Dynamic peak regulation method for thermal power plant utilizing high-temperature heat energy storage

Abstract

The invention provides a dynamic peak regulation method of a thermal power plant using high-temperature heat energy storage, and belongs to the technical field of dynamic peak regulation of thermal power plants. The current heat storage capacity and the residual heat storage capacity of a heat storage tank are calculated through an intelligent peak regulation decision maker, the optimal steam extraction proportion and the heat storage release strategy are determined based on a dynamic peak regulation game model, and intermediate-stage steam of a steam turbine is extracted to the heat storage tank for heat storage according to the optimal steam extraction proportion in the power grid load valley period. The heat storage tank heat release system is started in the peak load period of the power grid to increase the power generation power to achieve power compensation, the optimal steam extraction proportion is adjusted in real time through an intelligent peak regulation decision maker, and a heat storage efficiency monitoring system is established to monitor the operation state of the heat storage tank in real time and optimize operation parameters. The technical problem that the optimal steam extraction proportion cannot be determined in real time is solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of dynamic peak regulation of thermal power plants, and in particular, relates to a dynamic peak regulation method for thermal power plants using high-temperature thermal energy storage. BACKGROUND

[0002] The basic principle of traditional thermal power plants is that coal is burned to produce flue gas, the flue gas is used to heat water to produce steam, and the steam drives the turbine impeller to generate electricity. In order to meet the peak regulation demand caused by the fluctuation of power grid load, the existing technology extracts a part of the steam to the heat storage tank for heat storage during the low load period of the power grid, and releases the heat of the heat storage tank to heat the water to produce steam to drive the turbine impeller to generate electricity during the peak load period of the power grid. The steam does not need to be purified before entering the heat storage tank, thereby saving the purification cost. In actual operation, the proportion of steam extraction during the low load period is first limited by the maximum proportion to avoid affecting the normal rotation of the turbine impeller, and secondly can be adjusted in real time according to the power grid load, for example, when the power grid load further decreases, the steam extraction proportion can be appropriately increased, and when the power grid load rises, the extraction proportion can be reduced, but it cannot exceed the maximum safe proportion. However, in the existing technology, the determination of the steam extraction proportion mainly relies on the experience judgment of the operator and the pre-set fixed operation mode, and cannot make intelligent dynamic optimization decisions according to the power grid load prediction data, the real-time temperature and pressure state of the heat storage tank, and the turbine operation parameters, resulting in that the heat storage benefit cannot be maximized under complex and variable power grid load conditions. In the case of frequent fluctuation of power grid load, the fixed extraction proportion strategy often cannot fully utilize the heat storage capacity of the heat storage tank, and also cannot guarantee the optimal operation state of the turbine under various working conditions. That is to say, the existing technology has the technical problem of being unable to determine the optimal steam extraction proportion in real time. SUMMARY

[0003] Therefore, the present application provides a dynamic peak regulation method for thermal power plants using high-temperature thermal energy storage, which can solve the technical problem of being unable to determine the optimal steam extraction proportion in real time in the existing technology.

[0004] The application is implemented as follows: the application provides a dynamic peak shaving method for a thermal power plant using high-temperature thermal energy storage, the method comprising: collecting power grid load prediction data, heat storage tank temperature and pressure state data, and steam turbine operation parameter data in real time, and calculating the current heat storage amount and the remaining heat storage capacity of the heat storage tank through an intelligent peak shaving decision maker; based on the power grid load prediction data and the heat storage tank temperature and pressure state data, determining an optimal steam extraction ratio and a heat storage release strategy through a dynamic peak shaving game model, the dynamic peak shaving game model comprising an upper model aiming to maximize the power grid peak shaving benefit and a lower model aiming to maximize the steam turbine operation safety; during a power grid load valley period, extracting the steam at the intermediate stage of the steam turbine to the heat storage tank according to the optimal steam extraction ratio, the extracted steam transferring heat to the high-temperature heat storage medium through a heat exchanger, and the condensed water returning to the steam turbine circulation system; during a power grid load peak period, starting the heat storage tank heat release system, the high-temperature heat storage medium heating the feed water through the heat exchanger to generate steam, the steam driving the steam turbine impeller to increase the power generation power, and realizing power compensation during the power grid load peak period.

[0005] The steps of the dynamic peak shaving game model are as follows: the objective function of the upper model is the product of the power grid peak period power compensation amount and the current heat storage amount of the heat storage tank, plus the reciprocal term of the operation cost, minus the logarithmic term of the power grid load demand, plus the sine function term of the optimal steam extraction ratio, and the constraint conditions include the heat storage tank capacity constraint, the power grid load balance constraint, and the steam extraction ratio range constraint; the objective function of the lower model is the product of the steam turbine operation stability index and the steam temperature, plus the reciprocal term of the steam turbine vibration power, minus the cosine function term of the blade angle, plus the exponential function term of the optimal steam extraction ratio, and the constraint conditions include the steam turbine operation parameter safety range constraint, the steam temperature and pressure constraint, and the blade stress constraint.

[0006] The coupling term of the two objective functions is the optimal steam extraction ratio term, which represents the common influence of the steam extraction ratio on the power grid peak shaving benefit and the steam turbine operation safety.

[0007] The intelligent peak shaving decision maker is a deep network architecture based on a sequence-to-sequence model, comprising an input encoding layer, a multi-layer self-attention mechanism, a feedforward network layer, and an output decoding layer, the self-attention mechanism is used to process the correlation between the power grid load prediction data and the heat storage tank temperature and pressure state data, the feedforward network layer is used to extract nonlinear feature patterns, and the output decoding layer generates the current heat storage amount, the remaining heat storage capacity, and the optimal steam extraction ratio of the heat storage tank.

[0008] The intelligent peak shaving decision maker adopts a sliding time window strategy to dynamically adjust the upper limit of the extraction ratio according to the power grid load prediction data, and the sliding time window strategy is a data processing method that uses a fixed length time window to analyze historical data and updates the window content over time.

[0009] The training data set of the intelligent peak regulation decision maker is established, specifically historical power grid load data, heat storage tank operating state data, steam turbine operating parameter data, and corresponding heat storage tank current heat storage amount, residual heat storage capacity, and optimal steam extraction ratio label data are collected, the collected data is preprocessed including data cleaning, normalization, and time sequence alignment, and the preprocessed data is segmented according to a time window to form training samples.

[0010] The intelligent peak regulation decision maker is trained, specifically the model is supervised learning trained using the training set data, a mean square error loss function and an adaptive matrix estimator optimizer are used for parameter updating, the learning rate is set to 0.001, the batch size is 32, the number of training rounds is 100, and the validation set is used for model performance evaluation and early stopping mechanism to prevent overfitting during the training process.

[0011] The memory length parameter in the intelligent peak regulation decision maker is determined according to the power grid load prediction time window length, the heat storage tank heat capacity, and the steam turbine response time, when the power grid load prediction time window length is 4 hours, the heat storage tank heat capacity is 500 MWh, and the steam turbine response time is 15 minutes, the memory length parameter is set to 16 time segments.

[0012] The extraction ratio adjustment function is specifically based on the power grid load fluctuation amplitude, the heat storage tank temperature change rate, and the steam turbine power change rate to calculate an adjustment intensity value, when the adjustment intensity value belongs to the range of 0 to 0.3, a short memory length of 8 time segments is used, when the adjustment intensity value belongs to the range of 0.3 to 0.7, a medium memory length of 16 time segments is used, and when the adjustment intensity value belongs to the range of 0.7 to 1.0, a long memory length of 32 time segments is used.

[0013] The memory length parameter is the time sequence length considered by the intelligent peak regulation decision maker when processing historical data; the adjustment intensity value is a numerical value calculated by the extraction ratio adjustment function for adjusting the memory length parameter; the power grid load fluctuation amplitude is the degree of change of the power grid load within a certain time; the heat storage tank temperature change rate is the rate of change of the temperature in the heat storage tank over time; the steam turbine power change rate is the rate of change of the steam turbine output power over time; and the time segment is the basic time unit for the intelligent peak regulation decision maker to process data.

