Steam extraction and energy storage adjusting and optimizing method and system utilizing high-temperature fused salt heat storage system

Abstract

The invention provides a steam extraction energy storage adjustment optimization method and system using a high-temperature fused salt heat storage system, and relates to the technical field of energy storage, and the method comprises the steps: building a thermal coupling mathematical model of the high-temperature fused salt heat storage system and a turboset, determining key parameters affecting the steam extraction energy storage performance, and taking the power generation coal consumption rate as an optimization target, a dynamic optimization adjustment model of steam extraction energy storage is established, an improved non-dominated sorting genetic algorithm is applied to solve the established dynamic optimization adjustment model, an optimal solution set of steam extraction energy storage operation parameters under different working conditions is obtained, and a dynamic optimization adjustment strategy library of steam extraction energy storage is formed; and according to a power grid load prediction result, selecting an optimal adjustment strategy from the obtained dynamic optimization adjustment strategy library, cooperatively controlling the steam extraction process of the steam turbine and the fused salt heat storage process, and realizing flexible operation and efficient peak regulation of the thermal generator set under the condition of dynamic fluctuation of the power grid load.

Description

Technical Field

[0001] The present invention relates to energy storage technology, and in particular to a method and system for optimizing regulation of steam extraction energy storage using a high-temperature molten salt heat storage system. Background Art

[0002] As the proportion of renewable energy in the power system continues to increase, the peak-to-valley difference and load fluctuation of the power grid also increase, which puts higher requirements on the regulation capacity of traditional thermal power generating units. In order to improve the stability and reliability of the power grid and at the same time improve the economy and flexibility of thermal power generating units, the steam extraction storage regulation technology of the high-temperature molten salt heat storage system came into being.

[0003] Traditional thermal power generating units face the following problems during operation:

[0004] Limited peak-shaving capacity: The output regulation range of thermal power generating units is narrow, making it difficult to adapt to large fluctuations in grid load. Frequent start-stopping increases equipment wear: In order to cope with load changes, the units need to start and stop frequently, resulting in increased equipment wear and shortened life. Low energy efficiency: When operating at low load, the efficiency of the units is significantly reduced, resulting in energy waste. Poor economy: Frequent adjustment and non-rated operating conditions will lead to increased coal consumption and reduce the economy of the units. High environmental pressure: Load fluctuations lead to unstable combustion and increased pollutant emissions.

[0005] In order to solve the above problems, researchers proposed a solution that combines a high-temperature molten salt heat storage system with a thermal power generator. The basic principle of this solution is to extract part of the steam from the steam turbine and use it to heat the molten salt when the grid load is low, and store the heat energy; when the load is high, the stored heat energy is used to regenerate steam and drive the steam turbine to generate electricity. This method can improve the peak-shaving capacity of the unit, reduce the number of starts and stops, and improve energy efficiency.

[0006] However, the existing steam extraction storage regulation technology still has the following shortcomings:

[0007] Insufficient optimization of extraction parameters: The selection of extraction pressure, temperature and flow rate often relies on experience and lacks scientific optimization methods. Poor control of heat storage process: The control strategy of molten salt temperature and flow rate is not precise enough, which affects the heat storage efficiency. Poor coordination with the unit: The coordinated control between the extraction system and the main unit is not perfect, which may affect the safe and stable operation of the unit. Insufficient consideration of economic efficiency: The existing regulation strategy fails to fully consider economic factors such as changes in electricity prices and equipment wear. Limited prediction ability: The lack of accurate prediction of grid load and renewable energy output affects the formulation of regulation strategies. Insufficient real-time optimization capabilities: It is difficult for the existing system to dynamically adjust strategies according to real-time operating conditions and market conditions.

[0008] Therefore, it is urgent to develop an optimization method that can comprehensively consider multiple factors and realize coordinated control of steam extraction parameters, heat storage process and units, so as to give full play to the potential of high-temperature molten salt heat storage system and improve the regulation capacity and economic benefits of thermal power generating units. Summary of the invention

[0009] The embodiments of the present invention provide a method and system for optimizing regulation of steam extraction energy storage using a high-temperature molten salt heat storage system, which can solve the problems in the prior art.

[0010] According to a first aspect of the embodiments of the present invention,

[0011] Provides an extraction steam storage regulation optimization method using a high-temperature molten salt heat storage system, including:

[0012] Establish a thermal coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit, the thermal coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system; determine the key parameters that affect the extraction steam energy storage performance, the key parameters include the high-pressure cylinder extraction steam pressure, extraction steam temperature, extraction steam flow rate, high-temperature molten salt heat storage temperature, molten salt flow rate, and reheater outlet temperature;

[0013] Taking the power generation coal consumption rate as the optimization target, a dynamic optimization regulation model of extraction steam energy storage is established. The objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate. The constraints include the upper and lower limits of the extraction steam flow rate, the molten salt flow rate, the extraction steam temperature, the molten salt heat storage temperature and the reheater outlet temperature. The established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of the extraction steam energy storage operating parameters under different working conditions, thus forming a dynamic optimization regulation strategy library for extraction steam energy storage.

[0014] According to the grid load forecast results, the optimal regulation strategy is selected from the obtained dynamic optimization regulation strategy library, and the set values ​​of the high-pressure cylinder extraction pressure, extraction temperature and extraction flow are transmitted to the turbine extraction control system, and the set values ​​of the molten salt heat storage temperature and molten salt flow are transmitted to the high-temperature molten salt heat storage control system; the steam turbine extraction process and the molten salt heat storage process are coordinated to realize the flexible operation and efficient peak regulation of the thermal power generating units under the dynamic fluctuation of the grid load.

[0015] In an optional embodiment,

[0016] A thermal coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit is established. The thermal coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system, including:

[0017] Establishing a thermodynamic model of a steam turbine high-pressure cylinder, the thermodynamic model comprising a steam flow calculation formula based on the Stodola ellipse law and an isentropic efficiency correction formula considering partial load performance changes; calculating the steam flow and the corrected isentropic efficiency of the steam turbine high-pressure cylinder according to the thermodynamic model of the steam turbine high-pressure cylinder;

[0018] Establishing a heat transfer flow model of a high-temperature molten salt heat storage device, the heat transfer flow model of the high-temperature molten salt heat storage device includes a heat transfer equation of a heat exchanger based on a logarithmic mean temperature difference method and a modified Dittus-Boelter formula considering the flow characteristics of molten salt; using the heat transfer flow model of the high-temperature molten salt heat storage device, calculating the heat transfer power of the high-temperature molten salt heat storage device and the Nusselt number of the molten salt flow;

[0019] Establishing a mass energy balance equation for the steam turbine extraction and heat storage system, the mass energy balance equation includes a mass balance equation for the extraction point, an energy balance equation, and an energy accumulation rate equation for the high-temperature molten salt heat storage system; based on the mass energy balance equation, calculating the steam flow rate and specific enthalpy at each point in the system and the energy change rate of the high-temperature molten salt heat storage system;

[0020] The thermodynamic model of the steam turbine high-pressure cylinder, the heat transfer flow model of the high-temperature molten salt heat storage device and the mass energy balance equation of the steam turbine extraction and heat storage system are combined to construct a complete thermomechanical coupling mathematical model.

[0021] In an optional embodiment,

[0022] Taking the power generation coal consumption rate as the optimization target, a dynamic optimization regulation model of extraction steam energy storage is established. The objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate. The constraints include the upper and lower limits of the extraction steam flow rate, molten salt flow rate, extraction steam temperature, molten salt heat storage temperature and reheater outlet temperature. The established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of extraction steam energy storage operating parameters under different working conditions. The dynamic optimization regulation strategy library of extraction steam energy storage includes:

[0023] Initializing a population to generate an initial population including decision variables, wherein the decision variables include a high-pressure cylinder extraction steam pressure, an extraction steam temperature, an extraction steam flow rate, a high-temperature molten salt heat storage temperature, a molten salt flow rate, and a reheater outlet temperature;

[0024] The initial population is non-dominatedly sorted to divide the individuals into different non-dominated layers; uniformly distributed reference points are generated in the standardized target space; the distance from each individual to the nearest reference point is calculated, and individuals are selected to enter the next generation population according to the distance; the population diversity index is calculated, which is the average value of the Euclidean distances of all individuals to the population center;

[0025] According to the population diversity index, adaptively adjusting the crossover rate and the mutation rate, wherein the crossover rate decreases as the population diversity index increases, and the mutation rate increases as the population diversity index increases, and the adjustment of the crossover rate and the mutation rate both adopts an exponential function form; performing crossover and mutation operations on the selected individuals to generate a progeny population;

[0026] Select some individuals with a preset probability for simulated annealing optimization, generate new solutions in the neighborhood of the selected individuals, calculate the energy difference, calculate the probability of accepting the new solution based on the energy difference and the current temperature, and update the temperature according to the preset cooling coefficient;

[0027] Calculate the improved crowding, including calculating the target space crowding and the decision space crowding, and obtain the comprehensive crowding by weighted average, wherein the target space crowding and the decision space crowding consider the distances of adjacent individuals in the target space and the decision space respectively;

[0028] The population is divided into multiple sub-populations, with the number of sub-populations equal to the number of available processor cores. For each sub-population, objective function evaluation, constraint processing, non-dominated sorting, and crowding calculation are performed in parallel on a separate processor core; the evaluation results of each sub-population are summarized, and global non-dominated sorting and selection operations are performed; the parent population and the child population are merged, and the merged population is non-dominated sorted, and the top N individuals are selected to form the next generation population;

[0029] Determine whether the maximum number of iterations has been reached or the convergence condition has been met. If so, output the optimization result. If not, return to the step of calculating the distance from each individual to the nearest reference point. According to the output optimization result, determine the optimal operating parameters of the extraction steam storage system to achieve dynamic optimization control of the system.

