Steam extraction energy storage and fused salt heat storage method and system under deep peak regulation of power grid

Abstract

The invention provides a steam extraction energy storage and fused salt heat storage method and system under power grid deep peak regulation, and relates to the technical field of power grids, and the method comprises the steps: determining a power grid deep peak regulation period according to power grid load prediction data and renewable energy power generation prediction data; steam extraction operation is conducted on the steam turbine based on model predictive control, and extracted steam is introduced into the multi-stage steam extraction energy storage device; the steam passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger and exchanges heat with the fused salt in the fused salt storage tank; in the power utilization peak period of the power grid, the multi-stage steam extraction energy storage device and the fused salt heat storage device are controlled to release heat energy; a steam extraction energy storage and fused salt heat storage collaborative optimization model is established, and the steam extraction amount, the energy storage and release time and the fused salt heat storage capacity are optimized and adjusted according to power grid load characteristics, renewable energy source output characteristics and system equipment parameters; and through the collaborative optimization model, multi-target balance of power grid deep peak regulation, renewable energy consumption and system economy is realized.

Description

Technical Field

[0001] The present invention relates to power grid technology, and in particular to a method and system for extracting steam energy storage and molten salt heat storage under deep peak regulation of a power grid. Background Art

[0002] As the proportion of renewable energy in the power system continues to increase, the power grid faces severe peak load regulation challenges. The regulation capacity of traditional thermal power units is limited and it is difficult to meet the flexibility needs of large-scale renewable energy after grid connection. In order to achieve deep peak load regulation of the power grid, improve the absorption capacity of renewable energy, and ensure the economy of the system, it is necessary to explore new energy storage and flexible regulation technologies.

[0003] At present, the following technical means are mainly used to solve the problems of power grid peak regulation and renewable energy consumption:

[0004] Conventional pumped storage: By pumping water to a high reservoir when the power load is low, and releasing water to generate electricity when the load is peak, the time transfer of electricity is achieved. However, the site selection of pumped storage power stations is limited by geographical conditions, the construction period is long, and the investment cost is high.

[0005] Electrochemical energy storage: such as lithium-ion batteries and sodium-sulfur batteries, which have the advantages of fast response speed and high regulation accuracy. However, the cost of large-scale electrochemical energy storage systems is still high, and there are problems such as lifespan and safety.

[0006] Demand-side response: By encouraging users to actively adjust their electricity consumption behavior, flexible load regulation can be achieved. However, the scale and duration of demand-side response are limited, and it is difficult to meet the needs of deep peak regulation.

[0007] Flexibility transformation of conventional thermal power units: Improve the regulation capacity of thermal power units through technical transformation, such as reducing the minimum technical output, increasing the ramp rate, etc. However, this transformation often reduces the efficiency of the unit and increases operating costs.

[0008] Expansion of power system interconnection lines: By strengthening the interconnection of power grids, a wider range of power dispatching and balancing can be achieved. However, the construction of cross-regional transmission lines faces challenges such as large investment and long cycle.

[0009] In view of the shortcomings of the above technical solutions, researchers proposed the extraction steam energy storage technology, which uses the extraction steam of the thermal power unit to drive the compressor, compresses the air and stores it in the high-pressure gas tank, and then releases the air to drive the expander to generate electricity when needed. This technology can make full use of the equipment and sites of existing thermal power plants, and has the advantages of low investment cost and fast response speed. However, the simple extraction steam energy storage system still has problems such as low energy density and limited energy storage efficiency.

[0010] On the other hand, molten salt heat storage technology has been widely used in the field of solar thermal power generation, with the characteristics of large energy storage capacity and low cost. However, traditional molten salt heat storage systems are mainly used for day and night regulation and are difficult to adapt to the needs of deep peak regulation of the power grid. Summary of the invention

[0011] The embodiments of the present invention provide a method and system for extracting steam energy storage and molten salt heat storage under deep peak regulation of a power grid, which can solve the problems in the prior art.

[0012] According to a first aspect of the embodiments of the present invention,

[0013] Provides steam extraction energy storage and molten salt heat storage methods for deep peak regulation of power grid, including:

[0014] Obtaining grid load forecast data and renewable energy generation forecast data; determining a grid deep peak-shaving period according to the grid load forecast data and the renewable energy generation forecast data; performing steam extraction operation on a steam turbine based on model predictive control during the grid deep peak-shaving period, and introducing the extracted steam into a multi-stage steam extraction energy storage device;

[0015] The multi-stage steam extraction energy storage device comprises a high-pressure energy storage chamber, a medium-pressure energy storage chamber and a low-pressure energy storage chamber connected in sequence, and each energy storage chamber is provided with a phase change energy storage material; the extracted steam is passed through the high-pressure energy storage chamber, the medium-pressure energy storage chamber and the low-pressure energy storage chamber in sequence, and the thermal energy of the steam is used to cause the phase change energy storage material in each energy storage chamber to undergo a phase change, thereby realizing the storage of thermal energy; a molten salt heat storage device is provided at the outlet end of the multi-stage steam extraction energy storage device, and the molten salt heat storage device comprises a molten salt storage tank and a heat exchanger; the steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to perform heat exchange with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release thermal energy;

[0016] The released heat energy is used to heat feed water to generate high-temperature and high-pressure steam; the high-temperature and high-pressure steam is sent to a steam turbine to generate electricity to meet the peak load demand of the power grid; the changes in the power grid load are monitored in real time, and the charging and discharging processes of the multi-stage steam extraction energy storage device and the molten salt heat storage device are dynamically adjusted; a collaborative optimization model of steam extraction energy storage and molten salt heat storage is established, and the steam extraction amount, energy storage release time and molten salt heat storage capacity are optimized and adjusted according to the power grid load characteristics, renewable energy output characteristics and system equipment parameters; through the collaborative optimization model, a multi-objective balance of deep power grid peak regulation, renewable energy consumption and system economy is achieved.

[0017] In an optional embodiment,

[0018] During the deep peak load regulation period of the power grid, the steam turbine is extracted based on the model predictive control, and the extracted steam is introduced into the multi-stage steam extraction energy storage device, including:

[0019] A nonlinear state space model of a steam turbine-extraction system is established, wherein the nonlinear state space model includes a steam turbine model, an extraction system model and an energy storage device model; wherein:

[0020] The steam turbine model describes the relationship between state variables, control input, external disturbance and output variables; the steam extraction system model describes the relationship between the steam extraction amount and the flow coefficient, the effective flow area of ​​the steam extraction valve, the steam extraction pressure and the external pressure; the energy storage device model describes the relationship between the temperature change of the energy storage chamber and the inlet and outlet steam flow, specific enthalpy, heat loss, energy storage medium mass and specific heat capacity;

[0021] Based on the nonlinear state space model, a model predictive controller is designed, and the steam extraction operation of the steam turbine based on the model predictive control includes:

[0022] Define the prediction horizon and control horizon;

[0023] Constructing a predictive control objective function, wherein the predictive control objective function takes into account a power generation deviation, a steam extraction deviation, an energy storage efficiency deviation, and a control increment;

[0024] Set predictive control constraints, including extraction steam quantity constraint, extraction steam pressure constraint, extraction steam temperature constraint, load change rate constraint, extraction steam valve opening constraint and extraction steam valve opening change rate constraint;

[0025] A sequential quadratic programming algorithm is used to solve the model predictive control optimization problem to obtain an optimal control sequence, and steam extraction operation is performed on the steam turbine based on the optimal control sequence.

[0026] In an optional embodiment,

[0027] The method further comprises:

[0028] Linearizing the nonlinear state space model at the current working point; constructing a quadratic programming subproblem based on the linearized model; solving the quadratic programming subproblem using an interior point method to obtain a local optimal solution; updating the working point to obtain a global optimal control sequence;

[0029] Execute real-time control processes, including:

[0030] Collect steam turbine operation data, energy storage device status and short-term load forecast data; use extended Kalman filter to estimate the current state of the system based on the collected data; input the estimated current state and forecast information into the sequential quadratic programming algorithm to solve the model predictive control optimization problem and obtain the future control sequence;

[0031] Extract the first value of the future control sequence as the instruction of the current control cycle, and send the control instruction to the distributed control system through the OPC UA protocol; monitor key parameters in real time, including but not limited to vibration, temperature and pressure, execute multi-level safety protection logic, and realize rolling optimization control.

