Electric-thermal energy storage system control method and system based on multi-time scale rolling horizon optimization and price confidence

Abstract

This invention discloses a control method and system for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence, comprising the following steps: based on the electricity spot market forecast data, outputting the expected value of electricity price forecast for each discrete time period and the upper and lower boundaries of the electricity price confidence interval through a price forecast model; establishing a coupled model of the electric thermal system including a heat source, a thermal storage device, and a heat load; constructing a conservative equivalent electricity price for optimized scheduling, and establishing an optimization problem with the goal of minimizing the total operating cost of the system; generating and executing control commands using a two-timescale hierarchical rolling time-domain optimization: solving for a benchmark heat production and thermal storage plan covering a preset daily cycle at the first time scale, updating the electricity price forecast and the actual state of the thermal storage device at the second time scale, and resolving the optimization problem within a rolling window, and executing control commands only for the current time period control step.

Description

Technical Field

[0001] This invention relates to the fields of electricity market and energy economics, and in particular to a control method and system for electric thermal energy storage systems based on multi-timescale rolling time-domain optimization and price confidence. Background Technology

[0002] Traditional single-timescale optimization is ill-suited to the complex time-varying characteristics and market price uncertainties of multi-energy flow interactions involving electricity, heat, and gas in the energy internet. This paper addresses this by employing a rolling time-domain optimization mechanism to dynamically adjust the heating power, operating status, and thermal energy storage level of the electric thermal energy storage system within each control cycle. By combining price confidence levels with a quantitative assessment of the reliability of short-term electricity market price forecasts and refining the optimization objective function, the system's flexibility and responsiveness across multiple timescales are enhanced, energy purchase costs are reduced, and renewable energy integration rates are increased. Furthermore, it ensures multi-objective synergistic optimization of economic efficiency, reliability, and environmental benefits within the price fluctuation confidence interval. Ultimately, this constructs an intelligent electric thermal energy storage control system adapted to the electricity spot market, high-proportion renewable energy penetration, and demand-side response requirements, providing crucial technical support for the low-carbon transformation of the energy system and the efficient operation of integrated energy services.

[0003] Although existing technologies recognize the use of thermal energy storage for peak shaving and valley filling, their strategies are relatively static and fail to dynamically adjust the temporal relationship between heating and heat consumption based on rolling information across multiple time scales. The system is slow to adjust when dealing with sudden intraday changes and cannot achieve continuous optimization. Therefore, a control method and system for electric thermal energy storage based on rolling time-domain optimization across multiple time scales and price confidence is proposed. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a control method and system for electric thermal energy storage systems based on multi-timescale rolling time-domain optimization and price confidence.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A control method for electric thermal energy storage systems based on multi-timescale rolling time-domain optimization and price confidence includes the following steps: Obtain electricity spot market forecast data and heat load forecast data for a predetermined future time period. Based on the electricity spot market forecast data, output the expected electricity price forecast for each discrete time period t through a price forecast model. and the upper boundary of the confidence interval for electricity prices and lower boundary ; A coupled model of an electrothermal system is established, comprising a heat source, a thermal storage device, and a heat load. The coupled model includes a thermal energy state update equation, a thermal power balance equation, and capacity and power constraints for the heat source and the thermal storage device. Construct a conservative equivalent electricity price for optimized scheduling, such that the equivalent electricity price... satisfy: An optimization problem is established with the goal of minimizing the total system operating cost, which includes the expected electricity purchase cost calculated based on the equivalent electricity price, equipment operation and maintenance cost, insufficient heating penalty cost, and the conditional value at risk (CVaR) risk penalty term for the electricity purchase cost. A two-timescale, layered, rolling time-domain optimization method is used to generate and execute control commands: At the first timescale, a baseline heat production and storage plan covering a preset daily cycle is obtained; at the second timescale, electricity price forecasts and the actual status of the heat storage device are updated, and the optimization problem is resolved within a rolling window, executing only the control commands for the current time period's control step. These control commands include the heat production power setpoint of the heat source. and / or start / stop status and the set value of the heat release mass flow rate of the heat storage device and the opening degree of the heat release valve The opening degree of the heat release valve Based on the real-time temperature of the heat storage medium Sure.