[0030] In an optional embodiment,

[0031] According to the grid load forecast results, the optimal regulation strategy is selected from the obtained dynamic optimization regulation strategy library, and the set values ​​of the high-pressure cylinder extraction steam pressure, extraction steam temperature, and extraction steam flow are transmitted to the turbine extraction steam control system, and the set values ​​of the molten salt heat storage temperature and molten salt flow are transmitted to the high-temperature molten salt heat storage control system, including:

[0032] A deep learning model of long short-term memory network and attention mechanism is used to predict power grid load, wherein the deep learning model includes an input layer, a long short-term memory network layer, an attention layer, a fully connected layer and an output layer, wherein the long short-term memory network layer is provided with 3 layers, each layer includes 128 neurons, and the fully connected layer includes 2 layers, respectively including 64 and 32 neurons;

[0033] A dynamic optimization regulation strategy library is constructed using a dual deep Q network algorithm, the dual deep Q network algorithm includes an action value network and a target network, the action value network and the target network are both composed of three fully connected layers, the number of neurons is 256, 128 and 64 respectively, and the activation function is a ReLU function; the reward function of the dynamic optimization regulation strategy library takes into account the improvement of power generation efficiency, the improvement of peak load capacity, the equipment wear cost and the operating cost;

[0034] Based on the grid load forecasting results, a fuzzy comprehensive evaluation method is used to select the optimal regulation strategy from the dynamic optimization regulation strategy library, wherein the fuzzy comprehensive evaluation method includes establishing a fuzzy relationship matrix and determining the index weights, wherein the index weights are determined by a hierarchical analysis method;

[0035] The high-pressure cylinder extraction pressure set value, extraction temperature set value, and extraction flow set value in the optimal regulation strategy are transmitted to the turbine extraction control system through the OPC UA protocol, and the molten salt heat storage temperature set value and the molten salt flow set value are transmitted to the high-temperature molten salt heat storage control system through the OPC UA protocol. The OPC UA protocol uses an X.509 certificate for identity authentication and uses the AES-256 encryption algorithm to encrypt communication data;

[0036] In an optional embodiment,

[0037] The coordinated control of the steam turbine extraction process and the molten salt heat storage process enables flexible operation and efficient peak load regulation of thermal power generating units under dynamic fluctuations of power grid load, including:

[0038] A model predictive control method is used to realize the coordinated control of the steam turbine extraction process and the molten salt heat storage process. The objective function of the model predictive control method includes an output tracking error term and a control increment penalty term, and the constraints include an extraction pressure range, an extraction temperature range, an extraction flow range, a molten salt temperature range, a molten salt flow range, and a control increment limit.

[0039] An improved fast gradient method is used to solve the model predictive control optimization problem. The improved fast gradient method includes converting the original problem into a quadratic programming problem, using the Nesterov accelerated gradient method for iterative solution, and adopting a hot start strategy. A recursive least squares algorithm is introduced to update the system model parameters online. The forgetting factor of the recursive least squares algorithm is set to 0.98 to achieve adaptive update of the system model.

[0040] In an optional embodiment,

[0041] The improved fast gradient method is used to solve the model predictive control optimization problem. The improved fast gradient method includes converting the original problem into a quadratic programming problem, using the Nesterov accelerated gradient method for iterative solution, and adopting a hot start strategy including:

[0042] The Nesterov accelerated gradient method includes the following sub-steps:

[0043] Initialize the decision variable vector, auxiliary variable vector, step size and momentum parameter; calculate the gradient of the objective function; update the decision variable vector and auxiliary variable vector; project the updated decision variable vector into the feasible domain; dynamically adjust the step size using the Barzilai-Borwein method; dynamically adjust the momentum parameter according to the number of iterations;

[0044] Implement the hot start strategy and use the optimal solution of the previous control cycle as the initial value of the current control cycle, including:

[0045] Save the optimal solution of the previous control cycle; construct a new initial decision variable vector, where the first N-1 control increments come from the optimal solution of the previous cycle and the last control increment is set to zero; use a larger step size for warm-up iterations to quickly approach the optimal solution.

[0046] According to a second aspect of the embodiments of the present invention,

[0047] Provide an extraction steam storage regulation optimization system using a high-temperature molten salt heat storage system, including:

[0048] The first unit is used to establish a thermodynamic coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit, the thermodynamic coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system; determine the key parameters that affect the extraction steam energy storage performance, the key parameters include the high-pressure cylinder extraction steam pressure, extraction steam temperature, extraction steam flow rate, high-temperature molten salt heat storage temperature, molten salt flow rate, and reheater outlet temperature;

[0049] The second unit is used to establish a dynamic optimization regulation model of extraction steam energy storage with the power generation coal consumption rate as the optimization target, wherein the objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate, and the constraints include the upper and lower limits of the extraction steam flow rate, the molten salt flow rate, the extraction steam temperature, the molten salt heat storage temperature and the reheater outlet temperature; the established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of the extraction steam energy storage operating parameters under different working conditions, thereby forming a dynamic optimization regulation strategy library of extraction steam energy storage;

[0050] The third unit is used to select the optimal regulation strategy from the obtained dynamic optimization regulation strategy library according to the grid load forecast result, transmit the set values ​​of high-pressure cylinder extraction pressure, extraction temperature and extraction flow to the turbine extraction control system, and transmit the set values ​​of molten salt heat storage temperature and molten salt flow to the high-temperature molten salt heat storage control system; coordinately control the turbine extraction process and the molten salt heat storage process to realize flexible operation and efficient peak regulation of thermal power generating units under dynamic fluctuations of grid load.

[0051] According to a third aspect of the embodiments of the present invention,

[0052] An electronic device is provided, comprising:

[0053] processor;

[0054] a memory for storing processor-executable instructions;

[0055] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0056] A fourth aspect of the embodiments of the present invention is:

[0057] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0058] By building a dynamic optimization regulation strategy library, the system can quickly select the most appropriate operating parameters based on the current operating conditions and predicted future load changes. This predictive control strategy can respond to load changes in advance and further improve the regulation capability and economy of the system. In short, this dynamic optimization regulation method of steam extraction energy storage based on the improved non-dominated sorting genetic algorithm provides strong technical support for the efficient operation of high-temperature molten salt heat storage systems, helps to improve the flexibility and economy of thermal power generating units, and contributes to the stable operation of the power grid and the large-scale access of renewable energy.

