grid through mathematical modeling, and minimize the comprehensive cost of the micro-grid under the conditions of satisfying the power balance constraint, the energy storage state of charge constraint and the device operation constraint.

[0081] Further, in step S2, the comprehensive cost is the sum of the total fuel cost, the power interaction cost, the system maintenance cost and the controllable unit start-stop cost, minus the heat energy sale income.

[0082] The total fuel cost includes the gas turbine operation cost and the fuel cell operation cost; the gas turbine operation cost is the unit cost of natural gas divided by the unit low price, and then multiplied by the total heat consumed by the gas turbine; the fuel cell operation cost is the unit cost of natural gas divided by the unit low price, and then multiplied by the total heat consumed by the fuel cell;

[0083] The power interaction cost is the average value of the purchase price and the sale price multiplied by the interaction power between the micro-grid and the upper power grid, and half of the difference between the purchase price and the sale price multiplied by the absolute value of the interaction power between the micro-grid and the upper power grid, and then summed up;

[0084] The system maintenance cost is the product of the maintenance cost of each customer of each unit in the micro-grid and the power, and then summed up;

[0085] The controllable unit start-stop cost is the sum of the start-up costs when the adjacent two start-stop states of the controllable unit are inconsistent;

[0086] The heat energy sale income is the product of the heat energy sale unit price and the heat load power.

[0087] Further, in step S2,

[0088] The power balance constraint includes:

[0089] The sum of the power of each customer of each unit in the micro-grid minus the charging power of the electric energy storage system plus the discharging power of the electric energy storage system and the interaction power between the micro-grid and the upper grid is equal to the sum of the load power of the micro-grid and the electric power of the electric boiler;

[0090] The heat generated by the gas turbine plus the heat generated by the electric boiler minus the absorption power of the thermal energy storage system plus the release power of the thermal energy storage system is equal to the thermal load power;

[0091] The heat generated by the gas turbine is equal to the ratio of the part of the power consumption of the gas turbine other than the power generation and the heat loss to the power generation efficiency;

[0092] The energy storage state of charge constraint includes:

[0093] The product of the maximum charging rate of the electric energy storage system and the capacity of the electric energy storage system is taken as the lower limit of the range of the power of the electric energy storage system, and the product of the maximum discharging rate of the electric energy storage system and the capacity of the electric energy storage system is taken as the upper limit of the range of the power of the electric energy storage system; the product of the maximum charging rate of the thermal energy storage system and the capacity of the thermal energy storage system is taken as the lower limit of the range of the power of the thermal energy storage system, and the product of the maximum discharging rate of the thermal energy storage system and the capacity of the thermal energy storage system is taken as the upper limit of the range of the power of the thermal energy storage system;

[0094] The product of the upper limit and the lower limit of the state of charge and the capacity of the electric energy storage system is taken as the upper limit and the lower limit of the range of the capacity of the electric energy storage system, respectively;

[0095] The initial capacity and the final capacity of the electric energy storage system are equal;

[0096] The capacity of the electric energy storage system is the remaining part of the capacity of the electric energy storage system at the last time after self-discharge, plus the product of the charging power and the charging efficiency of the electric energy storage system per unit time minus the quotient of the discharging power and the discharging efficiency; the capacity of the thermal energy storage system at the current time is the remaining part of the capacity of the thermal energy storage system at the last time after heat loss, plus the product of the heat absorption power and the heat absorption efficiency of the thermal energy storage system per unit time minus the quotient of the heat release power and the heat release efficiency;

[0097] The equipment operation constraint includes:

[0098] The upper limit power and the lower limit power of a certain controllable unit are set;

[0099] The upper and lower limit ranges of the power difference of a certain controllable unit at adjacent times are set as the rising slope rate and the falling slope rate.