[0032] In an optional embodiment,

[0033] The steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to exchange heat with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release heat energy, including:

[0034] A system dynamic model is established, which includes a molten salt storage tank model, a heat exchanger model and a steam turbine model; wherein the molten salt storage tank model adopts a multi-node method to divide the tank into a plurality of equal-thickness nodes in the vertical direction, and an energy balance equation is established for each node; the heat exchanger model adopts a heat transfer unit number method, and introduces a time-varying scaling factor to characterize the change in thermal resistance of the pipe wall; the steam turbine model combines the Stodola elliptic law and the Willand equation to describe the characteristics of the steam turbine; the molten salt storage tank model, the heat exchanger model and the steam turbine model are connected through state variables and control variables to construct a state space expression of the overall system;

[0035] Based on the system dynamic model, a long short-term memory network is used in combination with multi-source data to predict load demand, and the multi-source data includes historical load data, meteorological data and calendar information; the long short-term memory network includes an input layer, two hidden layers and an output layer, and the dropout technology is used to prevent overfitting; multiple long short-term memory network models with different initialization parameters are trained, and an integrated learning method is used to improve the prediction stability, and the final prediction result is the median of the outputs of multiple models;

[0036] Based on the load demand forecasting result, an improved non-dominated sorting genetic algorithm III is used for multi-objective optimization; the first multi-objective function corresponding to the multi-objective optimization includes maximizing power generation efficiency, minimizing peak load response time and maximizing equipment life index; the improved non-dominated sorting genetic algorithm III includes using adaptive crossover and mutation operators, introducing local search operators, and using domination concepts to control the size of non-dominated solution sets; and using an ideal solution similarity sorting method to select the best compromise solution from the Pareto optimal solution set;

[0037] The multi-stage steam extraction storage device and the molten salt heat storage device are controlled to release thermal energy based on the optimal compromise solution.

[0038] In an optional embodiment,

[0039] A collaborative optimization model of steam extraction energy storage and molten salt heat storage is established. According to the load characteristics of the power grid, the output characteristics of renewable energy, and the parameters of system equipment, the steam extraction volume, energy storage release time, and molten salt heat storage capacity are optimized and adjusted. Through the collaborative optimization model, a multi-objective balance of deep peak regulation of the power grid, renewable energy consumption, and system economy is achieved, including:

[0040] Establishing a steam extraction energy storage subsystem model, the steam extraction energy storage subsystem model includes a steam turbine output power model, an energy balance equation for an energy storage process, and an energy balance equation for an energy release process; wherein the steam turbine output power model considers steam flow, steam enthalpy, steam turbine efficiency, and generator efficiency; the energy balance equation for an energy storage process and the energy balance equation for an energy release process consider storage energy, energy storage efficiency, energy release efficiency, storage power, and energy release power, respectively;

[0041] Establishing a molten salt heat storage subsystem model, the molten salt heat storage subsystem model includes a heat change equation for a hot storage tank and a heat change equation for a cold storage tank; wherein the heat change equation takes into account molten salt flow, specific heat capacity, inlet and outlet temperatures, heat loss coefficient, storage tank surface area and ambient temperature;

[0042] Establishing a power grid load forecasting model, wherein the power grid load forecasting model adopts a time series analysis method and considers historical load data, wind speed, temperature, humidity and date type;

[0043] Establishing a renewable energy output model, the renewable energy output model includes a wind power output model and a photovoltaic output model; wherein the wind power output model takes into account air density, wind rotor swept area, power coefficient and wind speed; the photovoltaic output model takes into account photoelectric conversion efficiency, photovoltaic panel area, solar radiation intensity and battery panel temperature;

[0044] Constructing a collaborative optimization model, the collaborative optimization model includes a second multi-objective function and constraints; wherein the second multi-objective function comprehensively considers load tracking error, renewable energy curtailment and system operation cost; the constraints include power balance constraints, extraction steam storage system constraints, molten salt thermal storage system constraints and renewable energy output constraints;

[0045] Design an improved multi-objective particle swarm optimization algorithm to solve the collaborative optimization model, and select three typical solutions based on the Pareto optimal solution set obtained, corresponding to the optimization solutions focusing on grid peak regulation, renewable energy consumption, and economy respectively;

[0046] According to the selected typical solution, the optimal values ​​of steam extraction volume, energy storage release time and molten salt heat storage capacity are determined to achieve a multi-objective balance among deep peak regulation of the power grid, renewable energy consumption and system economy.

[0047] In an optional embodiment,

[0048] The method further comprises:

[0049] The improved multi-objective particle swarm optimization algorithm includes the following improvements:

[0050] Adopting adaptive inertia weight, which decreases nonlinearly with the increase of iteration number; introducing chaotic perturbation to perturb the global optimal solution; adopting fast non-dominated sorting algorithm to classify particles; using crowding degree calculation method to maintain the diversity of solutions;

[0051] Executing the improved multi-objective particle swarm optimization algorithm comprises the following steps:

[0052] Initialize the particle swarm and set algorithm parameters;

[0053] Calculate the objective function value of each particle;

[0054] Perform non-dominated sorting and congestion calculations;

[0055] Update individual optimal solutions and global optimal solutions;

[0056] Update particle velocity and position;

[0057] Perform chaotic perturbation on the global optimal solution;

[0058] Determine whether the termination condition is met. If so, output the Pareto optimal solution set.

[0059] According to a second aspect of the embodiments of the present invention,

[0060] Provides steam extraction energy storage and molten salt heat storage system for deep peak regulation of power grid, including:

[0061] The first unit is used to obtain grid load forecast data and renewable energy power generation forecast data; determine a grid deep peak regulation period according to the grid load forecast data and the renewable energy power generation forecast data; and within the grid deep peak regulation period, extract steam from the steam turbine based on model predictive control to introduce the extracted steam into a multi-stage steam extraction energy storage device;

[0062] The second unit is used for the multi-stage steam extraction energy storage device, which includes a high-pressure energy storage chamber, a medium-pressure energy storage chamber and a low-pressure energy storage chamber connected in sequence, and each energy storage chamber is provided with a phase change energy storage material; the extracted steam passes through the high-pressure energy storage chamber, the medium-pressure energy storage chamber and the low-pressure energy storage chamber in sequence, and the thermal energy of the steam is used to cause the phase change energy storage material in each energy storage chamber to undergo a phase change, thereby realizing the storage of thermal energy; a molten salt heat storage device is provided at the outlet end of the multi-stage steam extraction energy storage device, and the molten salt heat storage device includes a molten salt storage tank and a heat exchanger; the steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to perform heat exchange with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release thermal energy;

[0063] The third unit is used to utilize the released heat energy to heat feed water to generate high-temperature and high-pressure steam; send the high-temperature and high-pressure steam into the steam turbine to generate electricity to meet the peak load demand of the power grid; monitor the changes in the power grid load in real time, and dynamically adjust the charging and discharging processes of the multi-stage steam extraction energy storage device and the molten salt heat storage device; establish a collaborative optimization model of steam extraction energy storage and molten salt heat storage, and optimize and adjust the steam extraction amount, energy storage release time and molten salt heat storage capacity according to the power grid load characteristics, renewable energy output characteristics and system equipment parameters; through the collaborative optimization model, a multi-objective balance of deep power grid peak regulation, renewable energy consumption and system economy is achieved.

[0064] According to a third aspect of the embodiments of the present invention,

[0065] An electronic device is provided, comprising:

[0066] processor;

[0067] a memory for storing processor-executable instructions;

[0068] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0069] According to a fourth aspect of the embodiments of the present invention,

[0070] 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.

[0071] This application achieves precise control of the steam turbine extraction process through sophisticated model processing, efficient optimization algorithms and perfect real-time control processes. This control strategy can not only meet the needs of deep peak regulation of the power grid, but also maximize the energy storage efficiency while ensuring system safety, providing strong support for the flexibility and economy of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1It is a flow chart of a method for extracting steam energy storage and molten salt heat storage under deep peak regulation of a power grid according to an embodiment of the present invention;

[0073] Figure 2 It is a structural schematic diagram of the steam extraction energy storage and molten salt heat storage system under deep peak regulation of the power grid according to an embodiment of the present invention. DETAILED DESCRIPTION

[0074] 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.

[0075] 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.