[0059] The steam extraction energy storage system can dynamically adjust operating parameters according to the grid load forecast results, maximizing power generation efficiency and peak load regulation capacity while ensuring safe and stable operation of the system. For example, when a load trough is predicted, the system may increase the steam extraction volume and increase the molten salt heat storage temperature to prepare for subsequent peak load regulation. During peak load periods, the system will reduce the steam extraction volume, increase the molten salt heat release, and improve the power generation output of the unit. This intelligent control strategy not only improves the flexibility and economy of thermal power generating units, but also provides strong support for the stable operation of the power grid and the large-scale access of renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1A schematic diagram of a flow chart of a method for optimizing steam extraction energy storage regulation using a high-temperature molten salt heat storage system according to an embodiment of the present invention;

[0061] Figure 2 It is a structural schematic diagram of the steam extraction energy storage regulation and optimization system using a high-temperature molten salt heat storage system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0064] Figure 1 FIG. 1 is a flow chart of an optimization method for regulating steam extraction energy storage using a high-temperature molten salt heat storage system according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0065] S101. Establish a thermodynamic coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit, the thermodynamic coupling mathematical model includes thermodynamic models of the high-pressure cylinder, the medium-pressure cylinder and the low-pressure cylinder of the steam turbine, a heat transfer flow model of the high-temperature molten salt heat storage device, and a mass energy balance equation of the steam turbine extraction and heat storage system; determine key parameters that affect the extraction steam energy storage performance, the key parameters include the high-pressure cylinder extraction steam pressure, extraction steam temperature, extraction steam flow rate, high-temperature molten salt heat storage temperature, molten salt flow rate, and reheater outlet temperature;

[0066] S102. Taking the power generation coal consumption rate as the optimization target, a dynamic optimization regulation model of extraction steam energy storage is established, wherein the objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate, and the constraints include the upper and lower limits of the extraction steam flow rate, the molten salt flow rate, the extraction steam temperature, the molten salt heat storage temperature and the reheater outlet temperature; the established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of the extraction steam energy storage operating parameters under different working conditions, thereby forming a dynamic optimization regulation strategy library of extraction steam energy storage;

[0067] S103. According to the grid load forecast result, the optimal regulation strategy is selected from the obtained dynamic optimization regulation strategy library, and the set values ​​of the high-pressure cylinder extraction pressure, extraction temperature and extraction flow are transmitted to the steam turbine extraction control system, and the set values ​​of the molten salt heat storage temperature and molten salt flow are transmitted to the high-temperature molten salt heat storage control system; the steam turbine extraction process and the molten salt heat storage process are coordinated to realize the flexible operation and efficient peak regulation of the thermal power generating units under the dynamic fluctuation of the grid load.

[0068] In an optional embodiment,

[0069] A thermal coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit is established. The thermal coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system, including:

[0070] Establishing a thermodynamic model of a steam turbine high-pressure cylinder, the thermodynamic model comprising a steam flow calculation formula based on the Stodola ellipse law and an isentropic efficiency correction formula considering partial load performance changes; calculating the steam flow and the corrected isentropic efficiency of the steam turbine high-pressure cylinder according to the thermodynamic model of the steam turbine high-pressure cylinder;

[0071] Establishing a heat transfer flow model of a high-temperature molten salt heat storage device, the heat transfer flow model of the high-temperature molten salt heat storage device includes a heat transfer equation of a heat exchanger based on a logarithmic mean temperature difference method and a modified Dittus-Boelter formula considering the flow characteristics of molten salt; using the heat transfer flow model of the high-temperature molten salt heat storage device, calculating the heat transfer power of the high-temperature molten salt heat storage device and the Nusselt number of the molten salt flow;

[0072] Establishing a mass energy balance equation for the steam turbine extraction and heat storage system, the mass energy balance equation includes a mass balance equation for the extraction point, an energy balance equation, and an energy accumulation rate equation for the high-temperature molten salt heat storage system; based on the mass energy balance equation, calculating the steam flow rate and specific enthalpy at each point in the system and the energy change rate of the high-temperature molten salt heat storage system;

[0073] The thermodynamic model of the steam turbine high-pressure cylinder, the heat transfer flow model of the high-temperature molten salt heat storage device and the mass energy balance equation of the steam turbine extraction and heat storage system are combined to construct a complete thermomechanical coupling mathematical model.

[0074] Exemplarily, first, a thermodynamic model of a steam turbine high-pressure cylinder is established. The model includes a steam flow calculation formula based on the Stodola ellipse law and an isentropic efficiency correction formula that takes into account partial load performance changes. The Stodola ellipse law is used to describe the relationship between the inlet and outlet pressures of the steam turbine high-pressure cylinder and the steam flow. The isentropic efficiency correction formula takes into account the change in efficiency of the steam turbine when it is running at partial load. Through these two formulas, the steam flow and the corrected isentropic efficiency of the steam turbine high-pressure cylinder can be calculated.

[0075] For example, for a steam turbine with a rated power of 300MW, the steam parameters at the inlet of the high-pressure cylinder are 16.7MPa and 538℃, and the outlet pressure is 3.8MPa. The steam flow rate under rated conditions is calculated to be 250kg / s using the Stodola ellipse law. Through the isentropic efficiency correction formula, it can be obtained that the isentropic efficiency at 75% load is 87%, which is lower than 89% at rated load.

[0076] Next, a heat transfer flow model of the high-temperature molten salt heat storage device is established. The model includes the heat transfer equation of the heat exchanger based on the logarithmic mean temperature difference method and the modified Dittus-Boelter formula considering the flow characteristics of the molten salt. The logarithmic mean temperature difference method is used to calculate the heat transfer power of the heat exchanger, while the modified Dittus-Boelter formula is used to calculate the Nusselt number of the molten salt flow, thereby obtaining the convective heat transfer coefficient.

[0077] Taking a high-temperature molten salt heat storage device with a capacity of 1000MWh as an example, a mixed molten salt of 60% NaNO3 and 40% KNO3 is used as the heat storage medium. The heat transfer power under the design condition is calculated to be 50MW by the logarithmic mean temperature difference method. Using the modified Dittus-Boelter formula, the Nusselt number is calculated to be 120 when the molten salt flow rate is 2m / s, and the corresponding convective heat transfer coefficient is 6000W / (m 2 ·K).

[0078] Then, the mass and energy balance equations of the steam turbine extraction and heat storage system are established. This includes the mass balance equation of the extraction point, the energy balance equation, and the energy accumulation rate equation of the high-temperature molten salt heat storage system. The mass balance equation ensures the steam flow balance of each extraction point, the energy balance equation ensures the energy conservation of each part of the system, and the energy accumulation rate equation describes the change of the energy of the heat storage system over time.

[0079] Take a system with three-stage steam extraction as an example. The first stage steam extraction pressure is 3.8MPa and the flow rate is 20kg / s; the second stage steam extraction pressure is 1.7MPa and the flow rate is 15kg / s; the third stage steam extraction pressure is 0.5MPa and the flow rate is 10kg / s. The mass balance equation can be used to calculate the steam flow change between each stage. The energy balance equation shows that the total steam extraction energy is 100MW under rated conditions. The energy accumulation rate equation shows that during the energy storage process, the energy of the thermal storage system increases at a rate of 45MW.

[0080] Finally, the thermodynamic model of the high-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system are combined to construct a complete thermodynamic coupling mathematical model. This comprehensive model can describe the thermodynamic behavior of the entire system, including the performance of each cylinder of the steam turbine, the heat transfer characteristics of the heat storage device, and the mass and energy balance of the entire system.

[0081] Through this thermal-mechanical coupling mathematical model, the system performance under different operating conditions can be simulated. For example, when the grid load is low, the steam extraction can be increased to 25kg / s, and the thermal storage power can reach 55MW. At peak load, the steam extraction can be reduced to 5kg / s, and the stored thermal energy can be used to generate an additional 15MW of power output.

[0082] This thermodynamic coupling mathematical model lays the foundation for subsequent optimization control. By adjusting variables such as steam extraction parameters and molten salt flow, system performance can be optimized, energy efficiency can be improved, and the peak load regulation capacity of the power grid can be enhanced. For example, the optimal steam extraction strategy can be formulated in advance based on the predicted power grid load curve to maximize economic benefits while ensuring safe and stable operation of the unit.

[0083] In addition, the model can also be used for dynamic response analysis of the system. By introducing the time factor, the transient characteristics of the system when the load changes rapidly can be studied, providing a basis for the formulation of real-time control strategies. For example, it can be analyzed how the thermal storage system cooperates with the steam turbine to respond quickly and maintain the stability of the system when the load jumps from 50% to 100% within 5 minutes.