[0100] In step S3, the specific steps are:

[0101] S31: Build a data chain, including a collection layer and a preprocessing layer;

[0102] The collection layer is used to collect basic data, price signals and environmental data in real time;

[0103] The basic data includes customer ID, time period and real-time active power; the customer ID is a unique identifier of the user, which is used to distinguish different electricity consumers; the time period is used to provide a time scale for capturing the time variation of electricity consumption behavior; the real-time active power is used to reflect the power consumption of the customer in each time period;

[0104] The price signal includes the upper-layer power grid time-of-use price and demand response incentive price; the upper-layer power grid time-of-use price is used to define different price periods of peak, flat and valley; the demand response incentive price is formulated from the micro-grid;

[0105] The environmental data includes temperature and light intensity; temperature affects the electricity demand of customers; light intensity determines the output of photovoltaic power generation;

[0106] The preprocessing layer is used to repair missing values of data by using linear interpolation method, and to identify abnormal data by using quartile range algorithm to filter out noise points in the data;

[0107] S32: The random forest model is used for demand response participation state classification, the price response characteristics are used as classification characteristics, the demand elasticity coefficient matrix is used to describe demand response, so as to identify the demand response of the customer side;

[0108] The demand elasticity coefficient matrix is obtained by statistical analysis of the customer side electricity consumption under the historical time-of-use price; the power of each time period after the price adjustment of the time-of-use price is further obtained, as well as the probability density function of the adjustment power of the participating customers.

[0109] In step S4, the electricity purchase cost evaluation model is: the proportion of the difference between the actual situation and the ideal situation of a user participating in demand response to the electricity fee of the user participating in demand response in the ideal situation, subtracted from the initial satisfaction of the user;

[0110] The power supply power evaluation model is: the proportion of the difference between the actual situation and the ideal situation of a user not participating in demand response to the power demand of the user not participating in demand response in the ideal situation, subtracted from the initial satisfaction of the user;

[0111] The proportion is divided by a unit interval parameter to eliminate evaluation bias.

[0112] The embodiment uses a machine learning classifier based on a random forest model to construct an integrated classification recognition center, quantifies the time-of-use price response behavior of users through a demand elasticity coefficient matrix, accurately identifies the demand response of the micro-grid, improves the economy of the coordinated dispatching of multiple micro-grids, and realizes the function of balancing the user satisfaction and the economy of the micro-grid through coordinated power balance and minimized operation cost.

[0113] Embodiment 2

[0114] The steps of this embodiment are the same as those of Embodiment 1, except that each step is applied to actual power data of a specific example power grid. Specifically, the steps include the following steps:

[0115] S1: initialization, input information including microgrid operation data, customer side demand response data, machine learning input features and interactive collaborative scheduling data;

[0116] S2: build a master-slave management framework and economic optimization model including a master microgrid and a slave microgrid according to the input information; the master microgrid is used to generate a low-price power sales plan through optimal scheduling; the slave microgrid is used to adjust the power purchase strategy according to the plan to achieve economic collaboration of the multi-microgrid with the upper grid; the specific steps are as follows:

[0117] S21: build a master-slave management framework;

[0118] The master microgrid, as a power output party, has abundant renewable energy resources, equipped with distributed power sources such as wind power and photovoltaic power, as well as electric / thermal energy storage systems. Through optimal scheduling, it generates a power sales plan with the objective of minimizing the unit operation cost, aiming to sell the excess renewable energy at a lower price than the upper grid, realizing effective utilization of energy and maximizing profits. At the same time, the energy storage system is used to smooth the output fluctuation of renewable energy and ensure power supply reliability.

[0119] The slave microgrid, as a power input party, mainly faces industrial and residential customers. It analyzes customer electricity consumption behavior with the help of the integrated classification and identification center to identify the demand response participation labels of customers. According to these labels, for customers participating in demand response, low-price power from the slave microgrid is preferentially purchased; for customers not participating in demand response, stable power supply is supplemented through the upper grid.

[0120] The upper grid, as a system-level energy hub, provides peak-valley time-of-use price signals to guide the electricity consumption behavior of the microgrid. When there is a power imbalance between the master and slave microgrids, the upper grid provides adjustment support to ensure power balance and stable operation of the entire system.

[0121] The coordinated dispatch of master-slave microgrids follows a closed-loop process of "plan generation-cross-grid transaction-grid coordination". First, the master microgrid generates a power sale plan containing multiple time periods based on the analysis of the next day's renewable energy prediction, load curve and energy storage status, and shares it in real time to the slave microgrid through the master-slave management platform. Then, the slave microgrid matches the sale plan of the master microgrid according to the demand response label of the customers, and preferentially purchases low-price power for adjustable loads. For customers who do not participate in demand response, the power supply stability is guaranteed by the energy storage system or the upper grid. Finally, if the master microgrid is insufficient in power supply, the slave microgrid needs to purchase power from the upper grid to make up for the gap, and the upper grid will dynamically adjust the price according to the system load status. At the same time, the three parties synchronize the power data regularly to correct the prediction error and adjust the transaction plan of the subsequent time period in real time, ensuring the power balance and economic optimization between microgrids.