[0076] Figure 1 Schematic diagram of the process of steam extraction energy storage and molten salt heat storage method under deep peak regulation of power grid according to an embodiment of the present invention, as shown in FIG. Figure 1 As shown, the method includes:

[0077] S101. Obtain grid load forecast data and renewable energy generation forecast data; determine the grid deep peak-shaving period according to the grid load forecast data and the renewable energy generation forecast data; during the grid deep peak-shaving period, extract steam from the steam turbine based on model predictive control, and introduce the extracted steam into a multi-stage steam extraction energy storage device;

[0078] S102. The multi-stage steam extraction energy storage device comprises a high-pressure energy storage chamber, a medium-pressure energy storage chamber and a low-pressure energy storage chamber connected in sequence, and each energy storage chamber is provided with a phase change energy storage material; the extracted steam is passed through the high-pressure energy storage chamber, the medium-pressure energy storage chamber and the low-pressure energy storage chamber in sequence, and the thermal energy of the steam is used to cause the phase change energy storage material in each energy storage chamber to undergo a phase change, thereby realizing the storage of thermal energy; a molten salt heat storage device is provided at the outlet end of the multi-stage steam extraction energy storage device, and the molten salt heat storage device comprises a molten salt storage tank and a heat exchanger; the steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to perform heat exchange with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release thermal energy;

[0079] S103. Utilize the released heat energy to heat feed water to generate high-temperature and high-pressure steam; send the high-temperature and high-pressure steam into a steam turbine to generate electricity to meet the peak load demand of the power grid; monitor the changes in power grid load in real time, and dynamically adjust the charging and discharging processes of the multi-stage steam extraction energy storage device and the molten salt heat storage device; establish a collaborative optimization model of steam extraction energy storage and molten salt heat storage, and optimize and adjust the steam extraction volume, energy storage release time and molten salt heat storage capacity according to the power grid load characteristics, renewable energy output characteristics and system equipment parameters; through the collaborative optimization model, achieve a multi-objective balance of deep power grid peak regulation, renewable energy consumption and system economy.

[0080] In an optional embodiment,

[0081] During the deep peak load regulation period of the power grid, the steam turbine is extracted based on the model predictive control, and the extracted steam is introduced into the multi-stage steam extraction energy storage device, including:

[0082] A nonlinear state space model of a steam turbine-extraction system is established, wherein the nonlinear state space model includes a steam turbine model, an extraction system model and an energy storage device model; wherein:

[0083] The steam turbine model describes the relationship between state variables, control input, external disturbance and output variables; the steam extraction system model describes the relationship between the steam extraction amount and the flow coefficient, the effective flow area of ​​the steam extraction valve, the steam extraction pressure and the external pressure; the energy storage device model describes the relationship between the temperature change of the energy storage chamber and the inlet and outlet steam flow, specific enthalpy, heat loss, energy storage medium mass and specific heat capacity;

[0084] Based on the nonlinear state space model, a model predictive controller is designed, and the steam extraction operation of the steam turbine based on the model predictive control includes:

[0085] Define the prediction horizon and control horizon;

[0086] Constructing a predictive control objective function, wherein the predictive control objective function takes into account a power generation deviation, a steam extraction deviation, an energy storage efficiency deviation, and a control increment;

[0087] Set predictive control constraints, including extraction steam quantity constraint, extraction steam pressure constraint, extraction steam temperature constraint, load change rate constraint, extraction steam valve opening constraint and extraction steam valve opening change rate constraint;

[0088] A sequential quadratic programming algorithm is used to solve the model predictive control optimization problem to obtain an optimal control sequence, and steam extraction operation is performed on the steam turbine based on the optimal control sequence.

[0089] For example, during the deep peak load period of the power grid, steam extraction operation of the steam turbine based on model predictive control and introduction of the extracted steam into a multi-stage steam extraction energy storage device is an innovative energy management method. The core of this method is to establish an accurate system model and use model predictive control technology to achieve optimal steam extraction operation. The implementation process and technical details of this method are described in detail below.

[0090] Firstly, a nonlinear state space model of the steam turbine-extraction system is established. The model consists of three main parts: steam turbine model, extraction system model and energy storage device model.

[0091] The steam turbine model describes the dynamic characteristics of the steam turbine. The state variables include the pressure, temperature and speed of each stage of the steam turbine; the control input includes the main steam valve opening, the regulating stage nozzle opening and the extraction valve opening; the external disturbance includes the grid load change and the main steam parameter fluctuation; the output variables include the power generation and extraction parameters. The model uses mass balance, energy balance and momentum balance equations to describe the relationship between these variables.

[0092] The extraction system model focuses on describing the relationship between the extraction volume and related parameters. The extraction volume is mainly affected by the flow coefficient, the effective flow area of ​​the extraction valve, the extraction pressure and the external pressure. The flow coefficient is related to the Reynolds number and the opening of the extraction valve, which is usually obtained by table lookup or interpolation. The effective flow area is related to the geometry and opening of the extraction valve and can be determined experimentally. The difference between the extraction pressure and the external pressure is the key factor driving the extraction flow.

[0093] The accumulator model describes the transfer and conversion of energy during the accumulator process. The temperature change of the accumulator cavity is the core of the model, which is affected by factors such as the inlet and outlet steam flow, steam specific enthalpy, heat loss, mass of the accumulator medium and specific heat capacity. The inlet and outlet steam flow is provided by the extraction system model, and the steam specific enthalpy can be calculated from the steam properties. Heat loss takes into account the heat exchange between the accumulator and the environment, and is usually proportional to the surface area of ​​the accumulator cavity and the temperature difference. The mass and specific heat capacity of the accumulator medium determine the thermal capacity of the accumulator.

[0094] Based on the above nonlinear state space model, a model predictive controller is designed. The core idea of ​​model predictive control is to use the system model to predict the system behavior in the future at each sampling moment and solve the optimal control sequence.

[0095] First, define the prediction time domain and control time domain. The prediction time domain is usually selected as 3-5 times the time constant of the main dynamic characteristics of the system. For example, for a 300MW steam turbine, the prediction time domain can be selected as 30 minutes. The control time domain is determined according to the computing power of the controller and the system response speed, and is usually smaller than the prediction time domain, such as 10 minutes.

[0096] Then, the predictive control objective function is constructed. The objective function comprehensively considers multiple performance indicators: the power generation deviation reflects the load tracking capability, the steam extraction deviation reflects the precise control of the energy storage process, the energy storage efficiency deviation measures the effect of energy conversion, and the control increment is used to suppress the drastic changes of the control signal. These indicators form the objective function through weighted summation, and the weight coefficient can be adjusted according to actual needs.

[0097] Next, set the predictive control constraints. The extraction steam volume constraint ensures that the extraction steam volume is within the allowable range, such as 0-30% of the rated main steam flow. The extraction steam pressure constraint ensures that the extraction steam pressure meets the requirements of the energy storage device, usually 0.5-2MPa. The extraction steam temperature constraint ensures that the extraction steam temperature is suitable for the energy storage medium, such as the molten salt thermal storage system requires a temperature of not less than 270°C. The load change rate constraint ensures the safe and stable operation of the unit, such as the load change rate does not exceed 2% / minute. The extraction valve opening constraint and the opening change rate constraint take into account the physical limitations of the actuator, such as the valve opening 0-100%, and the opening change rate does not exceed 1% / second.

[0098] Finally, the sequential quadratic programming algorithm is used to solve the model predictive control optimization problem. The algorithm transforms the nonlinear optimization problem into a series of quadratic programming sub-problems and obtains the optimal control sequence through iterative solution. The key steps of the algorithm include linearization, constructing quadratic programming problems, solving and updating. In each sampling period, the optimization calculation is performed to obtain the optimal control sequence for a period of time in the future.

[0099] Based on the calculated optimal control sequence, the steam turbine is extracted. The control system uses the first control variable in the optimal control sequence as the input of the current control cycle to adjust the main steam valve opening, regulating stage nozzle opening and extraction valve opening of the steam turbine. At the same time, the monitoring system collects the operating data of the steam turbine and the energy storage device in real time to provide the latest status information for the optimization calculation of the next control cycle.

[0100] Through the above method, the precise control of steam extraction of steam turbines during the deep peak regulation period of the power grid can be achieved, and the power generation power, steam extraction volume and energy storage efficiency can be effectively balanced, providing strong support for the flexible regulation of the power grid. In practical applications, this method can be further optimized and adjusted according to specific unit parameters and operating requirements to achieve the best peak regulation effect.

[0101] In an optional embodiment,

[0102] The method further comprises:

[0103] Linearizing the nonlinear state space model at the current working point; constructing a quadratic programming subproblem based on the linearized model; solving the quadratic programming subproblem using an interior point method to obtain a local optimal solution; updating the working point to obtain a global optimal control sequence;

[0104] Execute real-time control processes, including:

[0105] Collect steam turbine operation data, energy storage device status and short-term load forecast data; use extended Kalman filter to estimate the current state of the system based on the collected data; input the estimated current state and forecast information into the sequential quadratic programming algorithm to solve the model predictive control optimization problem and obtain the future control sequence;

[0106] Extract the first value of the future control sequence as the instruction of the current control cycle, and send the control instruction to the distributed control system through the OPC UA protocol; monitor key parameters in real time, including but not limited to vibration, temperature and pressure, execute multi-level safety protection logic, and realize rolling optimization control.

[0107] For example, based on the model predictive control, in order to further improve the control accuracy and real-time performance, this method introduces a more sophisticated optimization algorithm and real-time control process. The implementation process and specific details of these technical contents are described in detail below.