[0006] The above technical solution further includes: Furthermore, the acquisition of electricity spot market forecast data includes nationwide load forecast data, renewable energy output forecast data, and interconnection line plan data; The price prediction model employs a quantile regression forest model to generate the expected electricity price forecast at each time step. and the upper boundary of the confidence interval for electricity prices and lower boundary .

[0007] Furthermore, the coupling model includes the following mathematical constraints: ,in, For heat source heat production power, This refers to the heat release power of the thermal storage device. To provide heat charging power for the thermal storage device, The system will have a heat deficit and a corresponding penalty will be applied. Not less than 0 Given the predicted heat load, the state update equation for the thermal energy storage is: ,in, Discrete time period index The thermal energy stored in the thermal energy storage device during time period t (thermal energy state), expressed in units such as kWh or MJ. The thermal energy stored in the thermal storage device during time period t+1. For optimal heat charging efficiency, 0 < <1, The thermal power (heat power input to the thermal storage device) during time period t, in units such as kWth. For heat release efficiency, 0 < <1, This refers to the heat release power (heat output from the thermal storage device) over time period t, in units such as kWth. The coefficient for heat storage self-heat release / heat dissipation loss must satisfy... ≥0 (can be defined in relation to time step).

[0008] Furthermore, the heat load forecast data is obtained through machine learning prediction based on historical heat consumption data, weather forecast data, and user behavior patterns, outputting the heat load forecast values ​​for each future time period. And its prediction interval boundaries.

[0009] Furthermore, the specific steps for formulating the matching strategy are as follows: Construct an optimization function with the objective of minimizing the total operating cost of the electric heating system; The optimization function includes a conservative equivalent electricity price element. The calculation includes electricity purchase cost, equipment operation and maintenance cost, insufficient heating penalty, and CVaR risk penalty. The optimization process satisfies the following constraints: thermal power balance, energy state update of thermal storage device, capacity and charge / discharge power of thermal storage device, and operating power of heat source.

[0010] Furthermore, the electricity purchase cost term in the optimization function, calculated based on the conservative equivalent electricity price, is: ,in, The power purchased from the grid during time period t, and the heat generated by the heat source. With electricity purchase function The efficiency of the electro-thermal conversion is directly related to the heat source. Other costs include operation and maintenance costs and penalties for supply shortage risks; The CVaR risk penalty item is: And the random variable of electricity purchase cost in time period t Determined by the purchased power capacity and electricity price: ,in, This is a risk penalty term (a component of the objective function), with units such as yuan. For risk weighting coefficients, satisfying ≥0, For confidence level of The conditions under which the risk value is determined. Let be the confidence level parameter, satisfying 0 <a<1, Let be the random variable representing the electricity purchase cost during time period t. Let t be the power purchased by the power grid (an optimization decision variable), in units such as MW. The actual / random electricity price (uncertain amount) for time period t, in units such as yuan / MWh, and the CVaR risk penalty term adopts a confidence level. and risk weighting coefficient Configure it so that when the randomness of electricity prices increases the tail risk of electricity purchase costs, by increasing... The proportion in the objective function suppresses the power purchase decision for the corresponding time period. .

[0011] Furthermore, the decoupling time correlation of the thermal storage device is specifically manifested as follows: During lower off-peak or normal periods, based on the goal of minimizing electricity purchase costs, the system tends to increase the heat generation capacity of heat sources. The heat load that satisfies the heat power balance equation While demand is met, the surplus heat serves as Stored in a thermal storage device rise; Predicting electricity prices During higher periods, constrained by CVaR risk penalties and high electricity costs, the system tends to reduce... To reduce the amount of electricity purchased Prioritize increasing the heat release power of the thermal storage device. To achieve thermal power balance and increase the heat release power of the thermal storage device This is represented as the heat release mass flow rate setpoint. Or increase the opening of the heat dissipation valve .

[0012] Furthermore, the rolling optimization correction period for the second time scale is 15 minutes to 60 minutes; Get the latest electricity price forecast output at each correction. , , Actual thermal energy storage status of the thermal storage device and heat load forecast Within a rolling window, the control sequences for multiple future time periods are re-optimized, and only the control instructions for the current time period are executed.