[0014] Further, after the step of adjusting the optimal steam extraction ratio in real time by the intelligent peak regulation decision maker, a heat storage efficiency monitoring system is established to monitor the heat storage tank temperature distribution, pressure change, and heat loss in real time, and to optimize the high-temperature heat storage medium circulation flow and the heat exchanger operating parameters according to the monitoring results.

[0015] The high-temperature heat storage medium is molten salt or ceramic particle material, which is used for storing and releasing heat energy. The current heat storage amount of the heat storage tank is the total amount of heat energy stored in the heat storage tank, which is calculated by the intelligent peak shaving decision maker. The residual heat storage capacity is the heat energy capacity that the heat storage tank can continue to store, which is calculated by the intelligent peak shaving decision maker. The power grid load prediction data is prediction information of future load demand of the power grid, which is derived from the power grid dispatching system. The temperature and pressure state data of the heat storage tank is real-time monitoring data of temperature distribution and pressure change in the heat storage tank, which is derived from temperature and pressure monitoring sensors. The turbine operation parameter data includes turbine speed, power output, vibration parameter and blade angle operation state information. The optimal steam extraction ratio is the ratio of steam extraction amount to total steam amount calculated by the dynamic peak shaving game model. The heat storage release strategy is a control scheme for releasing stored heat energy of the heat storage tank during the peak period of power grid load. The upper limit of the extraction ratio is the maximum steam extraction ratio allowed under the condition of ensuring the safe operation of the steam turbine. The low valley period of the power grid load is a period of relatively low power grid load demand. The peak period of the power grid load is a period of relatively high power grid load demand. The power compensation is the increased power generation by releasing heat energy through the heat storage system.

[0016] The present application establishes a dynamic peak shaving game model and an intelligent peak shaving decision maker, combines power grid load prediction data and heat storage tank temperature and pressure state data, realizes real-time optimization decision of steam extraction ratio, and solves the problem of fixed extraction ratio in the prior art. The present application adopts a double-layer game model structure, the upper model takes maximizing the peak shaving benefit of the power grid as the target, the lower model takes maximizing the safety of the steam turbine operation as the target, the optimal steam extraction ratio is determined through coupling calculation of the two target functions, which ensures the safe operation of the steam turbine and maximizes the heat storage benefit. The sliding time window strategy of the intelligent peak shaving decision maker can dynamically adjust the upper limit of the extraction ratio according to the power grid load prediction data, realizes intelligent real-time adjustment of the steam extraction ratio, and solves the technical problem of being unable to determine the optimal steam extraction ratio in real time in the background art. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flowchart of the method of the present application.

[0018] Figure 2 The neural network structure diagram of the intelligent peak shaving decision maker involved in the present application.

[0019] Figure 3 The composition schematic diagram of the high-temperature heat storage system of the thermal power plant in embodiment 2.

[0020] Figure 4 The temperature distribution monitoring diagram in the heat storage tank in embodiment 2.

[0021] Figure 5 is a 24-hour power adjustment operation curve graph in Example 2.

[0022] Figure 6 is a molten salt temperature change process monitoring graph in Example 2. DETAILED DESCRIPTION

[0023] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0024] As Figure 1 shown is a flow chart of a dynamic peak shaving method of a thermal power plant using high-temperature heat storage energy provided by the present application, and the method comprises the following steps: S01, establishing a high-temperature heat storage system of a thermal power plant, comprising a heat storage tank, a steam extraction pipeline, a heat exchanger and a temperature and pressure monitoring sensor, wherein the heat storage tank is filled with a high-temperature heat storage medium, and the steam extraction pipeline is connected to an intermediate stage steam extraction port of a steam turbine and an inlet of the heat storage tank; S02, collecting real-time power grid load prediction data, heat storage tank temperature and pressure state data and steam turbine operation parameter data, and calculating the current heat storage amount and the remaining heat storage capacity of the heat storage tank through an intelligent peak shaving decision maker; S03, determining an optimal steam extraction ratio and a heat storage release strategy through a dynamic peak shaving game model based on the power grid load prediction data and the heat storage tank temperature and pressure state data, wherein the dynamic peak shaving game model comprises an upper model with the goal of maximizing power grid peak shaving efficiency and a lower model with the goal of maximizing steam turbine operation safety; S04, during a power grid load valley period, extracting intermediate stage steam of the steam turbine to the heat storage tank according to the optimal steam extraction ratio, and transferring heat from the extracted steam to the high-temperature heat storage medium through the heat exchanger, and returning condensed water to a steam turbine circulation system; S05, during a power grid load peak period, starting a heat storage tank heat release system, heating feed water to generate steam through the heat exchanger by the high-temperature heat storage medium, and driving a steam turbine impeller to increase power generation power by the steam, thereby realizing power compensation during the power grid load peak period; S06, adjusting the optimal steam extraction ratio in real time through the intelligent peak shaving decision maker, wherein the intelligent peak shaving decision maker adopts a sliding time window strategy, and dynamically adjusts an upper limit of the extraction ratio according to the power grid load prediction data; S07, establishing a heat storage efficiency monitoring system, monitoring heat storage tank temperature distribution, pressure change and heat loss in real time, and optimizing the circulation flow of the high-temperature heat storage medium and the operation parameters of the heat exchanger according to the monitoring results.

[0025] The existing technology is that coal-fired power plants burn coal to generate flue gas, the flue gas heats water to generate steam, the steam drives the turbine impeller to generate electricity, and part of the steam is extracted to the heat storage tank for heat storage during the low load period of the power grid. During the peak load period of the power grid, the heat storage tank releases heat to heat the water to generate steam to drive the turbine impeller to generate electricity. The steam does not need to be purified before entering the heat storage tank, thereby saving purification costs. The steam extraction ratio needs to be adjusted in real time according to the load of the power grid within the maximum proportion limit that ensures the safe operation of the turbine.

[0026] The dynamic peak regulation game model includes an upper model with the maximum peak regulation benefit of the power grid as the target and a lower model with the maximum safety of turbine operation as the target. The target function of the upper model is the product of the power compensation amount of the peak period of the power grid and the current heat storage amount of the heat storage tank plus the reciprocal term of the operation cost minus the logarithmic term of the load demand of the power grid plus the sine function term of the optimal steam extraction ratio. The power compensation amount of the peak period of the power grid is derived from the load prediction data of the power grid. The current heat storage amount of the heat storage tank is derived from the intelligent peak regulation decision maker. The operation cost is derived from the equipment operation parameters. The load demand of the power grid is derived from the load prediction data of the power grid. The optimal steam extraction ratio is used for the calculation of the lower model. The constraint conditions include the heat storage tank capacity constraint, the power grid load balance constraint, and the steam extraction ratio range constraint. The target function of the lower model is the product of the turbine operation stability index and the steam temperature plus the reciprocal term of the turbine vibration power minus the cosine function term of the blade angle plus the exponential function term of the optimal steam extraction ratio. The turbine operation stability index is derived from the turbine operation parameter data. The steam temperature is derived from the temperature and pressure monitoring sensor. The turbine vibration power is derived from the turbine operation parameter data. The blade angle is derived from the turbine operation parameter data. The optimal steam extraction ratio is used for the calculation of the upper model. The constraint conditions include the turbine operation parameter safety range constraint, the steam temperature and pressure constraint, and the blade stress constraint. The coupling term of the two target functions is the optimal steam extraction ratio term, which represents the common influence of the steam extraction ratio on the peak regulation benefit of the power grid and the safety of turbine operation.

[0027] The specific structure of the intelligent peak regulation decision maker is a deep network architecture based on a sequence-to-sequence model, which includes an input encoding layer, a multi-layer self-attention mechanism, a feedforward network layer, and an output decoding layer. The self-attention mechanism is used to process the correlation between the power grid load prediction data and the heat tank temperature and pressure state data. The feedforward network layer is used to extract nonlinear feature patterns. The output decoding layer generates the current heat storage amount, the remaining heat storage capacity, and the optimal steam extraction ratio of the heat tank. The intelligent peak regulation decision maker uses a sliding time window strategy to dynamically adjust the upper limit of the extraction ratio based on the power grid load prediction data.