[0108] First, the nonlinear state space model is linearized. Near the current operating point, the nonlinear model is approximated using Taylor expansion. Taking the steam turbine model as an example, it is assumed that the main steam pressure at the current operating point is 16.7MPa, the temperature is 538℃, and the power generation is 280MW. Near this operating point, a linearized model is established to simplify the complex nonlinear relationship into a set of linear equations. Although this linearization process will introduce certain errors, it can significantly reduce the computational complexity and provide the possibility for real-time control.

[0109] Based on the linearized model, a quadratic programming sub-problem is constructed. The nonlinear terms in the objective function are approximated as quadratic forms, and the constraints are converted into linear inequalities or equations. For example, the power generation deviation term can be expressed as the sum of squares of the deviations, and the steam extraction constraint can be expressed as a linear inequality. In this way, the originally complex nonlinear optimization problem is converted into a standard quadratic programming problem, which can be solved using mature optimization algorithms.

[0110] Use the interior point method to solve the quadratic programming subproblem. The interior point method transforms the constrained optimization problem into an unconstrained problem by introducing a barrier function. The algorithm starts from a strictly feasible interior point and gradually approaches the optimal solution through iteration. In each iteration, the barrier parameter is gradually reduced and eventually converges to the optimal solution of the original problem. For medium-sized problems, such as optimization problems with 50 decision variables and 100 constraints, the interior point method can usually converge to a high-precision solution within 100 milliseconds.

[0111] Update the working point to obtain the global optimal control sequence. Since a local linearization model is used, multiple iterations are required to approach the global optimal solution. After each quadratic programming subproblem is solved, the local optimal solution is used as the new working point and linearized and optimized again. After 5-10 iterations, the global optimal control sequence with satisfactory accuracy can usually be obtained.

[0112] Next, the real-time control process is executed. First, the turbine operation data, energy storage device status and short-term load forecast data are collected. The turbine operation data includes pressure, temperature, flow rate, etc. at each level, and the sampling period is 100 milliseconds. The energy storage device status includes energy storage chamber temperature, pressure, storage capacity, etc., and the sampling period is 1 second. The short-term load forecast data covers the next 4 hours, and the forecast accuracy is above 98%.

[0113] Based on the collected data, the extended Kalman filter is used to estimate the current state of the system. The extended Kalman filter obtains the optimal estimate of the system state by fusing the model prediction value and the measurement value through two steps of prediction and update. For the steam turbine system, the root mean square error of the state estimation can usually be controlled within 1%. This state estimation not only improves the control accuracy, but also effectively suppresses the influence of measurement noise.

[0114] The estimated current state and forecast information are input into the sequential quadratic programming algorithm to solve the model predictive control optimization problem and obtain the future control sequence. The optimization calculation takes into account the system behavior for the next 30 minutes, but only generates the control sequence for the next 10 minutes. The optimization objectives include minimizing the load tracking error, maximizing the energy storage efficiency, and minimizing the control change rate. By adjusting the weight coefficient, a balance can be achieved between different objectives.

[0115] From the optimized future control sequence, the first value is extracted as the instruction for the current control cycle. This control instruction includes the main steam valve opening, the regulating stage nozzle opening, and the extraction valve opening. Then, the control instruction is sent to the distributed control system (DCS) through the OPC UA protocol. The OPC UA protocol has high security and cross-platform characteristics, which can ensure the reliable transmission of control instructions.

[0116] While executing control instructions, key parameters are monitored in real time. These parameters include but are not limited to vibration, temperature and pressure. For example, when monitoring bearing vibration, the alarm value is 7.1mm / s and the trip value is 11.2mm / s; when monitoring steam turbine exhaust temperature, the alarm value is 60℃ and the trip value is 65℃; when monitoring main steam pressure, the allowable fluctuation range is ±5% of the rated value.

[0117] Based on real-time monitoring data, multi-level safety protection logic is executed. Safety protection is divided into three levels: the first level is early warning, which issues a prompt when the parameter approaches the alarm value; the second level is automatic adjustment, which automatically adjusts the control strategy when the parameter exceeds the alarm value but does not reach the trip value; the third level is emergency protection, which immediately executes the emergency shutdown procedure when the parameter reaches the trip value. This multi-level safety protection mechanism can ensure the safety of the system while maintaining the continuous operation of the system to the maximum extent.

[0118] Through the above real-time control process, rolling optimization control is realized. Each control cycle (usually 5 seconds) repeats data collection, state estimation, optimization calculation and instruction issuance. This rolling optimization method can respond to system state changes and external disturbances in a timely manner, ensuring the real-time and robustness of control.

[0119] In summary, this method achieves precise control of the steam turbine extraction process through sophisticated model processing, efficient optimization algorithm and perfect real-time control process. This control strategy can not only meet the needs of deep peak regulation of the power grid, but also maximize the energy storage efficiency while ensuring system safety, providing strong support for the flexibility and economy of the power system.

[0120] In an optional embodiment,

[0121] The steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to exchange heat with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release heat energy, including:

[0122] A system dynamic model is established, which includes a molten salt storage tank model, a heat exchanger model and a steam turbine model; wherein the molten salt storage tank model adopts a multi-node method to divide the tank into a plurality of equal-thickness nodes in the vertical direction, and an energy balance equation is established for each node; the heat exchanger model adopts a heat transfer unit number method, and introduces a time-varying scaling factor to characterize the change in thermal resistance of the pipe wall; the steam turbine model combines the Stodola elliptic law and the Willand equation to describe the characteristics of the steam turbine; the molten salt storage tank model, the heat exchanger model and the steam turbine model are connected through state variables and control variables to construct a state space expression of the overall system;

[0123] Based on the system dynamic model, a long short-term memory network is used in combination with multi-source data to predict load demand, and the multi-source data includes historical load data, meteorological data and calendar information; the long short-term memory network includes an input layer, two hidden layers and an output layer, and the dropout technology is used to prevent overfitting; multiple long short-term memory network models with different initialization parameters are trained, and an integrated learning method is used to improve the prediction stability, and the final prediction result is the median of the outputs of multiple models;

[0124] Based on the load demand forecasting result, an improved non-dominated sorting genetic algorithm III is used for multi-objective optimization; the first multi-objective function corresponding to the multi-objective optimization includes maximizing power generation efficiency, minimizing peak load response time and maximizing equipment life index; the improved non-dominated sorting genetic algorithm III includes using adaptive crossover and mutation operators, introducing local search operators, and using domination concepts to control the size of non-dominated solution sets; and using an ideal solution similarity sorting method to select the best compromise solution from the Pareto optimal solution set;

[0125] The multi-stage steam extraction storage device and the molten salt heat storage device are controlled to release thermal energy based on the optimal compromise solution.

[0126] For example, during the peak period of power consumption in the power grid, effectively controlling the release of heat energy from the multi-stage steam extraction storage device and the molten salt heat storage device is the key to ensuring the stable operation of the power grid. This method achieves fine control of the heat release process by establishing an accurate system dynamic model, making accurate load demand forecasts, and using a multi-objective optimization algorithm. The implementation process and technical details of this method are described in detail below.

[0127] First, a system dynamic model is established. The model includes a molten salt storage tank model, a heat exchanger model, and a steam turbine model. These three sub-models are interconnected through state variables and control variables, and together constitute the state space expression of the overall system.

[0128] The molten salt storage tank model adopts a multi-node method. Taking a molten salt storage tank with a capacity of 1,000 tons as an example, the tank is divided into 20 nodes of equal thickness in the vertical direction, and the thickness of each node is about 0.5 meters. An energy balance equation is established for each node, considering heat conduction, convection heat transfer and heat loss. Heat conduction between nodes is related to temperature gradient and thermal conductivity, convection heat transfer is related to flow rate and specific heat capacity, and heat loss is proportional to the difference between node temperature and ambient temperature. By solving the energy balance equations of these 20 nodes, the temperature distribution and heat storage inside the tank can be obtained.

[0129] The heat exchanger model adopts the heat transfer unit number method and introduces a time-varying scaling factor. Taking the shell and tube heat exchanger as an example, it is divided into 50 heat transfer units. The heat transfer efficiency of each heat transfer unit is related to the heat capacity ratio and the number of heat transfer units. The time-varying scaling factor is used to characterize the change in the thermal resistance of the tube wall, and the initial value is set to 0.0001m 2 K / W increases logarithmically with the operating time, increasing by 10% for every 1000 hours of operation. This method can accurately describe the degradation of heat exchanger performance over time.