[0122] S22: Establish an economic optimization model;

[0123] The economic optimization model is the core link of the master-slave management method, aiming to achieve the optimal allocation of internal resources of microgrids through mathematical modeling. Its goal is to minimize the comprehensive operation cost of microgrids under the conditions of meeting power balance and equipment operation constraints. With the goal of minimizing the comprehensive cost, the mathematical expression is:

[0124]

[0125] In the formula, is the total fuel cost; is the power interaction cost; is the system maintenance cost; is the controllable unit start-stop cost; is the heat sale income; is the unit time. The total fuel cost The calculation formula is as follows:

[0126]

[0127]

[0128]

[0129] In the formula, is the gas turbine operation cost; is the fuel cell operation cost; is the unit cost of natural gas; is the unit low price of natural gas; , are the power and efficiency of the gas turbine, respectively; , are the power and efficiency of the fuel cell, respectively. The power interaction cost The calculation formula is as follows:

[0130]

[0131] In the formula, , are the electricity prices of selling and purchasing, respectively; is the interaction power between the microgrid and the upper grid. The system maintenance cost The calculation formula is as follows:

[0132]

[0133] In the formula, , are the maintenance cost and power of the customer i of the unit w , respectively. The controllable unit start-stop cost The calculation formula is as follows:

[0134]

[0135] In the formula, , are the start-stop state and the primary start cost of the controllable unit j . The thermal energy selling income The calculation formula is as follows:

[0136]

[0137] In the formula, , are the unit price of thermal energy sales and the thermal load power.

[0138] The power balance constraint, the energy storage state of charge constraint and the unit operation constraint are considered. The power balance constraint is as follows:

[0139]

[0140]

[0141]

[0142] In the formula, , are the charging and discharging power of the electric energy storage system, respectively; is the load power of the microgrid; is the electric power of the electric boiler; is the heat generated by the gas turbine; is the heat generated by the electric boiler; , are the absorption and release power of the thermal energy storage system, respectively; , respectively the power generation efficiency and the heat loss rate of the gas turbine. The energy storage state of charge constraints are as follows:

[0143]

[0144]

[0145]

[0146]

[0147] where, , are the power of the electrical energy storage system, the thermal energy storage system, respectively; , are the capacity of the electrical energy storage system, the thermal energy storage system, respectively; , are the maximum charge and discharge rates of the electrical energy storage system, respectively; , are the maximum charge and discharge rates of the thermal energy storage system, respectively; , are the upper and lower limits of the state of charge, respectively; and are the initial and final capacities of the electrical energy storage system, respectively; , are the charge and discharge power of the electrical energy storage system, respectively; , are the heat absorption and release power of the thermal energy storage system, respectively; , are the self-discharge rate of the electrical energy storage system and the heat loss rate of the thermal energy storage system, respectively; , are the charge and discharge efficiency of the electrical energy storage system, respectively; , are the heat absorption and release efficiency of the thermal energy storage system, respectively. The unit commitment constraints are as follows:

[0148]

[0149]

[0150] where, , are the upper and lower limit power of the controllable unit i , respectively; , are the ramp-up and ramp-down rates, respectively; is the power of the controllable unit i .

[0151] S3: An integrated classification and identification center is built using a machine learning classifier based on a random forest model. The user's time-of-use electricity price response behavior is quantified through the demand elasticity coefficient matrix to solve anomalies such as missing data and incorrect labels, and to achieve high-precision identification of customer-side demand response participation labels.

[0152] S31: Establish a data link;

[0153] The data chain is divided into an acquisition layer and a preprocessing layer. The acquisition layer uses smart meters to collect the following three types of key data in real time:

[0154] Basic data: Customer ID serves as a unique identifier for users, accurately distinguishing different electricity users; time periods are divided into 96 time periods with precise 15-minute intervals, providing a time scale for capturing the temporal variation patterns of electricity consumption behavior; real-time active power directly reflects the customer's electricity consumption in each time period, serving as the core basic data for analyzing electricity consumption patterns.