[0028] In the intelligent peak regulation decision maker, the memory length parameter is determined based on the power grid load prediction time window length, the heat capacity of the heat tank, and the response time of the steam turbine. When the power grid load prediction time window length is 4 hours, the heat capacity of the heat tank is 500 MWh, and the response time of the steam turbine is 15 minutes, the memory length parameter is set to 16 time segments.

[0029] The extraction ratio adjustment function is designed to adjust the memory length parameter of the intelligent peak regulation decision maker. The extraction ratio adjustment function calculates an adjustment intensity value based on the power grid load fluctuation amplitude, the heat tank temperature change rate, and the steam turbine power change rate. When the adjustment intensity value is within the range of 0 to 0.3, a short memory length of 8 time segments is used. When the adjustment intensity value is within the range of 0.3 to 0.7, a medium memory length of 16 time segments is used. When the adjustment intensity value is within the range of 0.7 to 1.0, a long memory length of 32 time segments is used to adjust the memory length parameter of the intelligent peak regulation decision maker.

[0030] The high-temperature heat storage medium is molten salt or ceramic particle material, which is used to store and release heat energy. The current heat storage amount of the heat storage tank is the total amount of heat energy stored in the heat storage tank, which is calculated by the intelligent peak shaving decision maker. The remaining heat storage capacity is the heat energy capacity that the heat storage tank can continue to store, which is calculated by the intelligent peak shaving decision maker. The power grid load prediction data is the prediction information of future load demand of the power grid, which comes from the power grid dispatching system. The temperature and pressure state data of the heat storage tank is the real-time monitoring data of temperature distribution and pressure change in the heat storage tank, which comes from the temperature and pressure monitoring sensor. The steam turbine operating parameter data includes operating state information such as steam turbine speed, power output, vibration parameter and blade angle. The optimal steam extraction ratio is the ratio of steam extraction amount to total steam amount calculated by the dynamic peak shaving game model. The heat storage release strategy is the control scheme of releasing stored heat energy of the heat storage tank during the peak period of power grid load. The upper limit of extraction ratio is the maximum steam extraction ratio allowed under the condition of ensuring the safe operation of the steam turbine.

[0031] The sliding time window strategy is a data processing method that uses a fixed length time window to analyze historical data and updates the window content over time. The low valley period of power grid load is a period of relatively low power grid load demand. The peak period of power grid load is a period of relatively high power grid load demand. The power compensation is the increased power generation by releasing heat energy through the heat storage system. The memory length parameter is the time sequence length considered by the intelligent peak shaving decision maker when processing historical data. The adjustment intensity value is the numerical value used to adjust the memory length parameter calculated by the extraction ratio adjustment function. The power grid load fluctuation amplitude is the degree of change of power grid load within a certain time. The heat storage tank temperature change rate is the rate of change of temperature in the heat storage tank over time. The steam turbine power change rate is the rate of change of steam turbine output power over time. The time slice is the basic time unit for the intelligent peak shaving decision maker to process data.

[0032] The specific implementation of the above steps is described in detail below.

[0033] The specific implementation of step S01 is to first establish the structure of the heat storage tank, which adopts a double-layer insulation design. The inner layer is made of high-temperature resistant stainless steel material, and the outer layer is made of insulation material. The capacity of the heat storage tank is designed to be 500 MWh, the working temperature range is 550℃ to 650℃, and the working pressure range is 1.5MPa to 2.0MPa. Then install the steam extraction pipeline system, which connects the third stage extraction port of the medium-pressure cylinder of the steam turbine. The inner diameter of the pipeline is 800mm, which adopts insulation pipeline design and is equipped with a regulating valve to control the steam flow. The design pressure capacity of the extraction pipeline is 3.0MPa. Then install the heat exchanger system, which adopts a tube-shell structure with a heat exchange area of 2000 The heat exchange efficiency is greater than 85%, and the heat exchanger material is high-temperature resistant alloy steel. Finally, a temperature and pressure monitoring sensor network is deployed, with 20 temperature sensors and 10 pressure sensors arranged inside the heat storage tank. The sensor accuracy requires temperature measurement error to be less than ±2°C, pressure measurement error to be less than ±0.05MPa, and data acquisition frequency to be 1 per second. The high-temperature heat storage medium filled in the heat storage tank is molten salt material, and the molten salt composition is a mixture of 60% sodium nitrate and 40% potassium nitrate. Molten salt has good thermal stability and heat transfer performance, wide working temperature range, and heat storage density greater than 1.8MJ / kg.

[0034] The specific implementation of step S02 is to establish a data acquisition system to obtain power grid load prediction data from the power grid dispatching system through a data acquisition device. The prediction time window is 4 hours in the future, the data update frequency is every 15 minutes, and the prediction accuracy error is less than 5%. At the same time, temperature and pressure state data are collected from the heat storage tank monitoring system, including temperature values at 20 locations, pressure values at 10 locations, and molten salt flow data in the heat storage tank, with a data acquisition period of 1 second. In addition, operation parameter data are collected from the steam turbine control system, including steam turbine speed, power output, stage extraction pressure and temperature, vibration parameters, and blade angle, with a collection frequency of 10 times per second. The intelligent peak shaving decision maker uses a recurrent neural network algorithm to calculate the current heat storage amount of the heat storage tank, with input parameters including heat storage tank temperature distribution data, molten salt flow data, and heat exchanger inlet and outlet temperature difference. The current heat storage amount is calculated through a heat balance equation. The remaining heat storage capacity is calculated using a capacity estimation algorithm, which calculates the theoretical maximum heat storage amount based on the heat storage medium density, specific heat capacity, and temperature difference, and subtracts the current heat storage amount to obtain the remaining heat storage capacity. The calculation accuracy error is required to be less than 3%.

[0035] The specific implementation of step S03 is to construct a double-layer game optimization model. The upper model aims to maximize the peak shaving benefit of the power grid, and a multi-objective optimization function is established, including a power grid peak power compensation income item, a heat storage tank heat storage utilization rate item, a minimum operation cost item, and a power grid load balancing constraint item. The power grid peak power compensation amount is calculated by the difference between the peak load and the baseline load in the load prediction data, and the compensation power range is 50MW to 200MW. The current heat storage amount of the heat storage tank comes from the real-time calculation result of the intelligent peak shaving decision maker, and the heat storage amount range is 0 to 500MWh. The operation cost includes equipment depreciation cost, maintenance cost and energy consumption cost, and the cost calculation is based on the equipment operating hours and power output. The constraint condition sets the heat storage tank capacity constraint to be that the heat storage amount does not exceed 95% of the design capacity, the power grid load balancing constraint requires the deviation between the power generation and the load demand to be less than 2%, and the steam extraction ratio range constraint is 5% to 25%. The lower model aims to maximize the safety of the steam turbine operation, and a safety evaluation function is established, including a steam turbine operation stability index, a steam parameter safety index, a vibration level control index and a blade stress safety index. The steam turbine operation stability index is evaluated by the speed fluctuation rate, the power output stability and the stability of each stage pressure, and the stability index is required to be greater than 0.9. The steam temperature safety constraint is set to 540℃ to 580℃, and the steam pressure safety constraint is set to 1.2MPa to 2.2MPa. The steam turbine vibration power is controlled below 80% of the design value, and the blade stress constraint requires the stress value to be less than 70% of the allowable stress of the material. The two-layer model is coupled through the optimal steam extraction ratio, and an iterative solution algorithm is used to solve the optimal solution, and the iteration convergence condition is set to the change amount of the objective function being less than 0.001.

[0036] The specific implementation of step S04 is to start the heat storage mode during the low load period of the power grid. The determination criteria for the low load period of the power grid are that the current load is less than 75% of the daily average load and the duration is more than 30 minutes. The optimal steam extraction ratio calculated according to the game model controls the opening degree of the steam extraction valve. The steam extraction valve uses an electric regulating valve, the response time is less than 10 seconds, and the control accuracy is ±1%. The extracted medium-pressure steam enters the heat exchanger through the extraction pipeline. The steam flow is controlled within the range of 50 t / h to 200 t / h according to the extraction ratio. In the heat exchanger, the steam exchanges heat with the molten salt. After the steam transfers heat to the molten salt, it condenses into condensed water. The heat exchange process controls the steam inlet temperature to be 560°C±10°C, and the outlet temperature to be 180°C±5°C. The molten salt is heated in the heat exchanger. The inlet temperature is 550°C, and the outlet temperature reaches 620°C. The molten salt flow is controlled within the range of 300 t / h to 500 t / h. The condensed water is collected and returned to the turbine feedwater system through the backwater pipeline. The backwater temperature is 180°C, and the backwater flow is balanced with the steam extraction flow. During the heat storage process, the temperature rise rate of the heat storage tank is monitored in real time. When the temperature rise rate is 2°C / min to 5°C / min, the heat storage process is normal. When the heat storage capacity reaches 90% of the designed capacity, the extraction ratio is automatically reduced.