[0130] The turbine model combines the Stodola ellipse law and the Willeland equation to describe the turbine characteristics. The Stodola ellipse law is used to calculate the flow characteristics of each stage of the turbine, while the Willeland equation is used to estimate the efficiency of the turbine. Taking a 300MW subcritical turbine as an example, the nominal flow coefficients of the high-pressure cylinder, the intermediate-pressure cylinder, and the low-pressure cylinder are 0.02, 0.05, and 0.15, respectively, and the benchmark efficiency is 88%. Through these two equations, the actual power output and efficiency of the turbine can be calculated based on the inlet steam parameters and back pressure.

[0131] The above three sub-models are connected through common state variables (such as temperature, pressure, flow) and control variables (such as valve opening, pump speed) to construct the state space expression of the overall system. This expression contains about 100 state variables and 10 control variables, which can fully describe the dynamic characteristics of the system.

[0132] Next, based on the system dynamic model, the load demand is predicted by using a long short-term memory network combined with multi-source data, including historical load data for the past 30 days (15-minute sampling interval), weather forecast data for the next 24 hours (including temperature, humidity, wind speed), and calendar information (weekdays / weekends / holidays).

[0133] The structure of the LSTM network includes an input layer (the number of nodes is the number of features, about 20), two hidden layers (128 neurons in each layer) and an output layer (96 nodes for predicting the load in the next 24 hours). The dropout technology is used in the hidden layer, and the dropout rate is set to 0.2 to effectively prevent overfitting. The Adam optimizer is used in the training process, the initial value of the learning rate is 0.001, and the cosine annealing strategy is used for dynamic adjustment.

[0134] To improve the prediction stability, 10 LSTM network models with different initialization parameters were trained. These models used different random seeds to initialize weights and were trained on slightly different data subsets. The outputs of these 10 models were integrated using an ensemble learning method, and the final prediction result was taken as the median of the outputs of multiple models. This method can effectively reduce the random error of a single model and improve the robustness of the prediction. In practical applications, the mean absolute percentage error (MAPE) of this prediction method can be controlled within 3%.

[0135] Based on the load demand forecast results, an improved non-dominated sorting genetic algorithm III is used for multi-objective optimization. The first multi-objective function includes three sub-objectives: maximizing power generation efficiency, minimizing peak load response time, and maximizing equipment life index. The power generation efficiency target takes into account the efficiency of the turbine and the efficiency of the heat exchanger. The peak load response time target reflects the response speed of the system to load changes. The equipment life index is related to the number of temperature cycles and the amplitude of stress changes.

[0136] The improved non-dominated sorting genetic algorithm III contains the following key improvements:

[0137] 1. Adopt adaptive crossover and mutation operators. The initial value of crossover probability is set to 0.9, and the initial value of mutation probability is set to 0.1, which are dynamically adjusted according to population diversity. When population diversity decreases, the mutation probability is increased to improve exploration ability.

[0138] 2. Introduce the local search operator. Perform local search on the solutions of the Pareto front every 10 generations, and use the simulated annealing algorithm to find a better solution in the neighborhood of the solution.

[0139] 3. Use the concept of domination to control the size of the non-dominated solution set. Set the maximum number of non-dominated solutions to 100. When this number is exceeded, use crowding sorting to delete the redundant solutions.

[0140] The main parameters of the algorithm are set as follows: the population size is 200, the maximum number of iterations is 500, and the crossover and mutation operators use SBX and polynomial mutation. In each iteration, non-dominated sorting is performed first, then the crowding distance is calculated, and finally the environment selection is performed.

[0141] After the optimization is completed, a set of Pareto optimal solutions is obtained. In order to select the best compromise solution from these optimal solutions, the ideal solution similarity sorting method is used. First, the ideal point and the anti-ideal point are determined, and then the Euclidean distance of each solution to the ideal point and the Euclidean distance to the anti-ideal point are calculated. Finally, they are sorted according to relative proximity, and the solution with the largest relative proximity is selected as the best compromise solution.

[0142] The multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release heat energy based on the optimal compromise solution. The control strategy includes adjusting the opening of the steam extraction valve, controlling the speed of the molten salt circulation pump, and adjusting the bypass valve of the heat exchanger. For example, when the power generation needs to be increased quickly, the opening of the steam extraction valve can be increased, the speed of the molten salt circulation pump can be increased, and the opening of the heat exchanger bypass valve can be reduced to maximize the release of heat energy. Conversely, when the power generation needs to be reduced, the opposite operation is taken.

[0143] Through this sophisticated control strategy, the heat release process can be precisely regulated to meet the peak load demand of the power grid while ensuring the efficient operation of the system and the long-term reliability of the equipment. In practical applications, this method can control the peak load response time to less than 5 minutes, increase the power generation efficiency by 2-3 percentage points, and significantly extend the service life of key equipment.

[0144] In an optional embodiment,

[0145] A collaborative optimization model of steam extraction energy storage and molten salt heat storage is established. According to the load characteristics of the power grid, the output characteristics of renewable energy, and the parameters of system equipment, the steam extraction volume, energy storage release time, and molten salt heat storage capacity are optimized and adjusted. Through the collaborative optimization model, a multi-objective balance of deep peak regulation of the power grid, renewable energy consumption, and system economy is achieved, including:

[0146] Establishing a steam extraction energy storage subsystem model, the steam extraction energy storage subsystem model includes a steam turbine output power model, an energy balance equation for an energy storage process, and an energy balance equation for an energy release process; wherein the steam turbine output power model considers steam flow, steam enthalpy, steam turbine efficiency, and generator efficiency; the energy balance equation for an energy storage process and the energy balance equation for an energy release process consider storage energy, energy storage efficiency, energy release efficiency, storage power, and energy release power, respectively;

[0147] Establishing a molten salt heat storage subsystem model, the molten salt heat storage subsystem model includes a heat change equation for a hot storage tank and a heat change equation for a cold storage tank; wherein the heat change equation takes into account molten salt flow, specific heat capacity, inlet and outlet temperatures, heat loss coefficient, storage tank surface area and ambient temperature;

[0148] Establishing a power grid load forecasting model, wherein the power grid load forecasting model adopts a time series analysis method and considers historical load data, wind speed, temperature, humidity and date type;

[0149] Establishing a renewable energy output model, the renewable energy output model includes a wind power output model and a photovoltaic output model; wherein the wind power output model takes into account air density, wind rotor swept area, power coefficient and wind speed; the photovoltaic output model takes into account photoelectric conversion efficiency, photovoltaic panel area, solar radiation intensity and battery panel temperature;

[0150] Constructing a collaborative optimization model, the collaborative optimization model includes a second multi-objective function and constraints; wherein the second multi-objective function comprehensively considers load tracking error, renewable energy curtailment and system operation cost; the constraints include power balance constraints, extraction steam storage system constraints, molten salt thermal storage system constraints and renewable energy output constraints;

[0151] Design an improved multi-objective particle swarm optimization algorithm to solve the collaborative optimization model, and select three typical solutions based on the Pareto optimal solution set obtained, corresponding to the optimization solutions focusing on grid peak regulation, renewable energy consumption, and economy respectively;

[0152] According to the selected typical solution, the optimal values ​​of steam extraction volume, energy storage release time and molten salt heat storage capacity are determined to achieve a multi-objective balance among deep peak regulation of the power grid, renewable energy consumption and system economy.

[0153] For example, establishing a collaborative optimization model of steam extraction energy storage and molten salt heat storage is the key to achieving a multi-objective balance of deep peak regulation of the power grid, renewable energy consumption and system economy. The model comprehensively considers the load characteristics of the power grid, the output characteristics of renewable energy and the parameters of the system equipment, and optimizes the steam extraction volume, energy storage release time and molten salt heat storage capacity to improve the overall performance of the system. The establishment process and key technical points of the model are described in detail below.

[0154] First, the extraction steam storage subsystem model is established. The model includes the turbine output power model, the energy balance equation of the energy storage process and the energy balance equation of the energy release process. The turbine output power model takes into account the steam flow rate, steam enthalpy, turbine efficiency and generator efficiency. Taking a 300MW steam turbine as an example, the rated steam flow rate is 900t / h, the main steam pressure is 16.7MPa, the temperature is 538℃, the turbine efficiency is 88%, and the generator efficiency is 98.5%. The energy balance equation of the energy storage process takes into account the storage energy, energy storage efficiency and energy storage power, among which the energy storage efficiency can reach 85%. The energy balance equation of the energy release process takes into account the energy release efficiency and energy release power, and the energy release efficiency is about 80%. Through these equations, the dynamic characteristics of the extraction steam storage system can be accurately described.