[0013] A control system for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence, comprising: Source data and price confidence assessment module: Acquires electricity spot market forecast data and heat load forecast data for future periods, and outputs the expected electricity price forecast for each discrete period t from the price forecast model. and the upper boundary of the confidence interval for electricity prices and lower boundary , Conservative considerations prioritize equivalent electricity prices ; The electrothermal system coupling modeling module is used to establish a dynamic coupling model that includes a heat source, a thermal storage device, and a heat load. The dynamic coupling model includes the thermal energy state update equation and the thermal power balance constraint. The optimization matching and strategy formulation module is used to solve optimization problems with the goal of minimizing total operating costs. The objective function includes at least the electricity purchase cost term and the CVaR risk penalty term for the electricity purchase cost. ; Execution and Correction Module: Used to implement two-timescale hierarchical rolling time-domain optimization. It generates a baseline plan in the first timescale (usually 24 hours), updates the prediction in the second timescale with a rolling cycle of 15 to 60 minutes, and re-optimizes within the rolling window, executing only the current control step. It outputs the heat source's heat production power setpoint and start / stop status, as well as the heat storage device's heat release mass flow rate setpoint and heat release valve opening (based on the real-time temperature of the heat storage medium).

[0014] The present invention has the following beneficial effects: In this invention, one of the core strategies is to decouple the time relationship between heat generation and heat consumption processes using a thermal storage device. Through rolling time-domain optimization, the heat storage and release plan can be dynamically evaluated and adjusted to ensure sufficient heat storage during low-price periods and priority heat release during high-price or peak load periods. This dynamic decoupling capability makes the system more responsive and better able to adapt to the dual changes in market and demand, thereby unlocking greater cost-saving potential. Attached Figure Description

[0015] Fig. 1 The flowchart shows the control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence proposed in this invention. Fig. 2 This is a system block diagram of the control system for the electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence, as proposed in this invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figs. 1-2 As shown, this invention relates to a control method and system for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence, comprising the following steps: Obtain electricity spot market forecast data and heat load forecast data for a predetermined future time period. Based on the electricity spot market forecast data, output the expected electricity price forecast for each discrete time period t through a price forecast model. and the upper boundary of the confidence interval for electricity prices and lower boundary ; A coupled model of an electrothermal system is established, comprising a heat source, a thermal storage device, and a heat load. The coupled model includes a thermal energy state update equation, a thermal power balance equation, and capacity and power constraints for the heat source and the thermal storage device. Construct a conservative equivalent electricity price for optimized scheduling, such that the equivalent electricity price... satisfy: An optimization problem is established with the goal of minimizing the total system operating cost, which includes the expected electricity purchase cost calculated based on the equivalent electricity price, equipment operation and maintenance cost, insufficient heating penalty cost, and the conditional value at risk (CVaR) risk penalty term for the electricity purchase cost. A two-timescale, layered, rolling time-domain optimization method is employed to generate and execute control commands: At the first timescale, a baseline heat production and storage plan covering a preset daily cycle is obtained; at the second timescale, electricity price forecasts and the actual status of the heat storage device are updated, and the optimization problem is resolved within a rolling window, executing only the control commands for the current time period's control step. These control commands include the heat production power setpoint of the heat source. and / or start / stop status and the set value of the heat release mass flow rate of the heat storage device and the opening degree of the heat release valve The opening degree of the heat release valve Based on the real-time temperature of the heat storage medium Sure.

[0018] In one embodiment, the acquisition of electricity spot market forecast data includes grid-wide load forecast data, renewable energy output forecast data, and interconnection line plan data. More specifically, the grid-wide load forecast data reflects the total electricity demand of the system at various future times, denoted as... The new energy output forecast data includes the projected power generation of intermittent power sources such as wind power and photovoltaics, denoted as... The tie-line planning data reflects the planned exchange capacity across regional power grids, denoted as... The above-mentioned forecast data are aligned and integrated at the same time resolution (e.g., 15 minutes per point) to form a feature dataset describing the future net load of the system (total load minus renewable energy output and considering the impact of tie lines). The price prediction model employs a quantile regression forest model to generate the expected electricity price forecast at each time step. and the upper boundary of the confidence interval for electricity prices and lower boundary .