[0155] Price signals: The time-of-use pricing of the upper-level power grid clearly defines the three price periods: peak, flat, and valley. Peak-hour prices are higher to reflect the tight power supply and high costs, while valley-hour prices are lower to encourage users to use electricity during off-peak hours. This price difference guides users to adjust their electricity consumption behavior rationally. Demand response incentive pricing introduced from microgrids directly affects users' economic interests and becomes an important economic incentive for users to participate in demand response.

[0156] Environmental data: Temperature significantly impacts the electricity demand for residential air conditioning and heating equipment; high temperatures increase electricity consumption for air conditioning and low temperatures raise electricity consumption for heating equipment. Sunlight intensity directly determines the output of photovoltaic power generation; sufficient sunlight increases photovoltaic power output, and vice versa. These two types of environmental data are crucial for accurately predicting distributed energy output and changes in user-side electricity consumption response due to environmental factors.

[0157] Further repair and filtering are performed in the preprocessing layer. Missing values ​​are repaired using linear interpolation, which is based on the assumption of linear change in the data over time. When missing data is encountered, a linear function is constructed using known neighboring data points to extrapolate the missing values. Then, the interquartile range algorithm is used to identify outlier data, the data is sorted, and the first quartile is calculated. (Values ​​at the 25th percentile of the data) and the third quartile (Values ​​at 75% of the data), thus yielding the interquartile range. Set outlier boundaries as and Data points exceeding this boundary are identified as outliers and marked as "to be verified". These outliers can then be manually checked or further analyzed to filter out noise points in the data and improve data quality.

[0158] S32: Identify customer-side demand responses;

[0159] A random forest model was chosen for demand response participation state classification, with price response features selected as classification features. The demand elasticity coefficient, which helps describe the demand response, is defined as:

[0160]

[0161]

[0162]

[0163] In the formula, For customers m The demand elasticity coefficient; , They are time t Changes in power and time x Price changes; , The electricity consumption before and after the implementation of time-of-use pricing is respectively; , These represent the electricity prices before and after the implementation of time-of-use pricing. In the above formula, time... hour, As the self-elasticity coefficient, when the electricity price increases, the demand from end users decreases accordingly; when time... hour, These are the mutual elasticity coefficients. The demand elasticity coefficient matrix, composed of the self-elasticity coefficient and the mutual elasticity coefficient, is as follows:

[0164]

[0165] Elasticity coefficient matrix H This can be calculated through statistical analysis of customer-side electricity consumption under historical time-of-use pricing. Given... H In this case, the power consumption for each time period after the time-of-use electricity price adjustment is:

[0166]

[0167] probability density function It can be represented as:

[0168]

[0169] In the formula, To assist customers in adjusting power; , These are the mean and standard deviation of the response, respectively.

[0170] S4: Introduce electricity purchase cost assessment and power supply assessment models, dynamically adjust dispatch strategies based on actual user costs and power deviations, quantify satisfaction and balance economic efficiency and power supply reliability.

[0171] By constructing a dual-indicator evaluation system, supplemented by a dynamic closed-loop mechanism, we can achieve accurate assessment of customer satisfaction and timely strategy adjustments, thereby ensuring a positive customer electricity experience and stable system operation. The electricity purchase cost assessment is as follows:

[0172]

[0173] In the formula, For users under ideal circumstances m Electricity costs for participating in demand response; This refers to the electricity cost incurred by users participating in demand response under actual circumstances; Initial satisfaction level; The unit interval parameter is used to evaluate the deviation. The power supply is evaluated as follows:

[0174]

[0175] In the formula, For customers who do not participate in demand response under ideal conditions m The electricity demand; For customers who do not participate in demand response under actual conditions m The electricity demand; Initial satisfaction level; The unit interval parameter for evaluating bias.

[0176] To dynamically track customer satisfaction, the system calculates electricity purchase cost and power supply assessment indicators every 15 minutes (corresponding to the time interval for smart meter data collection). When the indicator value reaches a preset threshold, customer satisfaction is considered low, triggering the corresponding strategy adjustment process. Low-cost response strategy: Optimizes customer electricity costs and experience by adjusting demand response time periods. High-cost response strategy: Activates the gas turbine backup capacity of the main microgrid for situations with significant power supply gaps and high requirements for power supply reliability.