[0084] In an optional embodiment,

[0085] Taking the power generation coal consumption rate as the optimization target, a dynamic optimization regulation model of extraction steam energy storage is established. The objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate. The constraints include the upper and lower limits of the extraction steam flow rate, molten salt flow rate, extraction steam temperature, molten salt heat storage temperature and reheater outlet temperature. The established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of extraction steam energy storage operating parameters under different working conditions. The dynamic optimization regulation strategy library of extraction steam energy storage includes:

[0086] Initializing a population to generate an initial population including decision variables, wherein the decision variables include a high-pressure cylinder extraction steam pressure, an extraction steam temperature, an extraction steam flow rate, a high-temperature molten salt heat storage temperature, a molten salt flow rate, and a reheater outlet temperature;

[0087] The initial population is non-dominatedly sorted to divide the individuals into different non-dominated layers; uniformly distributed reference points are generated in the standardized target space; the distance from each individual to the nearest reference point is calculated, and individuals are selected to enter the next generation population according to the distance; the population diversity index is calculated, which is the average value of the Euclidean distances of all individuals to the population center;

[0088] According to the population diversity index, adaptively adjusting the crossover rate and the mutation rate, wherein the crossover rate decreases as the population diversity index increases, and the mutation rate increases as the population diversity index increases, and the adjustment of the crossover rate and the mutation rate both adopts an exponential function form; performing crossover and mutation operations on the selected individuals to generate a progeny population;

[0089] Select some individuals with a preset probability for simulated annealing optimization, generate new solutions in the neighborhood of the selected individuals, calculate the energy difference, calculate the probability of accepting the new solution based on the energy difference and the current temperature, and update the temperature according to the preset cooling coefficient;

[0090] Calculate the improved crowding, including calculating the target space crowding and the decision space crowding, and obtain the comprehensive crowding by weighted average, wherein the target space crowding and the decision space crowding consider the distances of adjacent individuals in the target space and the decision space respectively;

[0091] The population is divided into multiple sub-populations, with the number of sub-populations equal to the number of available processor cores. For each sub-population, objective function evaluation, constraint processing, non-dominated sorting, and crowding calculation are performed in parallel on a separate processor core; the evaluation results of each sub-population are summarized, and global non-dominated sorting and selection operations are performed; the parent population and the child population are merged, and the merged population is non-dominated sorted, and the top N individuals are selected to form the next generation population;

[0092] Determine whether the maximum number of iterations has been reached or the convergence condition has been met. If so, output the optimization result. If not, return to the step of calculating the distance from each individual to the nearest reference point. According to the output optimization result, determine the optimal operating parameters of the extraction steam storage system to achieve dynamic optimization control of the system.

[0093] Exemplarily, a dynamic optimization regulation model is established. The objective function of the model is to minimize the power generation coal consumption rate, and the constraints include the upper and lower limits of the extraction steam flow rate, molten salt flow rate, extraction steam temperature, molten salt heat storage temperature and reheater outlet temperature. For example, for a 300MW unit, the extraction steam flow rate can be set to 0-50kg / s, the molten salt flow rate range is 100-500kg / s, the extraction steam temperature range is 300-500℃, the molten salt heat storage temperature range is 290-565℃, and the reheater outlet temperature range is 535-565℃. These constraints ensure that the system operates within a safe and stable range.

[0094] Next, the improved non-dominated sorting genetic algorithm is used to solve the established dynamic optimization adjustment model. The specific implementation steps of the algorithm are as follows:

[0095] Initialize population: Generate an initial population containing decision variables. Decision variables include high-pressure cylinder extraction steam pressure, extraction steam temperature, extraction steam flow rate, high-temperature molten salt heat storage temperature, molten salt flow rate, and reheater outlet temperature. Assume that the population size is 100, and each individual contains 6 decision variables. For example, the possible decision variable values ​​of an individual are: extraction steam pressure 3.5MPa, extraction steam temperature 450℃, extraction steam flow rate 30kg / s, molten salt heat storage temperature 520℃, molten salt flow rate 300kg / s, and reheater outlet temperature 550℃.

[0096] Perform non-dominated sorting on the initial population: Divide the individuals into different non-dominated layers. Generate uniformly distributed reference points in the standardized target space, assuming that 100 reference points are generated. Calculate the distance from each individual to the nearest reference point, and select individuals to enter the next generation population based on the distance. Calculate the population diversity index, which is the average of the Euclidean distances of all individuals to the population center. For example, the calculated population diversity index is 0.8.

[0097] According to the population diversity index, the crossover rate and mutation rate are adaptively adjusted: the crossover rate decreases as the population diversity index increases, and the mutation rate increases as the population diversity index increases, and both are adjusted in the form of exponential functions. For example, when the population diversity index is 0.8, the calculated crossover rate is 0.7 and the mutation rate is 0.1. The selected individuals are crossover and mutation operations are performed to generate the offspring population.

[0098] Select some individuals for simulated annealing optimization with a preset probability: Assume that the preset probability is 0.2, that is, 20 individuals are randomly selected from 100 individuals for simulated annealing optimization. Generate a new solution in the neighborhood of the selected individual, calculate the energy difference, and calculate the probability of accepting the new solution based on the energy difference and the current temperature. For example, the initial temperature is set to 100 and the cooling coefficient is 0.95. After each iteration, the temperature is updated to the current temperature multiplied by the cooling coefficient.

[0099] Calculate the improved crowding: including calculating the target space crowding and decision space crowding, and obtaining the comprehensive crowding by weighted average. The target space crowding and decision space crowding consider the distance between adjacent individuals in the target space and decision space respectively. Assume that the weight of the target space crowding is 0.6 and the weight of the decision space crowding is 0.4.

[0100] Divide the population into multiple subpopulations to achieve parallel computing: Assuming that the number of available processor cores is 4, 100 individuals are evenly distributed into 4 subpopulations, each containing 25 individuals. For each subpopulation, objective function evaluation, constraint processing, non-dominated sorting, and crowding calculation are performed in parallel on a separate processor core. After completion, the evaluation results of each subpopulation are summarized to perform global non-dominated sorting and selection operations.

[0101] Merge the parent population and the offspring population, perform non-dominated sorting on the merged population, and select the first N individuals to form the next generation population. For example, if the parent and offspring populations each have 100 individuals, and there are 200 individuals after merging, then select the 100 individuals with the best sorting as the next generation population.

[0102] Determine whether the maximum number of iterations has been reached or the convergence condition has been met: Assume that the maximum number of iterations is set to 1000, and the convergence condition is that the change in the optimal solution for 50 consecutive generations is less than 1%. If the condition is met, the optimization result is output, otherwise, the step of calculating the distance from each individual to the nearest reference point is returned to continue iterative optimization.

[0103] According to the output optimization results, the optimal operating parameters of the extraction steam energy storage system are determined to achieve dynamic optimization control of the system. For example, the optimization results may show that under a certain operating condition, the optimal operating parameters are: extraction steam pressure 3.2MPa, extraction steam temperature 460℃, extraction steam flow 35kg / s, molten salt heat storage temperature 540℃, molten salt flow 350kg / s, reheater outlet temperature 555℃. This set of parameters can achieve the lowest power generation coal consumption rate while meeting all constraints.

[0104] Through this dynamic optimization and adjustment method, the optimal operating parameters of the steam extraction energy storage system can be quickly obtained for different load conditions and grid requirements. For example, during the low load period of the grid, the steam extraction volume may be increased to 40kg / s and the molten salt heat storage temperature may be increased to 560℃ to maximize the energy storage effect. During the peak load period, the steam extraction volume may be reduced to 10kg / s and the molten salt heat storage temperature may be reduced to 500℃ to improve power generation efficiency.

[0105] An important feature of this optimization method is that it can consider multiple objectives and constraints at the same time. Although the main optimization target is the power generation coal consumption rate, other performance indicators such as system response speed and equipment life can also be taken into account through non-dominated sorting and improved congestion calculation. This makes the optimization results more comprehensive and practical.

[0106] In addition, the parallel computing characteristics of this method enable it to quickly handle large-scale optimization problems. For complex power systems, dozens of decision variables and constraints may need to be considered, and traditional optimization methods may take several hours or even days to get results. Through parallel computing, the calculation time can be greatly shortened, achieving near real-time optimization control.

[0107] Finally, by building a dynamic optimization regulation strategy library, the system can quickly select the most appropriate operating parameters based on the current operating conditions and predicted future load changes. This predictive control strategy can respond to load changes in advance and further improve the system's regulation capability and economy.

[0108] In summary, this dynamic optimization regulation method of extracted steam energy storage based on the improved non-dominated sorting genetic algorithm provides strong technical support for the efficient operation of the high-temperature molten salt heat storage system, helps to improve the flexibility and economy of thermal power generating units, and contributes to the stable operation of the power grid and the large-scale access of renewable energy.