[0155] Secondly, a molten salt heat storage subsystem model is established. The model includes the heat change equation for the hot storage tank and the heat change equation for the cold storage tank. Taking a molten salt storage tank with a capacity of 1,000 tons as an example, the heat change equation takes into account the molten salt flow rate (maximum flow rate 100kg / s), specific heat capacity (about 1.5kJ / kg·K), inlet and outlet temperatures (565℃ for hot storage tank and 290℃ for cold storage tank), heat loss coefficient (0.5W / m 2 K), tank surface area (about 300m 2 ) and the ambient temperature (assumed to be 25°C). Through these equations, the temperature change and heat storage of the molten salt in the tank can be accurately calculated.

[0156] Next, a grid load forecasting model is established. The model uses a time series analysis method, taking into account historical load data, wind speed, temperature, humidity, and date type. Specifically, historical load data for the past 30 days (15-minute sampling interval), weather forecast data for the next 24 hours (including wind speed, temperature, humidity), and date type information (weekdays / weekends / holidays) are used as input. The ARIMA (Autoregressive Integrated Moving Average) model is combined with a machine learning algorithm (such as random forest) for load forecasting. The mean absolute percentage error (MAPE) of the model can be controlled within 2.5%.

[0157] Then, a renewable energy output model is established, including a wind power output model and a photovoltaic output model. The wind power output model takes into account the air density (1.225 kg / m3 ), wind rotor swept area (taking a 2MW wind turbine as an example, about 6400m 2 ), power factor (maximum value is about 0.45) and wind speed. The photovoltaic output model takes into account the photovoltaic conversion efficiency (monocrystalline silicon cell is about 20%), photovoltaic panel area (taking a 1MW power station as an example, about 6000m 2 ), solar radiation intensity and panel temperature. These two models can accurately predict the output of renewable energy based on meteorological conditions.

[0158] Based on the above sub-models, a collaborative optimization model is constructed. The model includes multi-objective functions and constraints. The multi-objective function comprehensively considers load tracking error, renewable energy curtailment and system operation cost. Load tracking error reflects the system's ability to respond to changes in grid load, renewable energy curtailment reflects the system's ability to absorb renewable energy, and system operation cost includes fuel cost, equipment maintenance cost and carbon emission cost. Constraints include power balance constraints (ensuring that power generation is balanced with load demand), steam extraction storage system constraints (such as steam extraction volume limit, energy storage capacity limit), molten salt thermal storage system constraints (such as tank capacity limit, temperature limit) and renewable energy output constraints (such as wind speed limit for wind turbine cut-in and cut-out).

[0159] In order to solve the collaborative optimization model, an improved multi-objective particle swarm optimization algorithm was designed. The main parameters of the algorithm are set as follows: the population size is 200, the maximum number of iterations is 1000, the inertia weight decreases linearly from 0.9 to 0.4, and the learning factors c1 and c2 are both 2. The improvements of the algorithm include: 1) using an adaptive mutation strategy, the mutation probability is dynamically adjusted according to the population diversity, and the initial value is 0.1; 2) introducing a crowding mechanism to maintain the diversity of solutions; 3) using an external archive to store non-dominated solutions, and the archive size is set to 100.

[0160] Based on the Pareto optimal solution set obtained, three typical solutions are selected, corresponding to the optimization schemes focusing on grid peak regulation, renewable energy consumption and economy. Taking a regional power grid as an example, it is assumed that the system includes 1000MW thermal power units, 500MW wind power and 300MW photovoltaic power. For the scheme focusing on grid peak regulation, the optimization results may show: the steam extraction volume is dynamically adjusted between 100-150t / h, the energy storage time is 1-5 am, the energy release time is 18-22 o'clock, and the molten salt heat storage capacity is 2000MWh. This configuration can achieve a peak regulation depth of more than 15%, significantly improving the peak regulation capacity of the system.

[0161] For the scheme focusing on renewable energy consumption, the optimization results may show that the steam extraction volume fluctuates between 80-120t / h, the energy storage time is dynamically adjusted with the wind power and photovoltaic output, and the molten salt heat storage capacity is increased to 2500MWh. This configuration can increase the renewable energy consumption rate to more than 95%, effectively reducing the phenomenon of wind and solar power abandonment.

[0162] For the economic-oriented solution, the optimization results may show that the steam extraction volume is controlled between 60-100t / h, the energy storage and release time are flexibly adjusted according to the fluctuation of electricity prices, and the molten salt heat storage capacity is 1800MWh. This configuration can reduce the annual operating cost by about 5% while ensuring the stable operation of the system.

[0163] By analyzing and weighing these three typical solutions, the final optimization solution can be determined. For example, a compromise solution can be selected: the steam extraction volume is 90-130t / h, the energy storage time is 2-6 am and 11-13 noon (corresponding to the photovoltaic peak), the energy release time is 18-23, and the molten salt heat storage capacity is 2200MWh. This solution can achieve a peak load depth of 12%, a renewable energy consumption rate of 93%, and a 3% reduction in annual operating costs.

[0164] By implementing this optimization plan, the three goals of deep peak regulation of the power grid, renewable energy consumption and system economy can be effectively balanced. This plan not only improves the flexibility of the power grid and the utilization rate of renewable energy, but also reduces the operating cost of the system through reasonable energy storage and release strategies, providing strong support for the efficient, clean and economical operation of the power system.

[0165] In an optional embodiment,

[0166] The method further comprises:

[0167] The improved multi-objective particle swarm optimization algorithm includes the following improvements:

[0168] Adopting adaptive inertia weight, which decreases nonlinearly with the increase of iteration number; introducing chaotic perturbation to perturb the global optimal solution; adopting fast non-dominated sorting algorithm to classify particles; using crowding degree calculation method to maintain the diversity of solutions;

[0169] Executing the improved multi-objective particle swarm optimization algorithm comprises the following steps:

[0170] Initialize the particle swarm and set algorithm parameters;

[0171] Calculate the objective function value of each particle;

[0172] Perform non-dominated sorting and congestion calculations;

[0173] Update individual optimal solutions and global optimal solutions;

[0174] Update particle velocity and position;

[0175] Perform chaotic perturbation on the global optimal solution;

[0176] Determine whether the termination condition is met. If so, output the Pareto optimal solution set.

[0177] For example, the improved multi-objective particle swarm optimization algorithm is the key to solving the collaborative optimization model of extraction steam energy storage and molten salt thermal storage. The algorithm has significantly improved the solution efficiency and solution quality through a number of innovative improvements. The following details the algorithm's improvements and execution steps.

[0178] First, an adaptive inertia weight strategy is adopted. The traditional linear decreasing inertia weight cannot balance the global search and local search capabilities well. The improved algorithm adopts a nonlinear decreasing adaptive inertia weight, and the weight value decreases nonlinearly with the increase of the number of iterations. Specifically, in the early stage of iteration, the inertia weight maintains a large value (such as 0.9) to promote global search; as the iteration proceeds, the weight value gradually decreases, but the speed of decrease gradually slows down, and finally stabilizes in the later stage of iteration (such as 0.4). This strategy can better balance exploration and development in the search process and improve the convergence performance of the algorithm.

[0179] Secondly, a chaotic perturbation mechanism is introduced. In order to prevent the algorithm from falling into the local optimum, the global optimal solution is chaotically perturbed. Logistic mapping is used as a chaotic sequence generator to generate pseudo-random numbers ranging from 0 to 1. After each iteration, each dimension of the global optimal solution is perturbed with a certain probability (such as 10%). The perturbation amplitude decreases with the number of iterations to ensure the stability of the algorithm in the later stage. This mechanism effectively enhances the algorithm's ability to jump out of the local optimum and improves the diversity of search.

[0180] Third, a fast non-dominated sorting algorithm is used. This algorithm can efficiently classify the solutions in the population and provide a basis for subsequent selection operations. In the specific implementation, the dominating solution set and the dominated number of each solution are first calculated. Then, the solutions with a dominated number of 0 are classified as the first frontier and temporarily removed from the population. This process is repeated until all solutions are classified. The time complexity of this method is O(MN 2 ), where M is the number of objective functions and N is the population size, which greatly improves the efficiency compared with traditional methods.

[0181] Fourth, the crowding calculation method is used to maintain the diversity of solutions. Within each frontier, the crowding of each solution is calculated. The crowding reflects the distribution density of the solution in the target space. When calculating, each objective function is first normalized, and then the distance between adjacent solutions is calculated. The crowding of the boundary solutions is set to infinity to ensure that they are retained. This method helps maintain the diversity of the solution set while maintaining convergence and avoids excessive concentration of solutions in certain areas.

[0182] Based on the above improvements, the specific steps of executing the multi-objective particle swarm optimization algorithm are as follows:

[0183] Initialize the particle swarm and set the algorithm parameters. Take the problem of solving the coordinated optimization of steam extraction energy storage and molten salt heat storage as an example. Assume that the decision variables include 24-hour steam extraction, energy storage release time, and molten salt heat storage capacity, a total of 26 variables. Initialize a population of 200 particles, each particle represents a possible solution. Set the maximum number of iterations to 1000, the initial inertia weight to 0.9, and the learning factors c1 and c2 to 2.