[0019] It should be noted that the specific analytical process for the quantile regression forest model prediction is as follows: The system automatically acquires forecast data from the publicly available information platform of the power trading center via API interface or file exchange at regular intervals (e.g., at fixed times each day) as input for the model. At the same time, the system connects to the local database to obtain historical electricity price data for the same period. Weather forecast data (such as temperature and wind speed) Including date type features (such as whether it is a weekday or a holiday), all data will be uniformly aligned to the same time resolution (such as 15 minutes per point), and missing value imputation and standardization will be performed to form the model input feature vector X(t); Model training phase: Training set construction: using historical data, where feature vectors Includes historical load, renewable energy output, interconnection plans, weather and date characteristics, with the target variable being the corresponding actual settlement price. ; Model training: Train a quantile regression forest model, which essentially constructs multiple regression trees, but in each leaf node, it stores the set of target values ​​of the samples that fall into that node, rather than a single mean; Quantile learning: For multiple preset quantile levels (For example, = 0.05, 0.5, 0.95τ (where 0.05, 0.5, 0.95 correspond to the lower quantile, median, and upper quantile respectively), the model learns how to predict the values ​​of the corresponding quantiles. This is achieved by optimizing the quantile loss function: ,in, This is the actual electricity price. These are the predicted quantile values; Online prediction and confidence interval generation: Before or during the optimization period, conduct online forecasting: Feature input: Construct feature vectors for each future time period from the acquired future time-period prediction data. ; Quantile prediction: Input a pre-trained quantile regression forest model. The model outputs a set of quantile predictions for each future time t, key components of which include: Electricity price forecast expectation : Take the median predicted value As a point estimate; Confidence interval upper and lower boundaries: Select the predicted value of the lower quantile as the lower boundary. The predicted value of the higher quantile is selected as the upper bound. This forms a confidence interval for future electricity prices. .

[0020] Interval width as a measure of uncertainty: the width of the confidence interval = - This directly reflects the uncertainty of the model's prediction of electricity prices at that future moment; the wider the range, the higher the prediction risk. Output and Application: Generated electricity price forecast expectation and the upper boundary of the confidence interval for electricity prices and lower boundary .

[0021] In one embodiment, the thermal energy state update equation is: ,in, Discrete time period index The thermal energy stored in the thermal energy storage device during time period t (thermal energy state), expressed in units such as kWh or MJ. The thermal energy stored in the thermal storage device during time period t+1. For optimal heat charging efficiency, 0 < <1, The thermal power (heat power input to the thermal storage device) during time period t, in units such as kWth. For heat release efficiency, 0 < <1, This refers to the heat release power (heat output from the thermal storage device) over time period t, in units such as kWth. The coefficient for heat storage self-heat release / heat dissipation loss must satisfy... ≥0 (can be defined in relation to time step).

[0022] In one embodiment, the heat load forecast data is obtained by machine learning from historical heat consumption data, weather forecast data, and user behavior patterns, and outputs the heat load forecast values ​​for each future time period. and its prediction interval boundary, the heat power balance equation is as follows ,in, Let be the heat output power of the heat source during time period t. This represents the heat release power of the thermal storage device during time period t. This represents the predicted heat load value for time period t. Let t be the charging power of the thermal storage device during time period t. This equation establishes the instantaneous power balance relationship between heat generation from the heat source, heat release from the thermal storage device, heat load demand, and charging power of the thermal storage device.