[0109] In an optional embodiment,

[0110] According to the grid load forecast results, the optimal regulation strategy is selected from the obtained dynamic optimization regulation strategy library, and the set values ​​of the high-pressure cylinder extraction steam pressure, extraction steam temperature, and extraction steam flow are transmitted to the turbine extraction steam control system, and the set values ​​of the molten salt heat storage temperature and molten salt flow are transmitted to the high-temperature molten salt heat storage control system, including:

[0111] A deep learning model of long short-term memory network and attention mechanism is used to predict power grid load, wherein the deep learning model includes an input layer, a long short-term memory network layer, an attention layer, a fully connected layer and an output layer, wherein the long short-term memory network layer is provided with 3 layers, each layer includes 128 neurons, and the fully connected layer includes 2 layers, respectively including 64 and 32 neurons;

[0112] A dynamic optimization regulation strategy library is constructed using a dual deep Q network algorithm, the dual deep Q network algorithm includes an action value network and a target network, the action value network and the target network are both composed of three fully connected layers, the number of neurons is 256, 128 and 64 respectively, and the activation function is a ReLU function; the reward function of the dynamic optimization regulation strategy library takes into account the improvement of power generation efficiency, the improvement of peak load capacity, the equipment wear cost and the operating cost;

[0113] Based on the grid load forecasting results, a fuzzy comprehensive evaluation method is used to select the optimal regulation strategy from the dynamic optimization regulation strategy library, wherein the fuzzy comprehensive evaluation method includes establishing a fuzzy relationship matrix and determining the index weights, wherein the index weights are determined by a hierarchical analysis method;

[0114] The high-pressure cylinder extraction pressure set value, extraction temperature set value, and extraction flow set value in the optimal regulation strategy are transmitted to the turbine extraction control system through the OPC UA protocol, and the molten salt heat storage temperature set value and the molten salt flow set value are transmitted to the high-temperature molten salt heat storage control system through the OPC UA protocol. The OPC UA protocol uses an X.509 certificate for identity authentication and uses the AES-256 encryption algorithm to encrypt communication data;

[0115] Exemplarily, a deep learning model with long short-term memory network and attention mechanism is used for power grid load forecasting. The structure of the model includes input layer, long short-term memory network layer, attention layer, fully connected layer and output layer. The input layer receives relevant factors such as historical load data, weather information, date features, etc. The long short-term memory network layer is set up with 3 layers, each containing 128 neurons, which are used to capture the long-term dependencies of time series data. The attention layer is used to weight the features of different time steps to improve the model's attention to key information. The fully connected layer includes 2 layers, containing 64 and 32 neurons respectively, for further feature extraction. The output layer gives the load forecast results for the next 24 hours.

[0116] For example, for a city's power grid load forecast, the input data includes hourly load data for the past 7 days (168 data points), temperature, humidity, wind speed forecasts for the current and next 24 hours, and date types (weekdays, weekends, holidays). The model training uses historical data from the past 3 years, using the Adam optimizer, with a learning rate of 0.001 and a batch size of 64. After 1,000 rounds of training, the model's mean absolute percentage error (MAPE) on the test set reached 2.5%, and the root mean square error (RMSE) was 15MW.

[0117] Next, a dual deep Q network algorithm is used to build a dynamic optimization adjustment strategy library. The algorithm includes an action value network and a target network. The two networks have the same structure, both consisting of three fully connected layers, with 256, 128 and 64 neurons respectively, and the activation function is the ReLU function. The action value network is used to evaluate the value of each action in the current state, and the target network is used to generate the target Q value to improve the stability of training.

[0118] The state space of the dynamic optimization regulation strategy library includes the current load level, load change trend, current values ​​of various parameters of the steam extraction storage system, etc. The action space includes the adjustment of the high-pressure cylinder steam extraction pressure, steam extraction temperature, steam extraction flow, molten salt heat storage temperature and molten salt flow. The reward function comprehensively considers the improvement of power generation efficiency, the improvement of peak-shaving capacity, the equipment wear cost and the operating cost. For example, the weight of the power generation efficiency improvement can be set to 0.4, the weight of the peak-shaving capacity improvement can be set to 0.3, the weight of the equipment wear cost can be set to 0.2, and the weight of the operating cost can be set to 0.1.

[0119] During the training process, the ε-greedy strategy was used for exploration. The initial ε value was set to 1 and gradually decayed to 0.01 as the training progressed. The experience replay buffer size was set to 10,000, and the target network parameters were updated every 4 steps. The training lasted for 1 million steps, and finally a model was obtained that could quickly give the optimal adjustment strategy based on the current state.

[0120] Based on the grid load forecast results, the fuzzy comprehensive evaluation method is used to select the optimal regulation strategy from the dynamic optimization regulation strategy library. First, a fuzzy relationship matrix is ​​established, and the evaluation indicators include power generation efficiency, peak-shaving capacity, equipment life impact and economy. For each candidate strategy, the membership value of each indicator is given based on expert experience and historical data. For example, the membership value of a strategy may be: power generation efficiency 0.8, peak-shaving capacity 0.7, equipment life impact 0.6, and economy 0.75.

[0121] The indicator weights are determined by the hierarchical analysis method. First, a judgment matrix is ​​constructed to compare the relative importance of each indicator. Assume that after expert evaluation, the weight vector obtained is: power generation efficiency 0.35, peak load capacity 0.30, equipment life impact 0.20, and economy 0.15. Multiply the weight vector with the fuzzy relationship matrix to obtain the comprehensive evaluation score of each candidate strategy. Select the strategy with the highest score as the optimal regulation strategy.

[0122] Finally, the parameter setting values ​​in the optimal adjustment strategy are transmitted to the corresponding control system through the OPC UA protocol. OPC UA (OPC Unified Architecture) is an industrial communication protocol with good cross-platform and security. In this embodiment, X.509 certificates are used for identity authentication to ensure that the identities of both parties in communication are authentic. The communication data is encrypted using the AES-256 encryption algorithm to protect the security of data transmission.

[0123] Specifically, the high-pressure cylinder extraction pressure set value, extraction temperature set value, and extraction flow set value are transmitted to the turbine extraction control system through the OPC UA protocol. For example, the optimal strategy at a certain moment may be: extraction pressure 3.2MPa, extraction temperature 460℃, extraction flow 35kg / s. These parameters are encoded as OPC UA data points and sent to the OPC UA server of the turbine control system through an encrypted channel.

[0124] Similarly, the molten salt heat storage temperature setpoint and the molten salt flow setpoint are transmitted to the high-temperature molten salt heat storage control system via the OPC UA protocol. For example, the molten salt heat storage temperature in the optimal strategy may be 540°C and the molten salt flow rate may be 350kg / s. These data are also transmitted to the controller of the heat storage system via a secure OPC UA channel.

[0125] After receiving these set values, the control system will adjust the corresponding actuators through PID controllers or more advanced model predictive control algorithms, such as adjusting valve openings and pump speeds, so that the actual operating parameters of the system gradually approach the set values. At the same time, the actual operating data will be fed back to the upper control system through the OPC UA protocol for monitoring and the next round of optimization calculations.

[0126] This dynamic optimization and adjustment method based on deep learning and reinforcement learning has the following advantages:

[0127] The power grid load forecasting model can accurately capture the load change trend and provide a reliable basis for optimizing decision-making. Through the long short-term memory network and attention mechanism, the model can effectively handle long-term dependencies and focus on key information to improve forecasting accuracy.

[0128] The dynamic optimization regulation strategy library constructed by the dual deep Q network algorithm can learn complex nonlinear decision rules and adapt to the optimal control strategy under different working conditions. By considering the reward function of multiple objectives, the comprehensive optimization of power generation efficiency, peak load regulation capacity and economy is achieved.

[0129] The fuzzy comprehensive evaluation method introduces expert knowledge in the process of strategy selection, which improves the interpretability and reliability of decision-making. The indicator weights are determined by the hierarchical analysis method, making the evaluation results more objective and reasonable.

[0130] The OPC UA protocol is used for data transmission, ensuring the interoperability and security of system integration. X.509 certificate authentication and AES-256 encryption protect the secure transmission of control instructions and prevent unauthorized access and data tampering.

[0131] Through this method, the steam extraction energy storage system can dynamically adjust the operating parameters according to the grid load forecast results, while ensuring the safe and stable operation of the system, while maximizing the power generation efficiency and peak load regulation capacity. For example, when the load trough period is predicted, the system may increase the steam extraction volume and increase the molten salt heat storage temperature to prepare for the subsequent peak load regulation. During the peak load period, the system will reduce the steam extraction volume, increase the molten salt heat release, and improve the power generation output of the unit.

[0132] This intelligent control strategy not only improves the flexibility and economy of thermal power generating units, but also provides strong support for the stable operation of the power grid and the large-scale access of renewable energy.

[0133] In an optional embodiment,

[0134] The coordinated control of the steam turbine extraction process and the molten salt heat storage process enables flexible operation and efficient peak load regulation of thermal power generating units under dynamic fluctuations of power grid load, including:

[0135] A model predictive control method is used to realize the coordinated control of the steam turbine extraction process and the molten salt heat storage process. The objective function of the model predictive control method includes an output tracking error term and a control increment penalty term, and the constraints include an extraction pressure range, an extraction temperature range, an extraction flow range, a molten salt temperature range, a molten salt flow range, and a control increment limit.

[0136] An improved fast gradient method is used to solve the model predictive control optimization problem. The improved fast gradient method includes converting the original problem into a quadratic programming problem, using the Nesterov accelerated gradient method for iterative solution, and adopting a hot start strategy. A recursive least squares algorithm is introduced to update the system model parameters online. The forgetting factor of the recursive least squares algorithm is set to 0.98 to achieve adaptive update of the system model.