[0184] Calculate the objective function value of each particle. For each particle, calculate the values ​​of the three objective functions: load tracking error, renewable energy curtailment, and system operating cost. For example, for a particle, the calculated 24-hour average load tracking error is 2%, the renewable energy curtailment is 50MWh, and the system daily operating cost is 1 million yuan.

[0185] Perform non-dominated sorting and crowding calculation. Use the fast non-dominated sorting algorithm to sort the 200 particles and obtain multiple non-dominated frontiers. Then calculate the crowding of particles in each frontier. Assume that the first frontier contains 50 solutions, calculate the crowding of these 50 solutions and sort them in descending order of crowding.

[0186] Update the individual optimal solution and the global optimal solution. For each particle, if the current position dominates its individual optimal solution, update the individual optimal solution. Select the solution with the highest congestion from the first frontier as the global optimal solution. If there are multiple non-dominated solutions, you can randomly select one as the global optimal solution.

[0187] Update particle speed and position. Update the speed and position of each particle according to the individual optimal solution and the global optimal solution. Use adaptive inertia weight when updating speed, and the weight value is dynamically adjusted according to the current number of iterations. After the position is updated, check whether it exceeds the boundary. If so, pull it back to the boundary.

[0188] Perform chaotic perturbation on the global optimal solution. The global optimal solution is perturbed with a probability of 10%. For example, the steam extraction volume is randomly perturbed by ±5%, the energy storage release time is randomly adjusted by ±30 minutes, and the molten salt heat storage capacity is randomly changed by ±100MWh. The perturbation amplitude decreases with the number of iterations.

[0189] Determine whether the termination condition is met. Check whether the maximum number of iterations (1000) has been reached or there is no significant improvement in the Pareto frontier after 50 consecutive iterations. If the termination condition is met, output the Pareto optimal solution set; otherwise, return to the second step to continue iterating.

[0190] By executing the above-mentioned improved multi-objective particle swarm optimization algorithm, the collaborative optimization problem of extraction steam energy storage and molten salt thermal storage can be effectively solved. For example, in a certain optimization, the algorithm converged at the 876th iteration and obtained a Pareto optimal solution set containing 100 non-dominated solutions. These solutions correspond to different trade-offs, such as a solution focusing on peak load regulation (average load tracking error of 1.5%, renewable energy abandonment of 100MWh, and system daily operating cost of 1.1 million yuan), a solution focusing on renewable energy consumption (average load tracking error of 2%, renewable energy abandonment of 30MWh, and system daily operating cost of 1.15 million yuan) and a solution focusing on economy (average load tracking error of 2.5%, renewable energy abandonment of 80MWh, and system daily operating cost of 950,000 yuan).

[0191] This improved multi-objective particle swarm optimization algorithm not only improves the solution efficiency, but also obtains a more diverse and high-quality Pareto optimal solution set, providing decision makers with a wealth of options. By analyzing the characteristics and performance of these options, the most suitable operation strategy can be selected according to actual needs to achieve a multi-objective balance between deep peak regulation of the power grid, renewable energy consumption and system economy.

[0192] Figure 2 Schematic diagram of the structure of the steam extraction energy storage and molten salt heat storage system under deep peak regulation of the power grid according to an embodiment of the present invention. Figure 2 As shown, the system comprises:

[0193] The first unit is used to obtain grid load forecast data and renewable energy power generation forecast data; determine a grid deep peak regulation period according to the grid load forecast data and the renewable energy power generation forecast data; and within the grid deep peak regulation period, extract steam from the steam turbine based on model predictive control to introduce the extracted steam into a multi-stage steam extraction energy storage device;

[0194] The second unit is used for the multi-stage steam extraction energy storage device, which includes a high-pressure energy storage chamber, a medium-pressure energy storage chamber and a low-pressure energy storage chamber connected in sequence, and each energy storage chamber is provided with a phase change energy storage material; the extracted steam passes through the high-pressure energy storage chamber, the medium-pressure energy storage chamber and the low-pressure energy storage chamber in sequence, and the thermal energy of the steam is used to cause the phase change energy storage material in each energy storage chamber to undergo a phase change, thereby realizing the storage of thermal energy; a molten salt heat storage device is provided at the outlet end of the multi-stage steam extraction energy storage device, and the molten salt heat storage device includes a molten salt storage tank and a heat exchanger; the steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to perform heat exchange with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release thermal energy;

[0195] The third unit is used to utilize the released heat energy to heat feed water to generate high-temperature and high-pressure steam; send the high-temperature and high-pressure steam into the steam turbine to generate electricity to meet the peak load demand of the power grid; monitor the changes in the power grid load in real time, and dynamically adjust the charging and discharging processes of the multi-stage steam extraction energy storage device and the molten salt heat storage device; establish a collaborative optimization model of steam extraction energy storage and molten salt heat storage, and optimize and adjust the steam extraction amount, energy storage release time and molten salt heat storage capacity according to the power grid load characteristics, renewable energy output characteristics and system equipment parameters; through the collaborative optimization model, a multi-objective balance of deep power grid peak regulation, renewable energy consumption and system economy is achieved.

[0196] According to a third aspect of the embodiments of the present invention,

[0197] An electronic device is provided, comprising:

[0198] processor;

[0199] a memory for storing processor-executable instructions;

[0200] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0201] According to a fourth aspect of the embodiments of the present invention,

[0202] 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.

[0203] 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.

[0204] 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. The method of extracting steam energy storage and molten salt heat storage under deep peak regulation of power grid is characterized in that: include: Obtain grid load forecast data and renewable energy generation forecast data; Determine a deep peak-shaving period of the power grid according to the power grid load forecast data and the renewable energy power generation forecast data; within the deep peak-shaving period of the power grid, extract steam from the steam turbine based on model predictive control, and introduce the extracted steam into a multi-stage steam extraction energy storage device; The multi-stage steam extraction energy storage device comprises a high-pressure energy storage chamber, a medium-pressure energy storage chamber and a low-pressure energy storage chamber connected in sequence, and each energy storage chamber is provided with a phase change energy storage material; the extracted steam is passed through the high-pressure energy storage chamber, the medium-pressure energy storage chamber and the low-pressure energy storage chamber in sequence, and the thermal energy of the steam is used to cause the phase change energy storage material in each energy storage chamber to undergo a phase change, thereby realizing the storage of thermal energy; a molten salt heat storage device is provided at the outlet end of the multi-stage steam extraction energy storage device, and the molten salt heat storage device comprises a molten salt storage tank and a heat exchanger; the steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to perform heat exchange with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release thermal energy; The released heat energy is used to heat feed water to generate high-temperature and high-pressure steam; the high-temperature and high-pressure steam is sent to a steam turbine to generate electricity to meet the peak load demand of the power grid; the changes in the power grid load are monitored in real time, and the charging and discharging processes of the multi-stage steam extraction energy storage device and the molten salt heat storage device are dynamically adjusted; a collaborative optimization model of steam extraction energy storage and molten salt heat storage is established, and the steam extraction amount, energy storage release time and molten salt heat storage capacity are optimized and adjusted according to the power grid load characteristics, renewable energy output characteristics and system equipment parameters; through the collaborative optimization model, a multi-objective balance of deep power grid peak regulation, renewable energy consumption and system economy is achieved.

2. The method according to claim 1, characterized in that During the deep peak load regulation period of the power grid, the steam turbine is extracted based on the model predictive control, and the extracted steam is introduced into the multi-stage steam extraction energy storage device, including: A nonlinear state space model of a steam turbine-extraction system is established, wherein the nonlinear state space model includes a steam turbine model, an extraction system model and an energy storage device model; wherein: The steam turbine model describes the relationship between state variables, control input, external disturbance and output variables; the steam extraction system model describes the relationship between the steam extraction amount and the flow coefficient, the effective flow area of ​​the steam extraction valve, the steam extraction pressure and the external pressure; the energy storage device model describes the relationship between the temperature change of the energy storage chamber and the inlet and outlet steam flow, specific enthalpy, heat loss, energy storage medium mass and specific heat capacity; Based on the nonlinear state space model, a model predictive controller is designed, and the steam extraction operation of the steam turbine based on the model predictive control includes: Define the prediction horizon and control horizon; Constructing a predictive control objective function, wherein the predictive control objective function takes into account a power generation deviation, a steam extraction deviation, an energy storage efficiency deviation, and a control increment; Set predictive control constraints, including extraction steam quantity constraint, extraction steam pressure constraint, extraction steam temperature constraint, load change rate constraint, extraction steam valve opening constraint and extraction steam valve opening change rate constraint; A sequential quadratic programming algorithm is used to solve the model predictive control optimization problem to obtain an optimal control sequence, and steam extraction operation is performed on the steam turbine based on the optimal control sequence.