[0037] The specific implementation of step S05 is to start the heat release mode during the peak load period of the power grid. The determination criteria for the peak load period of the power grid are that the current load is greater than 120% of the daily average load and the predicted duration is more than 20 minutes. The heat release system of the heat storage tank is started. The high-temperature molten salt is extracted from the upper part of the heat storage tank by the molten salt circulating pump. The temperature of the molten salt is 610°C to 630°C, and the flow is controlled within the range of 200 t / h to 400 t / h. The high-temperature molten salt exchanges heat with the feedwater in the heat exchanger. The inlet temperature of the feedwater is 160°C. After heat exchange, superheated steam with a temperature of 540°C is generated. The steam pressure is 1.8 MPa to 2.0 MPa. The superheated steam enters the high-pressure cylinder or the medium-pressure cylinder of the steam turbine through the steam pipeline, drives the turbine impeller to increase the power generation power, and the power increase is calculated according to the steam flow. The typical power increase is 80 MW to 150 MW. The temperature of the molten salt decreases to 580°C after releasing heat in the heat exchanger, and returns to the lower part of the heat storage tank through the backflow pipeline, forming a molten salt circulation system. The heat release process control uses a proportional-integral-derivative control algorithm. According to the demand of the power grid load, the molten salt flow and the steam generation are dynamically adjusted. The control response time is less than 30 seconds. When the temperature of the heat storage tank decreases to 570°C or the heat storage capacity is less than 10% of the total capacity, the heat release mode is automatically stopped to ensure the safe operation of the system.

[0038] The specific implementation of step S06 is to realize real-time optimization and adjustment of the steam extraction ratio through an intelligent peak shaving decision maker. The decision maker uses a sliding time window strategy to process historical data, and the time window length is set to 4 hours and the window sliding step is 15 minutes. According to the volatility of the power grid load prediction data, the upper limit of the extraction ratio is dynamically adjusted. When the load fluctuation rate is less than 10%, the upper limit of the extraction ratio is set to 20%; when the load fluctuation rate is between 10% and 20%, the upper limit of the extraction ratio is set to 15%; and when the load fluctuation rate is greater than 20%, the upper limit of the extraction ratio is set to 10%. The intelligent peak shaving decision maker uses a reinforcement learning algorithm for online learning, and evaluates the long-term benefits of different extraction ratios through a state-action value function. The learning rate is set to 0.01 and the discount factor is set to 0.95. The decision maker inputs the current power grid load, the state of the thermal storage tank, the operating parameters of the steam turbine, and the future 4-hour load prediction, and outputs the optimal extraction ratio and thermal storage release strategy. The extraction ratio adjustment adopts a smooth transition strategy, with a single adjustment amplitude not exceeding 2% and an adjustment interval not less than 5 minutes, to avoid the impact of frequent adjustments on the operating stability of the steam turbine.

[0039] The specific implementation of step S07 is to establish a thermal storage efficiency monitoring system to monitor the temperature distribution in the thermal storage tank in real time through a distributed sensor network. Eight layers of temperature measurement points are arranged vertically in the thermal storage tank, with four radial measurement points on each layer, totaling 32 temperature monitoring points. The temperature distribution monitoring uses wireless sensor network technology, and the sensor nodes have data processing and wireless communication functions, with a data transmission frequency of once per minute. Pressure change monitoring is achieved through 10 pressure sensors arranged at the top, middle and bottom of the thermal storage tank to monitor the pressure change trend and pressure distribution uniformity in the thermal storage tank. Heat loss monitoring is achieved through heat flux density sensors to measure the heat flux density of the outer wall of the thermal storage tank, calculate the overall heat loss rate, and control the heat loss rate target to be less than 2% / day. According to the monitoring results, a fuzzy control algorithm is used to optimize the molten salt circulation flow rate. When the temperature distribution non-uniformity is greater than 5°C, the molten salt circulation flow rate is increased, and when the heat loss rate exceeds the set threshold, the thermal insulation system operating state is adjusted. The heat exchanger operating parameter optimization uses a particle swarm optimization algorithm to maximize the heat exchange efficiency, and optimizes the flow ratio of the heat exchanger tube side and shell side. The optimization variables include molten salt flow rate, feed water flow rate and heat exchanger heat transfer coefficient, and the constraint conditions include flow range limitation and temperature difference limitation.

[0040] The detailed structure of the intelligent peak shaving decision maker is based on a sequence-to-sequence deep learning architecture, which includes four main components: an input encoding layer, a multi-layer self-attention mechanism, a feedforward network layer, and an output decoding layer. The input encoding layer is responsible for processing multi-source heterogeneous data, including power grid load prediction data, heat storage tank temperature and pressure state data, and steam turbine operating parameter data. An embedding layer is used to map different types of data to a unified feature space with an embedding dimension of 256. Position encoding is generated using the sine and cosine functions to maintain temporal information, and the encoding length is consistent with the input sequence length. The multi-layer self-attention mechanism includes 6 self-attention layers, each containing 8 attention heads with a dimension of 32. The multi-head attention mechanism captures the correlation between different data sources and the temporal dependence relationship. The self-attention calculation uses a scaled dot-product attention mechanism, which extracts features through linear transformations of the query matrix, key matrix, and value matrix. A dropout rate of 0.1 is set to prevent overfitting. The feedforward network layer uses a two-layer fully connected network, with 1024 neurons in the first layer and an activation function of ReLU. The second layer has 256 neurons to extract non-linear feature patterns and enhance the model's expression ability. Layer normalization is applied to the output of each sub-layer, and residual connections are used to alleviate the gradient vanishing problem. The output decoding layer includes three independent output branches that generate the current heat storage amount, remaining heat storage capacity, and optimal steam extraction ratio of the heat storage tank, respectively. Each branch uses a fully connected layer to achieve the output, and the output layer activation functions use linear, ReLU, and Sigmoid functions, respectively.

[0041] The detailed steps of establishing the intelligent peak shaving decision maker training dataset include four stages of data collection, preprocessing, sample construction and dataset division. In the data collection stage, 12 months of continuous operation data are obtained from the power plant historical operation database, including per-minute level power grid load data, heat storage tank temperature and pressure data, steam turbine operation parameter data and corresponding heat storage label data, with a total of about 500,000 records. The power grid load data includes real-time load value, load prediction value and load change rate, the heat storage tank data includes temperature values of 32 positions, pressure values of 10 positions and molten salt flow data, the steam turbine data includes rotation speed, power, pressure and temperature of each stage and vibration parameters. The label data is obtained by combining heat balance calculation and expert annotation, the heat storage calculation is based on molten salt temperature and mass, and the extraction proportion label is based on historical optimal operation experience. In the data preprocessing stage, first, the data cleaning is performed, and the abnormal data and missing data during sensor failure are deleted, and the abnormal data judgment standard is the value exceeding 3 times the standard deviation of the normal range. Then, the data normalization processing is performed, and the minimum and maximum normalization method is adopted to scale all data to the range of 0 to 1, so as to keep the numerical range of different data sources consistent. The time sequence alignment processing ensures that the time stamps of all data sources are consistent, and the linear interpolation method is adopted to fill a small amount of missing data points. In the sample construction stage, the continuous data is divided into training samples according to the sliding window method, each sample contains 4 hours of historical data as the input sequence and the target value of the next time point as the output sequence, and the total number of samples reaches 100,000. The input sequence length is set to 240 time steps, each time step contains an 85-dimensional feature vector, and the output sequence contains 3-dimensional target variables. In the dataset division stage, the samples are divided according to the time sequence, the first 80% of the data is used as the training set for model parameter learning, the middle 10% of the data is used as the validation set for hyperparameter tuning and early stopping mechanism, and the last 10% of the data is used as the test set for model performance evaluation, ensuring that the time distribution of the test data is consistent with the actual application scenario.