[0137] Exemplarily, the present invention provides a method for coordinated control of a steam extraction energy storage system of a thermal power generator set based on model predictive control, which is used to achieve flexible operation and efficient peak load regulation under dynamic fluctuations of power grid load. The method mainly includes using model predictive control to achieve coordinated control of the steam extraction process of the steam turbine and the molten salt heat storage process, and using an improved fast gradient method to solve the optimization problem.

[0138] First, a model predictive control method is established. The objective function of this method includes an output tracking error term and a control increment penalty term. The output tracking error term is used to measure the deviation between the actual output of the system and the expected output, and the control increment penalty term is used to limit the drastic changes in the control quantity to ensure the smooth operation of the system. The constraints include the extraction pressure range, extraction temperature range, extraction flow range, molten salt temperature range, molten salt flow range and control increment limit. For example, for a 300MW unit, the extraction pressure range can be set to 2-5MPa, the extraction temperature range is 300-500℃, the extraction flow range is 0-50kg / s, the molten salt temperature range is 290-565℃, and the molten salt flow range is 100-500kg / s. The control increment limit can be set to a rated value that does not change by more than 10% in each control cycle.

[0139] Next, the improved fast gradient method is used to solve the model predictive control optimization problem. The method includes the following steps:

[0140] Convert the original problem into a quadratic programming problem: By linearizing and discretizing the objective function and constraints, the original nonlinear optimization problem is converted into a standard quadratic programming problem. This step can greatly simplify the problem-solving process and improve computational efficiency.

[0141] Use Nesterov accelerated gradient method for iterative solution: Nesterov accelerated gradient method is an improved gradient descent algorithm with a faster convergence speed. In practical applications, the initial step size can be set to 0.01 and the maximum number of iterations can be set to 100. By continuously updating the gradient and momentum terms, the optimal solution is gradually approached.

[0142] Adopting the hot start strategy: using the optimization result of the previous moment as the initial value of the current moment optimization can significantly reduce the number of iterations and speed up the solution. For example, if the optimal extraction steam flow rate of the previous moment is 30kg / s, then this value can be used as the initial guess value of the current moment optimization.

[0143] In order to adapt to the changes in system parameters and improve control accuracy, a recursive least squares algorithm is introduced to update the system model parameters online. The forgetting factor of the recursive least squares algorithm is set to 0.98, which means that the algorithm will gradually "forget" old data and rely more on new observation data, thereby achieving adaptive updates of the system model.

[0144] The specific implementation process is as follows:

[0145] First, obtain the current state of the system and the load forecast value for a period of time in the future. For example, you can obtain the current state variables such as extraction steam pressure, temperature, flow rate, and molten salt temperature and flow rate, and obtain the load forecast value every 15 minutes in the next 4 hours.

[0146] Then, based on the system model and the predicted load changes, the system response in the future time window is calculated. Assuming the prediction time window is 4 hours and the control cycle is 15 minutes, the system response at 16 time points needs to be calculated.

[0147] Next, construct the objective function. The objective function includes the output tracking error term and the control increment penalty term. The output tracking error can be defined as the sum of the squared differences between the predicted output and the desired output, and the control increment penalty term can be defined as the sum of the squared changes in the control quantity. The control performance and system stability can be balanced by adjusting the weights of these two terms. For example, the weight of the output tracking error term can be set to 1, and the weight of the control increment penalty term can be set to 0.1.

[0148] Construct constraints. According to the physical limitations and safety requirements of the system, set the value range and change rate limit of each variable. For example, the change rate of the extraction steam pressure can be limited to no more than 0.1MPa per minute.

[0149] The improved fast gradient method is used to solve the optimization problem. First, the problem is transformed into a standard quadratic programming form, and then the Nesterov accelerated gradient method is used for iterative solution. In each iteration, the objective function value and gradient are calculated, and the search direction and step size are updated. When the maximum number of iterations is reached or the convergence condition is met, the iteration stops.

[0150] After the optimal control sequence is obtained, only the control action of the first control cycle is executed. For example, if the optimization result shows that the optimal extraction steam flow sequence in the next 4 hours is [35,36,38,37,...] kg / s, only the 35 kg / s setpoint of the first control cycle is executed.

[0151] While executing control actions, the system model parameters are updated using a recursive least squares algorithm. New system input and output data are collected, the prediction error is calculated, and then the model parameters are updated. For example, if the actual extraction steam flow response is found to be slower than the model prediction, the time constant parameters in the model are adjusted accordingly.

[0152] Repeat the above steps and re-solve the optimization problem at the beginning of each control cycle to achieve rolling optimization control.

[0153] Through this coordinated control method, the steam turbine extraction process and molten salt heat storage process can be flexibly adjusted according to the dynamic changes of the grid load. For example, when the grid load suddenly decreases, the controller will increase the steam extraction volume, while increasing the molten salt flow rate and heat storage temperature to achieve rapid energy transfer and storage. Conversely, when the load suddenly increases, the controller will reduce the steam extraction volume, reduce the molten salt flow rate, release the stored heat energy, and increase power generation output.

[0154] A key advantage of this collaborative control strategy is the ability to anticipate load changes. By taking into account load forecasts for a period of time in the future, the controller can make adjustments in advance to avoid delayed response of the system. For example, if a load peak is predicted to occur in 2 hours, the controller will increase the heat storage of molten salt in advance to prepare for the upcoming high-load operation.

[0155] Furthermore, by updating the system model parameters online, the method is able to adapt to slowly changing system characteristics, such as performance changes caused by equipment aging, seasonal changes, etc. This ensures the long-term stability of the control performance.

[0156] In short, this collaborative control method based on model predictive control achieves flexible operation and efficient peak load regulation of thermal power generating units under dynamic fluctuations of power grid load by optimizing the steam extraction process of steam turbines and the molten salt heat storage process at the same time. It not only improves the peak load regulation capacity and economy of the units, but also provides strong support for the stable operation of the power grid and the large-scale access of renewable energy.

[0157] In an optional embodiment,

[0158] The improved fast gradient method is used to solve the model predictive control optimization problem. The improved fast gradient method includes converting the original problem into a quadratic programming problem, using the Nesterov accelerated gradient method for iterative solution, and adopting a hot start strategy including:

[0159] The Nesterov accelerated gradient method includes the following sub-steps:

[0160] Initialize the decision variable vector, auxiliary variable vector, step size and momentum parameter; calculate the gradient of the objective function; update the decision variable vector and auxiliary variable vector; project the updated decision variable vector into the feasible domain; dynamically adjust the step size using the Barzilai-Borwein method; dynamically adjust the momentum parameter according to the number of iterations;

[0161] Implement the hot start strategy and use the optimal solution of the previous control cycle as the initial value of the current control cycle, including:

[0162] Save the optimal solution of the previous control cycle; construct a new initial decision variable vector, where the first N-1 control increments come from the optimal solution of the previous cycle and the last control increment is set to zero; use a larger step size for warm-up iterations to quickly approach the optimal solution.

[0163] Exemplarily, the present invention provides a method for solving a model predictive control optimization problem using an improved fast gradient method, the method comprising converting the original problem into a quadratic programming problem, using the Nesterov accelerated gradient method for iterative solution, and adopting a hot start strategy. This method can effectively improve the solution speed and convergence, and is suitable for real-time control of a steam extraction storage system of a thermal power generator set.

[0164] First, the original model predictive control optimization problem is transformed into a standard quadratic programming problem. This step includes linearizing the system model, discretizing the prediction time domain, and constructing the objective function and constraints. For example, for a system with two control variables, extraction steam flow and molten salt flow, the control sequence in the next 4 hours can be represented as a 96-dimensional decision variable vector (assuming the control cycle is 5 minutes). The objective function can be expressed as the quadratic form of this decision variable vector plus a linear term, and the constraints are expressed as linear inequalities.

[0165] Next, the improved Nesterov accelerated gradient method is used for iterative solution. This process includes the following key steps:

[0166] Initialize the decision variable vector, auxiliary variable vector, step size, and momentum parameter. The decision variable vector can be initialized to an all-zero vector or the optimal solution of the previous cycle. The auxiliary variable vector is initialized to the decision variable vector. The initial step size can be set to a small value, such as 0.01. The initial value of the momentum parameter can be set to 0.9.

[0167] Calculate the gradient of the objective function. This step requires calculating the partial derivatives of the objective function with respect to the decision variables. For quadratic programming problems, the gradient can be calculated efficiently using matrix operations.

[0168] Update the decision variable vector and auxiliary variable vector. Use the calculated gradient, combined with the current step size and momentum parameters, to update the decision variable vector and auxiliary variable vector. This step is the core of the Nesterov accelerated gradient method, which accelerates convergence by introducing momentum terms.