[0023] It should be noted that the specific analysis process for obtaining heat load forecast data is as follows: Collect and integrate multi-source data to form the basic dataset for model training. The basic dataset mainly includes: Historical heat consumption data over a long period of time, i.e., hourly or higher frequency records of heat power, supply and return water temperature and flow rate of the heating system in the past; Meteorological forecast data and measured data for corresponding historical periods, such as ambient temperature, humidity, wind speed, solar radiation intensity, and holiday information; And tag data that reflects user behavior patterns, such as work schedules, operating hours of large public buildings, and typical heating habits identified through data analysis; Based on the comprehensive training dataset, temporal feature engineering is performed to extract and construct effective features for prediction. These effective features include, but are not limited to: Time-series lag characteristics, moving averages, and periodic trend characteristics generated based on historical heat consumption data; Based on current and future forecast values ​​of meteorological forecast data, and their deviation characteristics from historical data for the same period; Date type characteristics based on calendar and user behavior patterns; The constructed feature set and the corresponding future target period heat load value are used as samples, divided into training set and validation set. Machine learning prediction algorithms, such as gradient boosting decision tree or long short-term memory neural network, are used to train the model. The hyperparameters are adjusted through the validation set to obtain the trained heat load prediction model. In practical applications, the prediction process is initiated periodically: The latest real-time historical heat consumption data, high-precision weather forecast data for the next few days, and known recent plans (such as special events that affect user behavior patterns) are obtained. This latest data is then input into the trained prediction model to perform multi-step forward prediction. The prediction process is usually carried out in a rolling manner, that is, each prediction forecasts the baseline heat load power value for each time period (such as every 15 minutes or hour) within the next 24 to 168 hours (1-7 days), thereby forming a continuous trend of heat load power changes for each time period in the future. To characterize the uncertainty of the forecast, the possible fluctuation range of the heat load is further generated, specifically: By utilizing the statistical distribution of historical prediction errors, or through ensemble learning methods (such as quantile regression), the upper and lower confidence boundaries of the baseline prediction value for each time period are calculated. The output of this fluctuation range is either the possible upper and lower limits of the heat load power for future time periods, or its probability distribution is directly output. Ultimately, the system will output complete heat load prediction data containing the baseline trend and fluctuation range.

[0024] In one embodiment, the specific steps for formulating the matching strategy are as follows: Construct an optimization function with the objective of minimizing the total operating cost of the electric heating system; The optimization function includes a conservative equivalent electricity price element. The calculation includes electricity purchase cost, equipment operation and maintenance cost, insufficient heating penalty, and CVaR risk penalty. The optimization process satisfies the following constraints: thermal power balance, energy state update of thermal storage device, capacity and charge / discharge power of thermal storage device, and operating power of heat source.

[0025] In one embodiment, the electricity purchase cost item in the optimization function, calculated based on the conservative equivalent electricity price, is: ,in, The power purchased from the grid during time period t, and the heat generated by the heat source. With electricity purchase function The efficiency of the electro-thermal conversion is directly related to the heat source. ; The CVaR risk penalty item is: And the random variable of electricity purchase cost in time period t Determined by the purchased power capacity and electricity price: ,in, This is a risk penalty term (a component of the objective function), with units such as yuan. For risk weighting coefficients, satisfying ≥0, For confidence level of The conditions under which the risk value is determined. Let be the confidence level parameter, satisfying 0 <a<1, Let be the random variable representing the electricity purchase cost during time period t. Let t be the power purchased by the power grid (an optimization decision variable), in units such as MW. The actual / random electricity price (uncertain amount) for time period t, in units such as yuan / MWh, and the CVaR risk penalty term adopts a confidence level. and risk weighting coefficient Configure it so that when the randomness of electricity prices increases the tail risk of electricity purchase costs, by increasing... The proportion in the objective function suppresses the power purchase decision for the corresponding time period. .

[0026] In one embodiment, the decoupling time correlation of the thermal storage device is specifically manifested as follows: During lower off-peak or normal periods, based on the goal of minimizing electricity purchase costs, the system tends to increase the heat generation capacity of heat sources. The heat load that satisfies the heat power balance equation While demand is met, the surplus heat serves as Stored in a thermal storage device rise; Predicting electricity prices During higher periods, constrained by CVaR risk penalties and high electricity costs, the system tends to reduce... To reduce the amount of electricity purchased Prioritize increasing the heat release power of the thermal storage device. To achieve thermal power balance and increase the heat release power of the thermal storage device This is represented as the heat release mass flow rate setpoint. Or increase the opening of the heat dissipation valve .