[0177] Electricity consumption information from 50 customers of a certain power grid was collected, and their participation in demand response was recorded, such as... Figure 2The customers 1-25 are set as label "1", indicating having demand response, and the customers 26-50 are set as label "0", indicating not having demand response. Assuming that the ratio of the test set is 20%, the random forest (RF), support vector machine (SVM) and gradient boosting decision tree (GBDT) are respectively used to classify and predict the customer demand response participation type, the true positive (TP), false positive (FP), true negative (TN) and false negative (FN) in the confusion matrix are used for evaluation, and the identification accuracy is described as follows:

[0178]

[0179] The machine learning performance comparison and customer satisfaction evaluation of the embodiment are listed as follows:

[0180] Table 1

[0181]

[0182] Table 1 shows the performance comparison of different machine learning methods. As can be seen from Table 1, the F1 value of the random forest method is 0.85, which is higher than that of the SVM, 0.800. In addition, the PR value of the random forest method is 1.000, which is higher than that of the SVM, 0.800. In this process, the most critical goal is to obtain the highest identification accuracy as possible. Therefore, RF is selected as the identification method.

[0183] Table 2

[0184]

[0185] Table 2 shows the customer satisfaction and information before and after the machine learning identification and demand response dynamic adjustment. As can be seen from Table 2, the customer 25 who is willing to participate in demand response and is identified as a non-participant has the largest decrease in satisfaction, and the electricity purchase cost evaluation is only 1.147. When the customer who does not participate in demand response is identified as a demand response participant, the customer 48 has a power supply evaluation of only 3.487.

[0186] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the application.

[0187] Embodiment 3

[0188] The embodiment is used to realize the principle of the above-mentioned method embodiment, and a microgrid demand response identification and optimal scheduling system considering satisfaction is constructed, which includes an initialization submodule, a modeling submodule, a demand identification submodule and an evaluation submodule;

[0189] The initialization submodule is used for initialization, and the information of the microgrid is input;

[0190] a modeling submodule configured to construct a master-slave management architecture and an economic optimization model of the micro-grid according to input information, for coordinating power balance and minimizing operation cost, respectively;

[0191] a demand identification submodule configured to construct an integrated classification identification center called by the master-slave management architecture by using a machine learning classifier based on a random forest model, to quantify user time-of-use price response behavior through a demand elasticity coefficient matrix, and to identify a customer-side demand response participation label;

[0192] an evaluation submodule configured to introduce a power purchase cost evaluation model and a power supply power evaluation model through the customer-side demand response participation label, to dynamically adjust a scheduling strategy based on actual cost and power deviation of the user, to quantify satisfaction, and to balance economy and power supply reliability.

[0193] Each submodule is mainly used to implement each step of the method embodiment, and will not be described here.

[0194] It should be noted that, according to the needs of implementation, each step / component described in the present application can be split into more steps / components, or two or more steps / components or part of the operation of the steps / components can be combined into a new step / component, to achieve the purpose of the present application.

[0195] The embodiment also includes a processor, a communication interface, a memory, and a communication bus; the processor, the communication interface, and the memory complete communication among each other through the communication bus; the memory stores a computer program, and when the program is executed by the processor, the processor executes the steps of the method for considering satisfaction of micro-grid demand response identification and optimal scheduling.

[0196] The embodiment also provides a computer readable storage medium, which stores executable instructions, and the instructions are executed by a processor to enable the processor to implement the method for considering satisfaction of micro-grid demand response identification and optimal scheduling.

[0197] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects.

[0198] Moreover, the present application can adopt the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0199] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flows or block diagram block or blocks.

[0200] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart or flows or block diagram block or blocks. Figure 1 Figure 1 The functions specified in the flowchart or flows or block diagram block or blocks can be implemented with either software or hardware, or a combination of both.

[0201] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart or flows or block diagram block or blocks. Figure 1 Figure 1 The functions specified in the flowchart or flows or block diagram block or blocks can be implemented with either software or hardware, or a combination of both.

[0202] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flows or block diagram block or blocks. Figure 1 Figure 1 The functions specified in the flowchart or flows or block diagram block or blocks can be implemented with either software or hardware, or a combination of both.