[0042] The key technical ideas of the application include a dynamic peak regulation game model, an intelligent peak regulation decision maker, a sliding time window adaptive strategy, and real-time monitoring and optimization of heat storage efficiency. The dynamic peak regulation game model considers the maximization of the peak regulation benefit of the power grid and the maximization of the safety of the operation of the steam turbine as two objectives through a double-layer optimization architecture. Compared with the traditional single-objective optimization method, the model can achieve higher economic benefits under the premise of ensuring equipment safety, avoids suboptimal solutions caused by target conflicts through coupled solution of the upper and lower models, and enables the heat storage system to find a globally optimal operation strategy under complex constraints. The intelligent peak regulation decision maker uses a deep learning sequence-to-sequence architecture to process multi-source time series data. Compared with the traditional rule-based control method, the decision maker has stronger non-linear mapping ability and adaptive learning ability, can automatically mine complex correlation patterns from historical operation data, captures long-term dependencies between different data sources through a multi-head self-attention mechanism, and realizes precise prediction of the state of the heat storage system and intelligent optimization of the control strategy. The sliding time window adaptive strategy dynamically adjusts the model memory length according to the load fluctuation characteristics of the power grid. Compared with the traditional method with a fixed window length, the strategy can better adapt to the time-varying characteristics of the power grid load, uses a short memory window during the load stable period to improve the response speed, and uses a long memory window during the load severe fluctuation period to enhance the prediction stability. Through the hierarchical control of the adjustment intensity value, the strategy realizes the smooth transition of the memory length. The real-time monitoring and optimization system of heat storage efficiency realizes fine control of the heat storage process through a distributed sensor network and intelligent algorithms. Compared with the traditional extensive monitoring method, the system can timely detect problems such as a decrease in heat storage efficiency and an increase in heat loss, identifies temperature stratification in the heat storage tank through multi-point temperature monitoring, and significantly improves the overall efficiency of the heat storage system through circulation flow optimization of molten salt based on the monitoring results and adjustment of heat exchanger parameters. The synergistic effect of the four key technical ideas forms a complete intelligent heat storage peak regulation control system. The game model provides a macro optimization strategy, the intelligent decision maker realizes micro precise control, the adaptive window strategy ensures the rapid response of the system to environmental changes, and the real-time monitoring and optimization ensures the efficient and stable operation of the system. Compared with the traditional separate control method, this integrated intelligent control architecture can comprehensively improve the performance of the heat storage system and significantly enhance the peak regulation capacity of the power grid.

[0043] It should be noted that the present application also solves the following technical problems: the present application solves the technical problem that the heat storage efficiency in the traditional heat storage system cannot be monitored and optimized in real time. In the prior art, the heat storage system usually lacks accurate monitoring of the temperature distribution, pressure change and heat loss inside the heat storage tank, resulting in that the heat storage efficiency cannot be effectively controlled and optimized. The present application establishes a heat storage efficiency monitoring system to monitor the temperature distribution, pressure change and heat loss of the heat storage tank in real time, and optimizes the circulating flow of the high-temperature heat storage medium and the operating parameters of the heat exchanger according to the monitoring results, so as to realize real-time monitoring and dynamic optimization of the heat storage efficiency. The system can timely find abnormal conditions in the heat storage process, and by adjusting the circulating flow of the heat storage medium and the operating parameters of the heat exchanger, the heat storage efficiency is maximized, the heat loss is reduced, and the energy utilization efficiency of the whole system is improved. The present application also solves the technical problem that the memory length parameter in the intelligent peak regulation decision system cannot be adaptively adjusted according to the system operating state. The traditional intelligent decision system usually adopts a fixed memory length parameter, which cannot be dynamically adjusted according to the actual operating state such as power grid load fluctuation, heat storage tank temperature change and steam turbine power change, resulting in low decision accuracy. The present application designs an extraction ratio adjustment function, which calculates the adjustment intensity value based on the power grid load fluctuation amplitude, the heat storage tank temperature change rate and the steam turbine power change rate, and adopts different memory length parameters according to different adjustment intensity value ranges. When the adjustment intensity value is in the range of 0 to 0.3, a short memory length of 8 time segments is adopted; when the adjustment intensity value is in the range of 0.3 to 0.7, a medium memory length of 16 time segments is adopted; when the adjustment intensity value is in the range of 0.7 to 1.0, a long memory length of 32 time segments is adopted, so as to realize adaptive adjustment of the memory length parameter of the intelligent peak regulation decision maker, and improve the adaptability and decision accuracy of the decision system under different operating conditions.

[0044] Specifically, the principle of the present application is that the root cause of the present application capable of solving the above technical problems lies in that it establishes a complete intelligent decision system which can comprehensively consider three key factors of power grid load forecasting, heat storage tank state and steam turbine operation safety. First, the dynamic peak regulation game model realizes multi-objective optimization through a double-layer structure, the upper model considers the maximization of power grid peak regulation benefit, the objective function includes the product item of power grid peak period power compensation amount and current heat storage amount of the heat storage tank, the reciprocal item of operation cost, the logarithm item of power grid load demand and the sine function item of optimal steam extraction ratio, the lower model considers the maximization of steam turbine operation safety, the objective function includes the product item of steam turbine operation stability index and steam temperature, the reciprocal item of steam turbine vibration power, the cosine function item of blade angle and the exponential function item of optimal steam extraction ratio, the two objective functions are coupled through the optimal steam extraction ratio item to ensure that the decision result can meet the power grid peak regulation demand and guarantee the safe operation of the steam turbine. Second, the intelligent peak regulation decision maker adopts a deep network architecture based on a sequence-to-sequence model, processes the correlation between the power grid load forecasting data and the heat storage tank temperature and pressure state data through a multi-layer self-attention mechanism, extracts nonlinear feature patterns through a feedforward network layer, and generates the current heat storage amount, the remaining heat storage capacity and the optimal steam extraction ratio of the heat storage tank through an output decoding layer to realize intelligent decision of a complex multivariate system. Finally, the sliding time window strategy dynamically adjusts the upper limit of the extraction ratio according to the power grid load forecasting data, the extraction ratio adjustment function calculates the adjustment intensity value based on the power grid load fluctuation amplitude, the heat storage tank temperature change rate and the steam turbine power change rate, adopts different memory length parameters according to different adjustment intensity values to realize the adaptive adjustment capability of the decision system.

[0045] A specific embodiment 1 of the present application is provided below, and the specific implementation of each step in the embodiment 1 is described in detail as follows.

[0046] In this embodiment, the specific implementation of step S01 is the same as the foregoing, and will not be described in detail here.

[0047] The specific implementation of step S02 is to establish a data acquisition system and calculate the heat storage parameters through the intelligent peak regulation decision maker, wherein the current heat storage amount of the heat storage tank is calculated according to the following formula: ; In the formula, is the current heat storage amount of the heat storage tank, with the unit of MWh; is the molten salt density, with the value of 2100 kg / m3; ; is the molten salt volume, with the unit of m3; ; is the specific heat capacity of the molten salt, with the value of 1.5 kJ / (kg·K); is the average temperature in the heat storage tank, with the unit of K; The reference temperature is 823 K; The conversion coefficient from kJ to MWh.

[0048] wherein, The calculation by temperature sensor data is obtained: ; In the formula, The measured value of the i th temperature sensor, The total number of temperature sensors is 20.

[0049] The formula for calculating the remaining heat storage capacity is: ; In the formula, The remaining heat storage capacity is MWh; The maximum heat storage capacity of the heat storage tank is 500 MWh; The heat loss is MWh.

[0050] The specific implementation of step S03 is to construct a dynamic peak shaving game model, and the objective function of the upper model is represented as: ; In the formula, The objective function value of the upper model is; The power compensation amount of the power grid peak period is MW; The current heat storage amount of the heat storage tank is; The operation cost is yuan / h; The power grid load demand is MW; The optimal steam extraction ratio is 0.05 to 0.25; The reference power compensation amount is 100 MW by default; The reference heat storage amount is 500 MWh by default; The reference operation cost is 10000 yuan / h by default; The reference load demand is 1000 MW by default.