[0169] Project the updated decision variable vector into the feasible region. This step ensures that the updated solution satisfies all constraints. For simple bound constraints, out-of-range values ​​can be truncated directly to the bounds. For more complex constraints, a quadratic programming subproblem may need to be solved.

[0170] The Barzilai-Borwein method is used to dynamically adjust the step size. This method adaptively adjusts the step size according to the gradient change between two consecutive iterations, which can significantly improve the convergence speed. Specifically, the new step size can be estimated by dividing the gradient difference between the current iteration and the previous iteration by the decision variable difference.

[0171] Dynamically adjust the momentum parameter according to the number of iterations. As the iterations proceed, gradually increasing the momentum parameter can maintain stability in the early stage and accelerate convergence in the later stage. For example, the momentum parameter can be gradually increased from the initial 0.9 to 0.99.

[0172] Implement a hot start strategy and use the optimal solution of the previous control cycle as the initial value of the current control cycle. This strategy includes the following steps:

[0173] Save the optimal solution of the previous control cycle. At the end of each control cycle, save the optimal decision variable vector obtained to the memory or database.

[0174] Construct a new initial decision variable vector. Use the saved optimal solution of the previous cycle to construct a new initial value. Specifically, for a problem with a control time domain of N, the last N-1 control increments in the optimal solution of the previous cycle can be used as the first N-1 elements of the new initial value, and the last element is set to zero. This construction method ensures the continuity of the control sequence.

[0175] Use a larger step size for warm-up iterations. Before officially starting the Nesterov accelerated gradient method, you can use a larger step size (such as 0.1) to perform several warm-up iterations. This helps to quickly approach the optimal solution and reduce the overall number of iterations.

[0176] The following is a specific data case to illustrate the implementation process of this method:

[0177] Assume that we have a problem with a control horizon of 12 (corresponding to a 1-hour prediction horizon and a control period of 5 minutes), and the control variables include extraction steam flow and molten salt flow. Then the dimension of the decision variable vector is 24 (12*2).

[0178] In the initialization stage, we can set the decision variable vector to [30,30,...,30,200,200,...,200], where the first 12 elements represent the steam extraction flow rate (unit: kg / s) and the last 12 elements represent the molten salt flow rate (unit: kg / s). The initial step size is set to 0.01 and the momentum parameter is set to 0.9.

[0179] In each iteration, we first calculate the gradient of the objective function. Suppose in a certain iteration, the calculated gradient vector is [-0.5,-0.3,...,0.2,0.1,0.3,...,-0.1].

[0180] Then, we use this gradient to update the decision variable vector and the auxiliary variable vector. After the update, assume that the new decision variable vector is [30.5, 30.3, ..., 29.8, 199.9, 199.7, ..., 200.1].

[0181] Next, we need to project this vector into the feasible domain. Assuming that the range of the extraction steam flow rate is [0,50] kg / s and the range of the molten salt flow rate is [100,300] kg / s, the result after projection remains unchanged because all values ​​are within the feasible range.

[0182] Use the Barzilai-Borwein method to adjust the step size. Suppose the calculated new step size is 0.015.

[0183] Dynamically adjust the momentum parameter. If the current iteration is the 50th, we can adjust the momentum parameter to 0.95.

[0184] The above process is repeated until the maximum number of iterations (such as 1000) is reached or the convergence condition is met (such as the relative change of the objective function value is less than 1e-6).

[0185] At the beginning of the next control cycle, we use the warm start strategy. Assume that the optimal solution obtained in the previous cycle is [32,33,35,...,40,210,215,220,...,250]. We construct a new initial decision variable vector as [33,35,...,40,0,215,220,...,250,0], where the last two elements are set to 0.

[0186] Use a larger step size (such as 0.1) for 10 warmup iterations, and then switch back to the normal iteration process.

[0187] Through this improved fast gradient method, combined with Nesterov acceleration and hot start strategy, we can quickly solve the model predictive control optimization problem in each control cycle and provide a real-time optimal control sequence for the steam extraction storage system of the thermal power generator. This method not only improves the solution speed, but also enhances the robustness and convergence of the algorithm, can effectively deal with the nonlinear and time-varying characteristics of the system model, and realize flexible operation and efficient peak regulation under the dynamic fluctuation of the power grid load.

[0188] Figure 2 FIG. 1 is a schematic diagram of a steam extraction energy storage regulation optimization system using a high-temperature molten salt heat storage system according to an embodiment of the present invention. Figure 2 As shown, the system comprises:

[0189] The first unit is used to establish a thermodynamic coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit, the thermodynamic coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system; determine the key parameters that affect the extraction steam energy storage performance, the key parameters include the high-pressure cylinder extraction steam pressure, extraction steam temperature, extraction steam flow rate, high-temperature molten salt heat storage temperature, molten salt flow rate, and reheater outlet temperature;

[0190] The second unit is used to establish a dynamic optimization regulation model of extraction steam energy storage with the power generation coal consumption rate as the optimization target, wherein the objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate, and the constraints include the upper and lower limits of the extraction steam flow rate, the molten salt flow rate, the extraction steam temperature, the molten salt heat storage temperature and the reheater outlet temperature; the established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of the extraction steam energy storage operating parameters under different working conditions, thereby forming a dynamic optimization regulation strategy library of extraction steam energy storage;

[0191] The third unit is used to select the optimal regulation strategy from the obtained dynamic optimization regulation strategy library according to the grid load forecast result, transmit the set values ​​of high-pressure cylinder extraction pressure, extraction temperature and extraction flow to the turbine extraction control system, and transmit the set values ​​of molten salt heat storage temperature and molten salt flow to the high-temperature molten salt heat storage control system; coordinately control the turbine extraction process and the molten salt heat storage process to realize flexible operation and efficient peak regulation of thermal power generating units under dynamic fluctuations of grid load.

[0192] According to a third aspect of the embodiments of the present invention,

[0193] An electronic device is provided, comprising:

[0194] processor;

[0195] a memory for storing processor-executable instructions;

[0196] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0197] According to a fourth aspect of the embodiments of the present invention,

[0198] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0199] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0200] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A steam extraction energy storage regulation optimization method using a high-temperature molten salt heat storage system, characterized in that: include: Establish a thermal coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit, the thermal coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system; determine the key parameters that affect the extraction steam energy storage performance, the key parameters include the high-pressure cylinder extraction steam pressure, extraction steam temperature, extraction steam flow rate, high-temperature molten salt heat storage temperature, molten salt flow rate, and reheater outlet temperature; Taking the power generation coal consumption rate as the optimization target, a dynamic optimization regulation model of extraction steam energy storage is established. The objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate. The constraints include the upper and lower limits of the extraction steam flow rate, the molten salt flow rate, the extraction steam temperature, the molten salt heat storage temperature and the reheater outlet temperature. The established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of the extraction steam energy storage operating parameters under different working conditions, thus forming a dynamic optimization regulation strategy library for extraction steam energy storage. According to the grid load forecast results, the optimal regulation strategy is selected from the obtained dynamic optimization regulation strategy library, and the set values ​​of the high-pressure cylinder extraction pressure, extraction temperature and extraction flow are transmitted to the turbine extraction control system, and the set values ​​of the molten salt heat storage temperature and molten salt flow are transmitted to the high-temperature molten salt heat storage control system; the steam turbine extraction process and the molten salt heat storage process are coordinated to realize the flexible operation and efficient peak regulation of the thermal power generating units under the dynamic fluctuation of the grid load.

2. The method according to claim 1, characterized in that A thermal coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit is established. The thermal coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system, including: Establishing a thermodynamic model of a steam turbine high-pressure cylinder, the thermodynamic model comprising a steam flow calculation formula based on the Stodola ellipse law and an isentropic efficiency correction formula considering partial load performance changes; calculating the steam flow and the corrected isentropic efficiency of the steam turbine high-pressure cylinder according to the thermodynamic model of the steam turbine high-pressure cylinder; Establishing a heat transfer flow model of a high-temperature molten salt heat storage device, the heat transfer flow model of the high-temperature molten salt heat storage device includes a heat transfer equation of a heat exchanger based on a logarithmic mean temperature difference method and a modified Dittus-Boelter formula considering the flow characteristics of molten salt; using the heat transfer flow model of the high-temperature molten salt heat storage device, calculating the heat transfer power of the high-temperature molten salt heat storage device and the Nusselt number of the molten salt flow; Establishing a mass energy balance equation for the steam turbine extraction and heat storage system, the mass energy balance equation includes a mass balance equation for the extraction point, an energy balance equation, and an energy accumulation rate equation for the high-temperature molten salt heat storage system; based on the mass energy balance equation, calculating the steam flow rate and specific enthalpy at each point in the system and the energy change rate of the high-temperature molten salt heat storage system; The thermodynamic model of the steam turbine high-pressure cylinder, the heat transfer flow model of the high-temperature molten salt heat storage device and the mass energy balance equation of the steam turbine extraction and heat storage system are combined to construct a complete thermomechanical coupling mathematical model.