[0203] The above embodiments are only used to illustrate the design ideas and characteristics of the present application, and the purpose is to enable the skilled in the art to understand the present application and to carry it out, and the protection scope of the present application is not limited to the above embodiments. Therefore, any equivalent changes or modifications made according to the principles and design ideas disclosed by the present application are within the protection scope of the present application.​​​

Claims

1. A microgrid demand response identification and optimized scheduling method considering satisfaction, characterized in that: Includes the following steps: S1: Initialization, inputting microgrid information; S2: Construct the microgrid master-slave management architecture and economic optimization model based on the input information, which are used to coordinate power balance and minimize operating costs, respectively. S3: Utilize a machine learning classifier based on a random forest model to build an integrated classification and identification center for use by the master-slave management architecture. Quantify users' time-of-use electricity price response behavior through the demand elasticity coefficient matrix and identify customer-side demand response participation tags. The specific steps are as follows: S31: Establish a data chain, including the acquisition layer and the preprocessing layer; The acquisition layer is used to collect basic data, price signals, and environmental data in real time; The basic data includes customer ID, time period, and real-time active power; the customer ID is a unique identifier for users, used to distinguish different electricity users; the time period is at fixed intervals to provide a time scale for capturing the temporal variation patterns of electricity consumption behavior; the real-time active power is used to reflect the customer's electricity consumption in each time period; Price signals include time-of-use pricing for the upper-level grid and demand response incentive pricing; the upper-level grid time-of-use pricing is used to define different pricing periods for peak, flat, and valley periods; the demand response incentive pricing is determined by the microgrid. Environmental data includes temperature and light intensity; temperature affects customers' electricity demand; light intensity determines the output of photovoltaic power generation. The preprocessing layer is used to repair missing values ​​in the data using linear interpolation and to identify outlier data using the interquartile range algorithm to filter out noise points in the data. S32: A random forest model is used to classify the state of demand response participation, with price response features as classification features and demand elasticity coefficient matrix used to describe demand response, thereby identifying customer-side demand response. The demand elasticity coefficient matrix is ​​obtained by statistical analysis of customer-side electricity consumption under historical time-of-use pricing; further, the power of each time period after the time-of-use pricing adjustment and the probability density function of the adjusted power of participating customers are obtained. S4: By introducing a power purchase cost assessment model and a power supply assessment model through customer-side demand response participation tags, the scheduling strategy is dynamically adjusted based on the user's actual cost and power deviation, quantifying satisfaction and balancing economy and power supply reliability.

2. The microgrid demand response identification and optimized scheduling method considering satisfaction as described in claim 1, characterized in that: In step S1, the microgrid information includes microgrid operation data, customer-side demand response data, machine learning input features, and interactive collaborative scheduling data.

3. The microgrid demand response identification and optimized scheduling method considering satisfaction as described in claim 1, characterized in that: In step S2, the microgrid master-slave management architecture includes a master microgrid, a slave microgrid, and an upper-level power grid. The main microgrid is the power supplier, including distributed generation and electric / thermal energy storage systems; the main microgrid is used to generate low-cost electricity sales plans through optimal scheduling. Microgrids are the source of electricity, used to adjust power purchase strategies according to electricity sales plans, and to achieve economic synergy between multiple microgrids and the upper-level grid. The upper-level power grid is a system-level energy hub used to provide peak-valley time-of-use pricing signals, guide the electricity consumption behavior of microgrids, and coordinate dispatch to ensure power balance and stable operation of the main microgrid and slave microgrids.

4. The microgrid demand response identification and optimized scheduling method considering satisfaction as described in claim 3, characterized in that: In step S2, the coordinated scheduling of the master microgrid and the slave microgrid follows a closed-loop process of plan generation, cross-grid transactions, and grid coordination. The specific steps are as follows: S21: Based on the analysis of the next day's renewable energy forecast, load curve, and energy storage status, the master microgrid generates an electricity sales plan that includes multiple time periods and shares it with the slave microgrid in real time through the master-slave management platform; S22: Based on the customer's demand response tag, the microgrid matches the main microgrid's power sales plan and prioritizes the purchase of low-priced electricity for adjustable loads; for customers who do not participate in demand response, the power supply stability is ensured through energy storage systems or the upper-level grid. S23: If the main microgrid's power supply is insufficient, the secondary microgrid will purchase electricity from the upper-level grid to make up the gap. The upper-level grid will dynamically adjust the electricity price according to the system load status. The main microgrid, secondary microgrid, and upper-level grid will periodically synchronize power data, correct prediction errors, and adjust the trading plan for subsequent periods in real time to ensure power balance and optimal economy among the microgrids.