[0051] The power compensation amount of the power grid peak period is calculated as: ; In the formula, The predicted peak load is, The reference load is.

[0052] The operation cost calculation formula is: ; In the formula,​ For fuel cost, For maintenance cost, For power consumption cost.

[0053] The lower model objective function is represented as: ; In the formula, is the value of the lower model objective function; is the steam turbine operation stability index, with a value ranging from 0 to 1; is the steam temperature, with the unit of K; is the steam turbine vibration power, with the unit of kW; is the blade angle, with the unit of radian; is a natural constant; : reference stability index, with a default value of 0.9; is the reference steam temperature, with a default value of 833 K (corresponding to 560℃); is the reference vibration power, with a default value of 50 kW; is the reference blade angle, with a default value of 0.5 radian; is the reference extraction ratio, with a default value of 0.15.

[0054] The steam turbine operation stability index is calculated as: ; In the formula, is the rotational speed stability index, is the power stability index, is the pressure stability index, is the weight coefficient, satisfying .

[0055] The specific implementation of step S04 is to store heat during the low load period of the power grid, and the steam extraction flow calculation formula is: ; In the formula, is the extraction steam flow, with the unit of kg / s; is the total steam flow of the steam turbine, with the unit of kg / s.

[0056] The heat exchange amount of the heat exchanger is calculated as: ; In the formula, is the heat exchange amount of the heat exchanger, with the unit of kW; is the inlet steam enthalpy, with the unit of kJ / kg; is the outlet condensate enthalpy, with the unit of kJ / kg.

[0057] The specific implementation of step S05 is to release heat during the peak load period of the power grid, and the power compensation amount calculation formula is: ; In the formula, is the increased power generation, with units of MW; is the turbine efficiency, with a value of 0.35 to 0.42; is the heat power released by the thermal storage tank, with units of MW.

[0058] The heat power released by the thermal storage tank is calculated as: ; Wherein is the heat power released by the thermal storage tank, with units of kW, is the mass flow rate of the molten salt, with units of kg / s, is the enthalpy of the high-temperature molten salt, with units of kJ / kg, is the enthalpy of the low-temperature molten salt, with units of kJ / kg. The enthalpy of the high-temperature molten salt and the enthalpy of the low-temperature molten salt are calculated by the following formulas: ; ; When the specific heat capacity of the molten salt changes little within the working temperature range, the average specific heat capacity can be used for simplified calculation, and the enthalpy calculation formula is simplified as: and ; Wherein is the enthalpy at the reference state, usually taking the enthalpy at the reference temperature as 0, with units of kJ / kg, is the reference temperature, with a default value of 273.15 K or 298.15 K, is the high-temperature molten salt temperature, with a typical value of 883-903 K, i.e., 610-630°C, is the low-temperature molten salt temperature, with a typical value of 853 K, i.e., 580°C, is the average constant-pressure specific heat capacity within the working temperature range, which is about 1.5 kJ / (kg·K) for a 60% sodium nitrate and 40% potassium nitrate mixed molten salt.

[0059] Alternatively, for more accurate calculation, the temperature dependence of the specific heat capacity needs to be considered, and the temperature relationship of the molten salt specific heat capacity can be used , wherein is the fitting coefficient of the molten salt specific heat capacity, with a typical value of , , for nitrate molten salt, and the corresponding accurate enthalpy calculation formula is: .

[0060] The specific implementation of step S06 is to adjust the extraction ratio in real time through an intelligent peak-shaving decision-maker. The dynamic adjustment formula for the upper limit of the extraction ratio is as follows: ; In the formula, This represents the upper limit of the extraction ratio; The baseline sampling ratio is 0.2. This is an adjustment factor with a value of 0.5; This refers to the power grid load fluctuation rate.

[0061] The formula for calculating power grid load fluctuation rate is: ; In the formula, For the first The load value at a given moment. This is the average load value. This represents the number of data points within the time window.

[0062] The specific implementation of step S07 is to establish a thermal storage efficiency monitoring system, and the formula for calculating the heat loss rate is: ; In the formula, This refers to the heat loss rate; The outer surface area of ​​the thermal storage tank is expressed in units of... ; The heat transfer coefficient is expressed in W / ( ·K); The temperature difference between the inside and outside is expressed in Kelvin (K). The time interval is expressed in hours (h).

[0063] The formula for calculating temperature distribution non-uniformity is: ; In the formula, This represents the standard deviation of the temperature distribution, used to evaluate the uniformity of temperature distribution.

[0064] The molten salt circulation flow rate optimization uses a proportional-integral-derivative (PID) control algorithm, and the control equation is: ; In the formula, To control the output signal; This is an error signal; These are the proportional, integral, and differential coefficients, respectively.

[0065] The formula for calculating the adjustment intensity value of the extraction proportional adjustment function is as follows: ; In the formula, is the intensity value, the numerical range is 0 to 1; is the weight coefficient, satisfying ; is the heat storage tank temperature change rate, unit: K / min; is the steam turbine power change rate, unit: MW / min.

[0066] wherein the parameter acquisition method is: obtained by statistical calculation of power grid load historical data; obtained by calculating the time derivative of temperature sensor data; obtained by calculating the time derivative of steam turbine power output data; weight coefficient determined by fitting historical operation data, and a typical value is .

[0067] The formula principles and effect explanations are as follows. The current heat storage tank heat storage amount calculation formula is based on the first law of thermodynamics, and the heat storage amount is calculated through the physical property parameters and temperature change of the molten salt. The formula considers key physical property parameters such as molten salt density and specific heat capacity, provides accurate heat storage amount calculation compared with the traditional empirical estimation method, realizes real-time accurate monitoring of the heat storage system state, and provides a reliable data basis for subsequent control decisions. The double-layer objective function of the dynamic peak regulation game model adopts a polynomial combination form. The upper model reflects the synergistic effect of heat storage amount and power compensation through the product term, the reciprocal term ensures effective control of the operation cost, the logarithmic term and the sinusoidal term respectively process the nonlinear characteristics of the load demand and the periodic constraint of the extraction ratio, and the lower model strengthens the coupling relationship between stability and temperature through the product term, the reciprocal term prevents excessive vibration power, and the cosine term and the exponential term respectively process the angle characteristics of the blade angle and the exponential growth characteristics of the extraction ratio. Compared with the traditional single-objective linear optimization method, the game model can realize multi-objective coordinated optimization under complex constraint conditions, and significantly improve the economy and safety of the system. The extraction ratio upper limit dynamic adjustment formula realizes adaptive control through linear adjustment of the load fluctuation rate. When the load fluctuation is severe, the extraction ratio upper limit is automatically reduced to ensure system stability, and when the load is stable, the extraction ratio is appropriately increased to increase the heat storage effect. Compared with the traditional method of fixed ratio limitation, the dynamic adjustment strategy can optimize system operation parameters in real time according to the state of the power grid, and effectively balances the contradiction between heat storage efficiency and operation safety. The heat loss rate calculation formula is based on the basic equation of heat transfer. The heat loss power is calculated by multiplying the surface area, the heat transfer coefficient and the temperature difference, and then the loss rate is obtained by dividing the heat storage amount. The formula accurately reflects the heat preservation performance and heat loss characteristics of the heat storage tank, and provides more accurate evaluation results compared with the simplified heat loss estimation method, providing a quantitative analysis basis for heat storage system efficiency optimization. The regulation intensity value calculation formula adopts a multivariate linear weighted combination form, and comprehensively considers three key factors of power grid load fluctuation, heat storage tank temperature change and steam turbine power change. The importance of different factors is adjusted through the weight coefficient. Compared with the traditional method of single variable control, the multivariate fusion strategy can more comprehensively reflect the system operation state, realize intelligent adaptive adjustment of the memory length parameter, and significantly improve the adaptability and response accuracy of the control system to complex working conditions.

[0068] It should be noted that the variables involved in the present application are explained in detail as shown in Table 1.