[0027] In one embodiment, the rolling optimization correction period for the second time scale is 15 minutes to 60 minutes; Get the latest electricity price forecast output at each correction. , , Actual thermal energy storage status of the thermal storage device and heat load forecast Within a rolling window, the control sequences for multiple future time periods are re-optimized, and only the control instructions for the current time period are executed.

[0028] A control system for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence, comprising: Source data and price confidence assessment module: Acquires electricity spot market forecast data and heat load forecast data for future periods, and outputs the expected electricity price forecast for each discrete period t from the price forecast model. and the upper boundary of the confidence interval for electricity prices and lower boundary And generate a conservative equivalent electricity price. ; The electrothermal system coupling modeling module is used to establish a dynamic coupling model that includes a heat source, a thermal storage device, and a heat load. The dynamic coupling model includes the thermal energy state update equation and the thermal power balance constraint. The optimization matching and strategy formulation module is used to solve optimization problems with the goal of minimizing total operating costs. The objective function includes at least the electricity purchase cost term and the CVaR risk penalty term for the electricity purchase cost. ; Execution and Correction Module: Used to implement two-timescale hierarchical rolling time-domain optimization, generate a baseline plan in the first time scale, update the prediction in the second time scale with a rolling cycle of 15 minutes to 60 minutes and re-optimize within the rolling window, and only execute the current control step, output the heat source heat generation power setpoint and start / stop status, as well as the heat storage device heat release mass flow rate setpoint and heat release valve opening (based on the real-time temperature of the heat storage medium).

[0029] All data obtained in this invention has been authorized by the user.

[0030] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence, characterized in that, Includes the following steps: Obtain electricity spot market forecast data and heat load forecast data for a predetermined future time period. Based on the electricity spot market forecast data, output the expected electricity price forecast for each discrete time period t through a price forecast model. and the upper boundary of the confidence interval for electricity prices and lower boundary ; A coupled model of an electrothermal system is established, comprising a heat source, a thermal storage device, and a heat load. The coupled model includes a thermal energy state update equation, a thermal power balance equation, and capacity and power constraints for the heat source and the thermal storage device. Construct a conservative equivalent electricity price for optimized scheduling, such that the equivalent electricity price... satisfy: An optimization problem is established with the goal of minimizing the total system operating cost, which includes the expected electricity purchase cost calculated based on the equivalent electricity price, equipment operation and maintenance cost, insufficient heating penalty cost, and the conditional value at risk (CVaR) risk penalty term for the electricity purchase cost. A two-timescale, layered, rolling time-domain optimization method is employed to generate and execute control commands: At the first timescale, a baseline heat production and storage plan covering a preset daily cycle is obtained; at the second timescale, electricity price forecasts and the actual status of the heat storage device are updated, and the optimization problem is resolved within a rolling window, executing only the control commands for the current time period's control step. These control commands include the heat production power setpoint of the heat source. and start / stop status and the set value of the heat release mass flow rate of the heat storage device and the opening degree of the heat dissipation valve The opening degree of the heat release valve Based on the real-time temperature of the heat storage medium Sure.

2. The control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree as described in claim 1, characterized in that, The electricity spot market forecast data obtained includes nationwide grid load forecast data, new energy output forecast data, and tie-line plan data. The price prediction model employs a quantile regression forest model to generate the expected value of the electricity price prediction at each time step. and the upper boundary of the confidence interval for electricity prices and lower boundary .

3. The control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree as described in claim 1, characterized in that, The coupling model includes the following mathematical constraints: ,in, For heat source heat production power, This refers to the heat release power of the thermal storage device. To provide heat charging power for the thermal storage device, The system will have a heat deficit and a corresponding penalty will be applied. Not less than 0 Given the predicted heat load, the state update equation for the thermal energy storage is: ,in, Discrete time period index The thermal energy stored in the thermal storage device during time period t. The thermal energy stored in the thermal storage device during time period t+1. For optimal heat charging efficiency, 0 < <1, The charging heat power during time period t, For heat release efficiency, 0 < <1, Let be the heat release power during time period t. The coefficient for heat storage self-heat release / heat dissipation loss must satisfy... ≥0.