5. The microgrid demand response identification and optimized scheduling method considering satisfaction as described in claim 1, characterized in that: In step S2, the economic optimization model is used to optimally allocate resources within the microgrid through mathematical modeling, minimizing the overall cost of the microgrid while satisfying power balance constraints, energy storage state of charge constraints, and equipment operation constraints.

6. The microgrid demand response identification and optimized scheduling method considering satisfaction as described in claim 5, characterized in that: In step S2, the comprehensive cost is the sum of total fuel cost, power interaction cost, system maintenance cost, and controllable unit start-up and shutdown cost, minus the revenue from the sale of thermal energy. Total fuel cost includes gas turbine operating cost and fuel cell operating cost; gas turbine operating cost is the unit cost of natural gas divided by the unit price, and then multiplied by the total heat consumed in the gas turbine; fuel cell operating cost is the unit cost of natural gas divided by the unit price, and then multiplied by the total heat consumed in the fuel cell. The power interaction cost is calculated by multiplying the average of the purchase price and the selling price of electricity by the interaction power between the microgrid and the upstream grid, and by multiplying half of the difference between the purchase price and the selling price by the absolute value of the interaction power between the microgrid and the upstream grid, and then summing them up. The system maintenance cost is calculated by separately calculating the maintenance cost and power of each customer in each unit of the microgrid, and then summing them up. The start-up and shutdown cost of a controllable unit is the sum of the start-up costs when the start-up and shutdown states of the controllable unit are inconsistent in two consecutive times. Revenue from the sale of heat energy is the product of the unit price of heat energy sold and the heat load power.

7. The microgrid demand response identification and optimized scheduling method considering satisfaction as described in claim 1, characterized in that: In step S4, the electricity purchase cost assessment model is: the proportion of the electricity cost of a user participating in demand response under the ideal situation to the user's initial satisfaction minus the difference between the electricity cost of a user participating in demand response under actual and ideal conditions. The power supply assessment model is: the percentage of a user's non-demand response power demand under ideal conditions, calculated by subtracting the difference between a user's initial satisfaction and the user's actual power demand under ideal conditions. Divide the percentage by the unit interval parameter to eliminate evaluation bias.

8. A microgrid demand response identification and optimized scheduling system considering satisfaction, characterized in that: The initialization submodule is used for initialization, inputting information about the microgrid; The modeling submodule is used to construct the microgrid's master-slave management architecture and economic optimization model based on the input information, which are used to coordinate power balance and minimize operating costs, respectively. The demand identification submodule is used to build an integrated classification and identification center for use by the master-slave management architecture by using a machine learning classifier based on the random forest model. It quantifies the user's time-of-use electricity price response behavior through the demand elasticity coefficient matrix and identifies the customer-side demand response participation tags. Establish a data chain, including an acquisition layer and a preprocessing layer; The acquisition layer is used to collect basic data, price signals, and environmental data in real time; The basic data includes customer ID, time period, and real-time active power; the customer ID is a unique identifier for users, used to distinguish different electricity users; the time period is at fixed intervals to provide a time scale for capturing the temporal variation patterns of electricity consumption behavior; the real-time active power is used to reflect the customer's electricity consumption in each time period; Price signals include time-of-use pricing for the upper-level grid and demand response incentive pricing; the upper-level grid time-of-use pricing is used to define different pricing periods for peak, flat, and valley periods; the demand response incentive pricing is determined by the microgrid. Environmental data includes temperature and light intensity; temperature affects customers' electricity demand; light intensity determines the output of photovoltaic power generation. The preprocessing layer is used to repair missing values ​​in the data using linear interpolation and to identify outlier data using the interquartile range algorithm to filter out noise points in the data. A random forest model is used to classify the state of demand response participation, with price response features as classification features and demand response described by a demand elasticity coefficient matrix, thereby identifying customer-side demand response. The demand elasticity coefficient matrix is ​​obtained by statistical analysis of customer-side electricity consumption under historical time-of-use pricing; further, the power of each time period after the time-of-use pricing adjustment and the probability density function of the adjusted power of participating customers are obtained. The evaluation submodule is used to introduce the electricity purchase cost evaluation model and the power supply evaluation model through customer-side demand response participation tags. Based on the actual user cost and power deviation, the scheduling strategy is dynamically adjusted to quantify satisfaction and balance economy and power supply reliability.

Images