[0069] Table 1 Variable explanation table

[0070] For better understanding and implementation of the present application, the following provides an embodiment 2 of a specific application scenario of the present application: a certain technical team implements a high-temperature thermal energy storage dynamic peak shaving system reconstruction on a 600 MW supercritical thermal power generating unit, the original peak shaving capacity of the unit is limited, and it is difficult to meet the rapid response requirement in the case of large grid load fluctuation. Based on the technical solution of the present application, the technical team designs and establishes a complete thermal storage peak shaving system. The high-temperature thermal energy storage system of the thermal power plant involved is as shown in Figure 3 .

[0071] In the system establishment process of step S01, the technical team first establishes a thermal storage tank structure. The thermal storage tank adopts a cylindrical design, with a diameter of 18 m, a height of 25 m, and an effective volume of 6000 . The inner layer of the thermal storage tank is made of 316L stainless steel material with a thickness of 12 mm, the outer layer is coated with 200 mm thick aluminum silicate fiber insulation material, and the outer insulation layer is provided with a stainless steel outer protective plate. The design working pressure of the thermal storage tank is 2.5 MPa, the design temperature is 680℃, and the thermal storage capacity reaches 500 MWh. The technical team fills 3600 t of binary molten salt in the thermal storage tank, the composition of the molten salt is 60% and 40% , the melting point is 220℃, and the working temperature range is 290℃ to 565℃.

[0072] The establishment of the steam extraction pipeline system includes a main extraction pipeline and an auxiliary pipeline. The main extraction pipeline is connected to the fourth stage extraction port of the intermediate pressure cylinder of the steam turbine, with a pipe diameter of 900 mm, a pipeline length of 85 m, and a design flow rate of 180 t / h. The auxiliary pipeline is connected to the fifth stage extraction port, with a pipe diameter of 600 mm and a design flow rate of 120 t / h. The extraction pipeline is entirely designed with insulation, with an insulation thickness of 150 mm, a pipeline material of P91 steel, and a design pressure bearing capacity of 3.5 MPa. Twelve electric regulating valves are installed at key positions of the pipeline, with a valve response time of 8 s and a control accuracy of ±0.5%.

[0073] The heat exchanger system includes a steam condenser and a molten salt heater. The steam condenser adopts a horizontal tube-shell structure, with a heat exchange area of 1800 , a stainless steel tube with a diameter of φ25×3 mm, and a number of 3200 tubes. The molten salt heater adopts a vertical tube-shell structure, with a heat exchange area of 2200 , and a design heat exchange efficiency of 88%. The heat exchanger shell material is 16MnR steel, the tube bundle material is TP347H stainless steel, the design pressure is 2.8 MPa, and the design temperature is 650℃.

[0074] The deployment of temperature and pressure monitoring sensor network was conducted according to the principle of hierarchical arrangement. Eight monitoring layers were arranged vertically in the thermal storage tank, and four temperature measuring points and two pressure measuring points were arranged in each layer, totaling 32 temperature sensors and 16 pressure sensors. The temperature sensor was a platinum resistance type, with a measurement range of 200-700°C and an accuracy of ±1.5°C. The pressure sensor was a diffused silicon type, with a measurement range of 0-3.0 MPa and an accuracy of ±0.03 MPa. The sensor data was collected to the central control system through wireless transmission, and the data acquisition frequency was 2 times per second.

[0075] In the data acquisition implementation of step S02, the technical team established a comprehensive data acquisition platform to obtain load prediction data from the power grid dispatching center. The prediction time span was 6 hours, and the data update interval was 10 minutes. The thermal storage tank state monitoring data included 32 temperature measuring points, 16 pressure measuring points, and molten salt flow data, with a collection period of 0.5 s. The turbine operation parameters included more than 200 monitoring parameters such as unit speed 3000 r / min, rated power 600 MW, steam extraction parameters of each stage, and blade angle. The intelligent peak shaving decision maker calculated the current heat storage capacity of the thermal storage tank in real time through a deep learning algorithm. The heat storage capacity was 385 MWh under typical operating conditions, and the remaining heat storage capacity was 115 MWh.

[0076] The dynamic peak shaving game model in step S03 was implemented using a double-layer optimization algorithm. The upper model aimed to maximize the peak shaving revenue of the power grid, and the lower model was constrained by the safe operation of the turbine. In the operation of a typical day, the power load reached a low value of 420 MW at 6 am. At this time, the game model calculated the optimal steam extraction ratio as 18%, corresponding to an extraction steam flow of 156 t / h. At 2 pm when the load reached a peak value of 580 MW, the model determined the optimal heat release strategy for the thermal storage, with a molten salt flow of 280 t / h, generating 130 t / h of superheated steam, and increasing the power generation capacity by 95 MW.

[0077] The operating parameter monitoring data shown in Table 2 reflects the performance of the system under different operating conditions.

[0078] Table 2 System operating parameter table under typical operating conditions

[0079] In the heat storage mode implementation of step S04, the system automatically starts the heat storage mode when the grid load is in the low valley period and the duration exceeds 45 minutes. The steam extraction valve adjusts the opening degree to 72% according to the calculation results of the game model, the extraction steam temperature is 545℃, the pressure is 1.85MPa, and the flow rate is stabilized at 156t / h. The extracted steam exchanges heat with the molten salt in the heat exchanger, and the steam condenses to a temperature of 185℃, and the condensed water returns to the unit through the backwater system. The temperature of the molten salt rises from 485℃ to 550℃ during the heat exchange process, and the heat storage power reaches 185MW. The heat storage process lasts for 3.5 hours, and the cumulative heat storage amount is 648MWh.

[0080] The heat release mode implementation of step S05 is carried out during the peak period of the grid load, and is started when the load exceeds the reference value of 120% and the predicted duration is greater than 30 minutes. The high-temperature molten salt is extracted from the top of the heat storage tank by the circulating pump, with a flow rate of 280t / h and a temperature of 548℃. The molten salt exchanges heat with the feed water in the heat exchanger, and the feed water temperature rises from 165℃ to 535℃, producing 130t / h of superheated steam with a pressure of 1.95MPa. The superheated steam enters the intermediate pressure cylinder of the steam turbine to drive the generator to increase the output power by 95MW. The temperature of the molten salt decreases to 465℃ after heat release, and returns to the bottom of the heat storage tank through the backflow system. The heat release process lasts for 2.8 hours, and the cumulative released heat is 784MWh.

[0081] In the intelligent peak shaving implementation of step S06, the intelligent peak shaving decision maker processes historical data with a 4-hour sliding time window, and dynamically adjusts the upper limit of the extraction ratio according to the grid load fluctuation characteristics. In the smooth period with a load fluctuation rate of 8%, the upper limit of the extraction ratio is set to 20%. When the load fluctuation rate rises to 15%, the upper limit is automatically adjusted to 16%. In the case of a load fluctuation rate of 25%, the upper limit is reduced to 12% to ensure system stability. The decision maker updates the optimization strategy every 5 minutes, and the average response time of the extraction ratio adjustment is 35s.

[0082] The heat storage efficiency monitoring data shown in Table 3 reflects the thermal performance of the system.

[0083] Table 3 Heat storage efficiency monitoring data table

[0084] In the implementation of the thermal storage efficiency monitoring system of step S07, the technical team monitors the temperature distribution in the thermal storage tank in real time through 32 temperature monitoring points and finds that the temperature stratification in the thermal storage tank is obvious, and the temperature in the upper part is 12°C higher than that in the lower part. The pressure monitoring shows that the pressure distribution in the thermal storage tank is uniform, and the maximum pressure difference is 0.08 MPa. The heat loss monitoring is realized through the heat flux density measurement of 16 measuring points on the tank wall, and the daily heat loss rate is calculated to be 1.8%, which meets the design requirements. According to the monitoring results, the system adopts a fuzzy control algorithm to optimize the molten salt circulation flow rate, and when the temperature non-uniformity exceeds 4°C, the circulation flow rate is increased by 15%, effectively improving the temperature distribution uniformity.