3. The method according to claim 1, characterized in that Taking the power generation coal consumption rate as the optimization target, a dynamic optimization regulation model of extraction steam energy storage is established. The objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate. The constraints include the upper and lower limits of the extraction steam flow rate, molten salt flow rate, extraction steam temperature, molten salt heat storage temperature and reheater outlet temperature. The established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of extraction steam energy storage operating parameters under different working conditions. The dynamic optimization regulation strategy library of extraction steam energy storage includes: Initializing a population to generate an initial population including decision variables, wherein the decision variables include a high-pressure cylinder extraction steam pressure, an extraction steam temperature, an extraction steam flow rate, a high-temperature molten salt heat storage temperature, a molten salt flow rate, and a reheater outlet temperature; The initial population is non-dominatedly sorted to divide the individuals into different non-dominated layers; uniformly distributed reference points are generated in the standardized target space; the distance from each individual to the nearest reference point is calculated, and individuals are selected to enter the next generation population according to the distance; the population diversity index is calculated, which is the average value of the Euclidean distances of all individuals to the population center; According to the population diversity index, adaptively adjusting the crossover rate and the mutation rate, wherein the crossover rate decreases as the population diversity index increases, and the mutation rate increases as the population diversity index increases, and the adjustment of the crossover rate and the mutation rate both adopts an exponential function form; performing crossover and mutation operations on the selected individuals to generate a progeny population; Some individuals are selected with a preset probability for simulated annealing optimization, new solutions are generated in the neighborhood of the selected individuals, the energy difference is calculated, the probability of accepting the new solution is calculated based on the energy difference and the current temperature, and the temperature is updated according to the preset cooling coefficient.

4. The method according to claim 3, characterized in that The method further comprises: Calculate the improved crowding, including calculating the target space crowding and the decision space crowding, and obtain the comprehensive crowding by weighted average, wherein the target space crowding and the decision space crowding consider the distances of adjacent individuals in the target space and the decision space respectively; The population is divided into multiple sub-populations, with the number of sub-populations equal to the number of available processor cores. For each sub-population, objective function evaluation, constraint processing, non-dominated sorting, and crowding calculation are performed in parallel on a separate processor core; the evaluation results of each sub-population are summarized, and global non-dominated sorting and selection operations are performed; the parent population and the child population are merged, and the merged population is non-dominated sorted, and the top N individuals are selected to form the next generation population; Determine whether the maximum number of iterations has been reached or the convergence condition has been met. If so, output the optimization result. If not, return to the step of calculating the distance from each individual to the nearest reference point. According to the output optimization result, determine the optimal operating parameters of the extraction steam storage system to achieve dynamic optimization control of the system.

5. The method according to claim 1, characterized in that According to the grid load forecast results, the optimal regulation strategy is selected from the obtained dynamic optimization regulation strategy library, and the set values ​​of the high-pressure cylinder extraction steam pressure, extraction steam temperature, and extraction steam flow are transmitted to the turbine extraction steam control system, and the set values ​​of the molten salt heat storage temperature and molten salt flow are transmitted to the high-temperature molten salt heat storage control system, including: A deep learning model of long short-term memory network and attention mechanism is used to predict power grid load, wherein the deep learning model includes an input layer, a long short-term memory network layer, an attention layer, a fully connected layer and an output layer, wherein the long short-term memory network layer is provided with 3 layers, each layer includes 128 neurons, and the fully connected layer includes 2 layers, respectively including 64 and 32 neurons; A dynamic optimization regulation strategy library is constructed using a dual deep Q network algorithm, the dual deep Q network algorithm includes an action value network and a target network, the action value network and the target network are both composed of three fully connected layers, the number of neurons is 256, 128 and 64 respectively, and the activation function is a ReLU function; the reward function of the dynamic optimization regulation strategy library takes into account the improvement of power generation efficiency, the improvement of peak load capacity, the equipment wear cost and the operating cost; Based on the grid load forecasting results, a fuzzy comprehensive evaluation method is used to select the optimal regulation strategy from the dynamic optimization regulation strategy library, wherein the fuzzy comprehensive evaluation method includes establishing a fuzzy relationship matrix and determining the index weights, wherein the index weights are determined by a hierarchical analysis method; The high-pressure cylinder extraction pressure set value, extraction temperature set value, and extraction flow set value in the optimal regulation strategy are transmitted to the turbine extraction control system through the OPC UA protocol, and the molten salt heat storage temperature set value and molten salt flow set value are transmitted to the high-temperature molten salt heat storage control system through the OPC UA protocol. The OPC UA protocol uses X.509 certificates for identity authentication and uses the AES-256 encryption algorithm to encrypt communication data.

6. The method according to claim 5, characterized in that The coordinated control of the steam turbine extraction process and the molten salt heat storage process enables flexible operation and efficient peak load regulation of thermal power generating units under dynamic fluctuations of power grid load, including: A model predictive control method is used to realize the coordinated control of the steam turbine extraction process and the molten salt heat storage process. The objective function of the model predictive control method includes an output tracking error term and a control increment penalty term, and the constraints include an extraction pressure range, an extraction temperature range, an extraction flow range, a molten salt temperature range, a molten salt flow range, and a control increment limit. An improved fast gradient method is used to solve the model predictive control optimization problem. The improved fast gradient method includes converting the original problem into a quadratic programming problem, using the Nesterov accelerated gradient method for iterative solution, and adopting a hot start strategy. A recursive least squares algorithm is introduced to update the system model parameters online. The forgetting factor of the recursive least squares algorithm is set to 0.98 to achieve adaptive update of the system model.

7. The method according to claim 6, characterized in that The improved fast gradient method is used to solve the model predictive control optimization problem. The improved fast gradient method includes converting the original problem into a quadratic programming problem, using the Nesterov accelerated gradient method for iterative solution, and adopting a hot start strategy including: The Nesterov accelerated gradient method includes the following sub-steps: Initialize the decision variable vector, auxiliary variable vector, step size and momentum parameter; calculate the gradient of the objective function; update the decision variable vector and auxiliary variable vector; project the updated decision variable vector into the feasible domain; dynamically adjust the step size using the Barzilai-Borwein method; dynamically adjust the momentum parameter according to the number of iterations; Implement the hot start strategy and use the optimal solution of the previous control cycle as the initial value of the current control cycle, including: Save the optimal solution of the previous control cycle; construct a new initial decision variable vector, where the first N-1 control increments come from the optimal solution of the previous cycle and the last control increment is set to zero; use a larger step size for warm-up iterations to quickly approach the optimal solution.

8. A steam extraction energy storage regulation optimization system using a high-temperature molten salt heat storage system is used to implement the method described in any one of claims 1 to 7, characterized in that: include: The first unit is used to establish a thermodynamic coupling mathematical model of the high-temperature molten salt heat storage system and the steam turbine unit, the thermodynamic coupling mathematical model includes the thermodynamic models of the high-pressure cylinder, the intermediate-pressure cylinder and the low-pressure cylinder of the steam turbine, the heat transfer flow model of the high-temperature molten salt heat storage device, and the mass energy balance equation of the steam turbine extraction and heat storage system; determine the key parameters that affect the extraction steam energy storage performance, the key parameters include the high-pressure cylinder extraction steam pressure, extraction steam temperature, extraction steam flow rate, high-temperature molten salt heat storage temperature, molten salt flow rate, and reheater outlet temperature; The second unit is used to establish a dynamic optimization regulation model of extraction steam energy storage with the power generation coal consumption rate as the optimization target, wherein the objective function of the dynamic optimization regulation model is to minimize the power generation coal consumption rate, and the constraints include the upper and lower limits of the extraction steam flow rate, the molten salt flow rate, the extraction steam temperature, the molten salt heat storage temperature and the reheater outlet temperature; the established dynamic optimization regulation model is solved by using an improved non-dominated sorting genetic algorithm to obtain the optimal solution set of the extraction steam energy storage operating parameters under different working conditions, thereby forming a dynamic optimization regulation strategy library of extraction steam energy storage; The third unit is used to select the optimal regulation strategy from the obtained dynamic optimization regulation strategy library according to the grid load forecast result, transmit the set values ​​of high-pressure cylinder extraction pressure, extraction temperature and extraction flow to the turbine extraction control system, and transmit the set values ​​of molten salt heat storage temperature and molten salt flow to the high-temperature molten salt heat storage control system; coordinately control the turbine extraction process and the molten salt heat storage process to realize flexible operation and efficient peak regulation of thermal power generating units under dynamic fluctuations of grid load.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method described in any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images