4. The control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree according to claim 1, characterized in that, The heat load forecast data is obtained through machine learning from historical heat consumption data, weather forecast data, and user behavior patterns, and outputs the heat load forecast values ​​for future time periods. And its prediction interval boundaries.

5. The control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree according to claim 1, characterized in that, The specific steps for formulating the matching strategy are as follows: Construct an optimization function with the objective of minimizing the total operating cost of the electric heating system; The optimization function includes a conservative equivalent electricity price element. The calculation includes electricity purchase cost, equipment operation and maintenance cost, insufficient heating penalty, and CVaR risk penalty. The optimization process satisfies the following constraints: thermal power balance, energy state update of thermal storage device, capacity and charge / discharge power of thermal storage device, and operating power of heat source.

6. The control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree according to claim 1, characterized in that, The electricity purchase cost item in the optimization function, calculated based on the conservative equivalent electricity price, is: ,in, The power purchased from the grid during time period t, and the heat generated by the heat source. With electricity purchase function The efficiency of the electro-thermal conversion is directly related to the heat source. Other costs include operation and maintenance costs and penalties for supply shortage risks; The CVaR risk penalty item is: And the random variable of electricity purchase cost in time period t Determined by the purchased power capacity and electricity price: ,in, As a risk penalty item, For risk weighting coefficients, satisfying ≥0, For confidence level of The conditions under which the risk value is determined. Let be the confidence level parameter, satisfying 0 <a<1, Let be the random variable representing the electricity purchase cost over time period t. The power purchased by the power grid during time period t. The actual / random electricity price for time period t, and the CVaR risk penalty term using a confidence level. and risk weighting coefficient Configure it so that when the randomness of electricity prices increases the tail risk of electricity purchase costs, by increasing... The proportion in the objective function suppresses the power purchase decision for the corresponding time period. .

7. The control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree according to claim 1, characterized in that, The specific manifestation of utilizing the thermal storage device to decouple time correlation is as follows: During lower off-peak or normal periods, based on the goal of minimizing electricity purchase costs, the system tends to increase the heat generation capacity of heat sources. The heat load that satisfies the heat power balance equation While demand is met, the surplus heat serves as Stored in a thermal storage device rise; Predicting electricity prices During higher periods, constrained by CVaR risk penalties and high electricity costs, the system tends to reduce... To reduce the amount of electricity purchased Prioritize increasing the heat release power of the thermal storage device. To achieve thermal power balance and increase the heat release power of the thermal storage device This is represented as the heat release mass flow rate setpoint. Or increase the opening of the heat dissipation valve .

8. The control method for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree according to claim 1, characterized in that, The rolling optimization correction cycle for the second time scale is 15 to 60 minutes; Get the latest electricity price forecast output at each correction. , , Actual thermal energy storage status of the thermal storage device and heat load forecast Within a rolling window, the control sequences for multiple future time periods are re-optimized, and only the control instructions for the current time period are executed.

9. A control system for an electric thermal energy storage system based on multi-timescale rolling time-domain optimization and price confidence degree for implementing the method of claim 1, characterized in that, include: Source data and price confidence assessment module: Acquires electricity spot market forecast data and heat load forecast data for future periods, and outputs the expected electricity price forecast for each discrete period t from the price forecast model. and the upper boundary of the confidence interval for electricity prices and lower boundary And generate a conservative equivalent electricity price. ; The electrothermal system coupling modeling module is used to establish a dynamic coupling model that includes a heat source, a thermal storage device, and a heat load. The dynamic coupling model includes the thermal energy state update equation and the thermal power balance constraint. The optimization matching and strategy formulation module is used to solve optimization problems with the goal of minimizing total operating costs. The objective function includes at least the electricity purchase cost term and the CVaR risk penalty term for the electricity purchase cost. ; Execution and Correction Module: Used to implement two-timescale hierarchical rolling time-domain optimization, generate a baseline plan in the first time scale, update the prediction in the second time scale with a rolling cycle of 15 minutes to 60 minutes and re-optimize within the rolling window, and only execute the current control step, output the heat source heat production power setpoint and start / stop status, as well as the heat storage device heat release mass flow rate setpoint and heat release valve opening.

Images