[0085] As shown in Figure 4 , the temperature distribution in the thermal storage tank shows obvious stratification characteristics, with higher temperature in the upper part and relatively lower temperature in the lower part, and the temperature gradient is most obvious in the middle region. As shown in Figure 5 , the power regulation curve of the system in a 24-hour operation cycle shows good peak regulation effect, effectively storing heat in the load valley period and releasing energy in time in the load peak period. As shown in Figure 6 , the variation of molten salt temperature with time reflects the dynamic characteristics of the heat storage and heat release processes, and the temperature change is smooth and controllable.

[0086] The extraction proportional regulation function shows good self-adaptive performance in actual operation, and the regulation intensity value is calculated according to the grid load fluctuation amplitude, the temperature change rate of the thermal storage tank, and the power change rate of the turbine. Under typical operating conditions, the load fluctuation amplitude is 0.12, the temperature change rate is 2.8°C / min, and the power change rate is 0.85 MW / min, and the calculated regulation intensity value is 0.42, corresponding to the control strategy using a medium memory length of 16 time segments.

[0087] The operation test results show that during the continuous 30-day operation of the thermal storage peak regulation system, the average thermal storage efficiency reaches 91.6%, the average heat release efficiency reaches 88.2%, and the system availability reaches 98.5%. The unit peak regulation response time is shortened from the traditional 45 minutes to 25 minutes, and the peak regulation amplitude is increased from 120 MW to 195 MW. Under the condition of rapid change of grid load, the system can complete the switching from heat storage mode to heat release mode within 30 minutes, effectively meeting the grid peak regulation demand.

[0088] Compared with the traditional thermal power peak shaving mode, the technical scheme of the present application brings significant technical progress. The traditional peak shaving mainly relies on changing the fuel supply and water flow to adjust the unit output, which has slow response speed and large impact on equipment, while the present application realizes the time transfer of heat energy through the physical energy storage of the heat storage medium, avoiding the thermal shock of frequent combustion adjustment on the boiler and steam turbine. The double-layer optimization architecture of the dynamic peak shaving game model takes into account the economy and safety at the same time, compared with the traditional single objective control, it can find the global optimal solution under complex constraint conditions, effectively balancing the contradiction between peak shaving benefit and equipment safety. The intelligent peak shaving decision maker has stronger environmental adaptability than the traditional rule-based control method based on the sequence prediction ability of deep learning, which can automatically learn the optimal control strategy from historical data and realize the adaptive adjustment of control parameters. The heat storage efficiency real-time monitoring system realizes the fine management and control of the heat storage process through the distributed sensor network and intelligent algorithm, compared with the traditional extensive monitoring method, it can timely find and handle the heat storage efficiency decline problem, significantly improving the overall performance and reliability of the system.

[0089] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage, used for dynamic peak-shaving using a high-temperature thermal energy storage system in the thermal power plant, wherein the high-temperature thermal energy storage system includes a thermal storage tank, a steam extraction pipeline, a heat exchanger, and temperature and pressure monitoring sensors; the thermal storage tank is filled with a high-temperature thermal storage medium; and the steam extraction pipeline connects the intermediate stage steam extraction port of the steam turbine to the inlet of the thermal storage tank; characterized in that... The method includes: real-time acquisition of grid load forecast data, thermal storage tank temperature and pressure status data, and turbine operating parameter data; and calculation of the current heat storage capacity and remaining heat storage capacity of the thermal storage tank through an intelligent peak-shaving decision-maker; based on the grid load forecast data and thermal storage tank temperature and pressure status data, determining the optimal steam extraction ratio and thermal storage release strategy through a dynamic peak-shaving game model, which includes an upper-level model aimed at maximizing grid peak-shaving benefits and a lower-level model aimed at maximizing turbine operating safety; during grid load off-peak periods, steam from the turbine intermediate stage is extracted to the thermal storage tank according to the optimal steam extraction ratio, and the extracted steam transfers heat to the high-temperature thermal storage medium through a heat exchanger, while the condensate is returned to the turbine circulation system; during grid load peak periods, the thermal storage tank thermal system is activated, and the high-temperature thermal storage medium heats the feedwater through a heat exchanger to generate steam, which drives the turbine impeller to increase power generation, thereby achieving power compensation during grid load peak periods.

2. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 1, characterized in that, The steps of the dynamic peak-shaving game model are as follows: the objective function of the upper-level model is the product of the power compensation amount during the peak period of the power grid and the current heat storage capacity of the thermal storage tank, plus the inverse term of the operating cost minus the logarithm of the power grid load demand, plus the sine function term of the optimal steam extraction ratio. The constraints include thermal storage tank capacity constraints, power grid load balance constraints, and steam extraction ratio range constraints. The objective function of the lower-level model is the product of the turbine operating stability index and the steam temperature, plus the inverse term of the turbine vibration power minus the cosine function term of the blade angle, plus the exponential function term of the optimal steam extraction ratio. The constraints include turbine operating parameter safety range constraints, steam temperature and pressure constraints, and blade stress constraints.

3. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 2, characterized in that, The coupling term of the two objective functions is the optimal steam extraction ratio term, which represents the combined impact of the steam extraction ratio on the grid peak-shaving efficiency and the turbine operation safety.

4. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 3, characterized in that, The intelligent peak-shaving decision-maker is specifically based on a deep network architecture of sequence-to-sequence model, including an input encoding layer, a multi-layer self-attention mechanism, a feedforward network layer, and an output decoding layer. The self-attention mechanism is used to process the correlation between power grid load forecast data and thermal storage tank temperature and pressure status data. The feedforward network layer is used to extract nonlinear feature patterns. The output decoding layer generates the current heat storage capacity, remaining heat storage capacity, and optimal steam extraction ratio of the thermal storage tank.

5. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 4, characterized in that, The intelligent peak-shaving decision-maker adopts a sliding time window strategy to dynamically adjust the upper limit of the extraction ratio based on the power grid load forecast data. The sliding time window strategy is a data processing method that uses a fixed-length time window to analyze historical data and updates the window content over time.

6. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 5, characterized in that, The training dataset for the intelligent peak-shaving decision-maker is established by collecting historical grid load data, thermal storage tank operating status data, turbine operating parameter data, and corresponding label data for the current heat storage capacity, remaining heat storage capacity, and optimal steam extraction ratio of the thermal storage tank. The collected data is preprocessed, including data cleaning, normalization, and time-series alignment. The preprocessed data is then segmented according to time windows to form training samples.

7. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 6, characterized in that, The training of the intelligent peak shaving decision-maker specifically involves supervised learning training of the model using training set data, employing a mean squared error loss function and an adaptive moment estimation optimizer for parameter updates, setting the learning rate to 0.001, the batch size to 32, and the number of training epochs to 100. During the training process, a validation set is used to evaluate model performance and an early stopping mechanism is implemented to prevent overfitting.

8. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 7, characterized in that, The memory length parameter in the intelligent peak shaving decision-maker is determined based on the grid load prediction time window length, the thermal capacity of the thermal storage tank, and the turbine response time. When the grid load prediction time window length is 4 hours, the thermal capacity of the thermal storage tank is 500 MWh, and the turbine response time is 15 minutes, the memory length parameter is set to 16 time segments.

9. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 8, characterized in that, The extraction proportional control function is specifically calculated based on the grid load fluctuation amplitude, the thermal storage tank temperature change rate, and the turbine power change rate to obtain the control intensity value. When the control intensity value is in the range of 0 to 0.3, a short memory length of 8 time segments is used; when the control intensity value is in the range of 0.3 to 0.7, a medium memory length of 16 time segments is used; and when the control intensity value is in the range of 0.7 to 1.0, a long memory length of 32 time segments is used.

10. The dynamic peak-shaving method for thermal power plants utilizing high-temperature thermal energy storage according to claim 9, characterized in that, The memory length parameter is the time series length considered by the intelligent peak shaving decision-maker when processing historical data; the adjustment intensity value is the value used to adjust the memory length parameter, calculated by the extraction proportional adjustment function; the grid load fluctuation amplitude is the degree of change of grid load within a certain period of time; the thermal storage tank temperature change rate is the rate of change of temperature inside the thermal storage tank over time; the turbine power change rate is the rate of change of turbine output power over time; and the time segment is the basic time unit for data processing by the intelligent peak shaving decision-maker.

Images