3. The method according to claim 2, characterized in that The method further comprises: Linearizing the nonlinear state space model at the current working point; constructing a quadratic programming subproblem based on the linearized model; solving the quadratic programming subproblem using an interior point method to obtain a local optimal solution; updating the working point to obtain a global optimal control sequence; Execute real-time control processes, including: Collect steam turbine operation data, energy storage device status and short-term load forecast data; use extended Kalman filter to estimate the current state of the system based on the collected data; input the estimated current state and forecast information into the sequential quadratic programming algorithm to solve the model predictive control optimization problem and obtain the future control sequence; Extract the first value of the future control sequence as the instruction of the current control cycle, and send the control instruction to the distributed control system through the OPC UA protocol; monitor key parameters in real time, including but not limited to vibration, temperature and pressure, execute multi-level safety protection logic, and realize rolling optimization control.

4. The method according to claim 1, characterized in that: introducing the steam after passing through the multi-stage extracted steam storage device into the heat exchanger to perform heat exchange with the molten salt in the molten salt storage tank; During the peak period of power consumption in the power grid, controlling the multi-stage steam extraction energy storage device and the molten salt heat storage device to release heat energy includes: A system dynamic model is established, which includes a molten salt storage tank model, a heat exchanger model and a steam turbine model; wherein the molten salt storage tank model adopts a multi-node method to divide the tank into a plurality of equal-thickness nodes in the vertical direction, and an energy balance equation is established for each node; the heat exchanger model adopts a heat transfer unit number method, and introduces a time-varying scaling factor to characterize the change in thermal resistance of the pipe wall; the steam turbine model combines the Stodola elliptic law and the Willand equation to describe the characteristics of the steam turbine; the molten salt storage tank model, the heat exchanger model and the steam turbine model are connected through state variables and control variables to construct a state space expression of the overall system; Based on the system dynamic model, a long short-term memory network is used in combination with multi-source data to predict load demand, and the multi-source data includes historical load data, meteorological data and calendar information; the long short-term memory network includes an input layer, two hidden layers and an output layer, and the dropout technology is used to prevent overfitting; multiple long short-term memory network models with different initialization parameters are trained, and an integrated learning method is used to improve the prediction stability, and the final prediction result is the median of the outputs of multiple models; Based on the load demand forecasting result, an improved non-dominated sorting genetic algorithm III is used for multi-objective optimization; the first multi-objective function corresponding to the multi-objective optimization includes maximizing power generation efficiency, minimizing peak load response time and maximizing equipment life index; the improved non-dominated sorting genetic algorithm III includes using adaptive crossover and mutation operators, introducing local search operators, and using domination concepts to control the size of non-dominated solution sets; and using an ideal solution similarity sorting method to select the best compromise solution from the Pareto optimal solution set; The multi-stage steam extraction storage device and the molten salt heat storage device are controlled to release thermal energy based on the optimal compromise solution.

5. The method according to claim 1, characterized in that Establish a collaborative optimization model of steam extraction energy storage and molten salt heat storage, and optimize the steam extraction volume, energy storage release time and molten salt heat storage capacity according to the grid load characteristics, renewable energy output characteristics and system equipment parameters; Through the collaborative optimization model, the multi-objective balance of deep peak regulation of the power grid, renewable energy consumption and system economy is achieved, including: Establishing a steam extraction energy storage subsystem model, the steam extraction energy storage subsystem model includes a steam turbine output power model, an energy balance equation for an energy storage process, and an energy balance equation for an energy release process; wherein the steam turbine output power model considers steam flow, steam enthalpy, steam turbine efficiency, and generator efficiency; the energy balance equation for an energy storage process and the energy balance equation for an energy release process consider storage energy, energy storage efficiency, energy release efficiency, storage power, and energy release power, respectively; Establishing a molten salt heat storage subsystem model, the molten salt heat storage subsystem model includes a heat change equation for a hot storage tank and a heat change equation for a cold storage tank; wherein the heat change equation takes into account molten salt flow, specific heat capacity, inlet and outlet temperatures, heat loss coefficient, storage tank surface area and ambient temperature; Establishing a power grid load forecasting model, wherein the power grid load forecasting model adopts a time series analysis method and considers historical load data, wind speed, temperature, humidity and date type; Establishing a renewable energy output model, the renewable energy output model includes a wind power output model and a photovoltaic output model; wherein the wind power output model takes into account air density, wind rotor swept area, power coefficient and wind speed; the photovoltaic output model takes into account photoelectric conversion efficiency, photovoltaic panel area, solar radiation intensity and battery panel temperature; Constructing a collaborative optimization model, the collaborative optimization model includes a second multi-objective function and constraints; wherein the second multi-objective function comprehensively considers load tracking error, renewable energy curtailment and system operation cost; the constraints include power balance constraints, extraction steam storage system constraints, molten salt thermal storage system constraints and renewable energy output constraints; Design an improved multi-objective particle swarm optimization algorithm to solve the collaborative optimization model, and select three typical solutions based on the Pareto optimal solution set obtained, corresponding to the optimization solutions focusing on grid peak regulation, renewable energy consumption, and economy respectively; According to the selected typical solution, the optimal values ​​of steam extraction volume, energy storage release time and molten salt heat storage capacity are determined to achieve a multi-objective balance among deep peak regulation of the power grid, renewable energy consumption and system economy.

6. The method according to claim 5, characterized in that The method further comprises: The improved multi-objective particle swarm optimization algorithm includes the following improvements: Adopting adaptive inertia weight, which decreases nonlinearly with the increase of iteration number; introducing chaotic perturbation to perturb the global optimal solution; adopting fast non-dominated sorting algorithm to classify particles; using crowding degree calculation method to maintain the diversity of solutions; Executing the improved multi-objective particle swarm optimization algorithm comprises the following steps: Initialize the particle swarm and set algorithm parameters; Calculate the objective function value of each particle; Perform non-dominated sorting and congestion calculations; Update individual optimal solutions and global optimal solutions; Update particle velocity and position; Perform chaotic perturbation on the global optimal solution; Determine whether the termination condition is met. If so, output the Pareto optimal solution set.

7. A steam extraction energy storage and molten salt heat storage system under deep peak regulation of a power grid, used to implement the method described in any one of claims 1 to 6, characterized in that: include: The first unit is used to obtain grid load forecast data and renewable energy generation forecast data; Determine a deep peak-shaving period of the power grid according to the power grid load forecast data and the renewable energy power generation forecast data; within the deep peak-shaving period of the power grid, extract steam from the steam turbine based on model predictive control, and introduce the extracted steam into a multi-stage steam extraction energy storage device; The second unit is used for the multi-stage steam extraction energy storage device, which includes a high-pressure energy storage chamber, a medium-pressure energy storage chamber and a low-pressure energy storage chamber connected in sequence, and each energy storage chamber is provided with a phase change energy storage material; the extracted steam passes through the high-pressure energy storage chamber, the medium-pressure energy storage chamber and the low-pressure energy storage chamber in sequence, and the thermal energy of the steam is used to cause the phase change energy storage material in each energy storage chamber to undergo a phase change, thereby realizing the storage of thermal energy; a molten salt heat storage device is provided at the outlet end of the multi-stage steam extraction energy storage device, and the molten salt heat storage device includes a molten salt storage tank and a heat exchanger; the steam after passing through the multi-stage steam extraction energy storage device is introduced into the heat exchanger to perform heat exchange with the molten salt in the molten salt storage tank; during the peak period of power consumption of the power grid, the multi-stage steam extraction energy storage device and the molten salt heat storage device are controlled to release thermal energy; The third unit is used to heat the feed water using the released heat energy to generate high-temperature and high-pressure steam; The high-temperature and high-pressure steam is sent into a steam turbine to generate electricity to meet the peak load demand of the power grid; the changes in the power grid load are monitored in real time, and the charging and discharging processes of the multi-stage steam extraction energy storage device and the molten salt heat storage device are dynamically adjusted; a collaborative optimization model of steam extraction energy storage and molten salt heat storage is established, and the steam extraction volume, energy storage release time and molten salt heat storage capacity are optimized and adjusted according to the power grid load characteristics, renewable energy output characteristics and system equipment parameters; through the collaborative optimization model, a multi-objective balance of deep power grid peak regulation, renewable energy consumption and system economy is achieved.

8. 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 according to any one of claims 1 to 6.

9. 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 6 is implemented.

Images