A method for load tracking of a pumped storage power station based on a GAPSO algorithm

Abstract

The application provides a kind of GAPSO algorithm-based power station load tracking method, including data acquisition and preprocessing, and related data of power grid and power station are collected and processed;A power grid load forecasting model based on GAPSO algorithm is constructed, variables are determined, and an optimized neural network is constructed;The charge-discharge strategy of energy storage power station is optimized, the objective function is established and solved;Real-time scheduling and response are carried out, data monitoring, decision-making and adjustment of power station state are carried out;Multi-objective optimization adjustment is carried out, function is constructed, solved and continuously improved.The application improves the load forecasting accuracy, optimizes the power station operation strategy, enhances the real-time scheduling and response capability, realizes the multi-objective comprehensive optimization, effectively improves the operation efficiency and adaptability of energy storage power station in power system.

Description

Technical Field

[0001] This invention relates to the field of power system energy storage technology, specifically to a load tracking method for power storage substations based on the GAPSO algorithm. Background Technology

[0002] With the continuous growth of electricity demand and the increasing complexity of power grid structure, the volatility and uncertainty of power grid load pose a severe challenge to the stable operation of the power system. Traditional power generation methods suffer from slow response speed and poor regulation flexibility when dealing with rapid load changes.

[0003] Traditional power generation methods, such as thermal and hydropower, have significant limitations in responding to rapid load changes due to the inherent characteristics of their equipment. For example, thermal power units have relatively slow start-up and regulation processes, making it difficult to respond promptly to load changes. While hydropower units have relatively fast regulation speeds, their regulation capabilities are constrained by factors such as water resource distribution and seasonal variations. Therefore, relying solely on traditional power generation methods is insufficient to meet the requirements of modern power grids for rapid load tracking and precise regulation.

[0004] As an emerging power regulation method, energy storage power stations can charge and store energy during off-peak hours and discharge to provide power support during peak hours, effectively alleviating grid load pressure and improving grid stability and reliability.

[0005] Meanwhile, energy storage power stations can also participate in ancillary services such as frequency regulation and backup power, providing multifaceted support for the safe and stable operation of the power grid. However, the key to achieving efficient operation of energy storage power stations lies in formulating reasonable load tracking strategies. Accurately predicting the trend of power grid load changes and optimizing the charging and discharging strategies of energy storage power stations accordingly is a key focus and challenge in current power sector research.

[0006] Currently, existing load tracking methods still have many shortcomings in terms of forecast accuracy, strategy optimization effect, and multi-objective coordination, and cannot fully leverage the advantages of energy storage power stations in the power system. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by proposing a load tracking method for pumped storage substations based on the GAPSO algorithm.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] A load tracking method for energy storage power stations based on the GAPSO algorithm includes the following steps: S1, data acquisition and preprocessing; S2, construction of a power grid load prediction model based on the GAPSO algorithm; S3, optimization of charging and discharging strategies for energy storage power stations; S4, real-time scheduling and response; S5, multi-objective optimization adjustment.

[0010] Preferably, step S1, data acquisition and preprocessing, includes the following steps:

[0011] S11. Collect data;

[0012] The data collected includes historical load data and real-time load data of the power grid, operating parameters of energy storage power stations (including battery capacity, charging and discharging efficiency, etc.), electricity price information, and other relevant influencing factors (such as weather data, time factors, etc.). Load data for different time periods are obtained from the power grid monitoring system, battery status information is obtained from the energy storage power station management system, and electricity price change data is obtained from the electricity market.

[0013] S12, Data preprocessing;

[0014] The collected data is cleaned to remove outliers and erroneous data;

[0015] The Raida criterion was used to identify and correct data points that significantly deviated from the normal range.

[0016] Missing data can be supplemented using methods such as mean interpolation and linear interpolation. If load data is missing at a certain moment, it can be supplemented based on the average value or linear relationship of load data before and after that moment. Data normalization can be performed to map the data to a specific interval (such as [0,1] or [-1,1]) to improve the efficiency of subsequent calculations.

[0017] Let the original data be Normalized data (For mapping to the interval [0,1]), where and These are the minimum and maximum values ​​of the data, respectively.

[0018] Preferably, step S2, the construction of the power grid load forecasting model based on the GAPSO algorithm, includes the following steps:

[0019] S21: Determine the input and output variables;

[0020] Constructing input vectors This includes load values ​​at six different times. arrive The current hour is The indicator variable for whether the day is a workday is: ,in It is 0 or 1; the temperature value is Humidity value Special event flag variables The output variable is That is, the load forecast value at a future time;

[0021] S22: Construct a neural network prediction model optimized by the GAPSO algorithm;

[0022] Let the particle The position vector is ,in Indicates the neural network's first... The weights of each connection, Indicates the first The threshold of a neuron;

[0023] Calculate the fitness value for each particle, using the mean squared error as the fitness function. Let the number of predicted samples be... The actual load value is The predicted load value is Then the fitness function ;

[0024] Forecast load value To pass the input vector The input is fed into the neural network and calculated through forward propagation, i.e. ,in The mapping function representing the neural network;

[0025] The velocity and position of the particles are updated based on the fitness value. During the update process, dynamic crossover and mutation probability are used for the genetic algorithm part.

[0026] Crossover probability:

[0027] ,

[0028] Mutation probability:

[0029] ,

[0030] in For the number of iterations, and These are the minimum crossover probability and the maximum mutation probability, respectively.

[0031] Initial crossover probability The initial mutation probability is set between 0.6 and 0.9, the minimum crossover probability is set between 0.3 and 0.5, and the initial mutation probability is... Set between 0.01 and 0.1 for the maximum mutation probability. Set between 0.1 and 0.3;

[0032] In the particle swarm optimization algorithm, inertia weights... , As the initial weights, It is a constant. The number of iterations, the learning factor Initial weights Set between 0.7 and 0.9, with the constant determined by the maximum number of iterations. calculate, ,in, ;

[0033] The speed update formula is:

[0034] ,

[0035] The position update formula is:

[0036] ,

[0037] in For particles In the During the nth iteration Dimensional speed, For particles In the During the nth iteration The position of the dimension For particles In the The individual extreme value at the nth iteration is... Dimension value, The global extremum is at the th Dimension value, and It is a random number that is uniformly distributed within the interval [0,1].

[0038] Preferably, step S3, optimizing the charging and discharging strategy of the energy storage power station, includes the following steps:

[0039] S31: Establish the objective function for optimization:

[0040] The objective function is established to minimize the operating cost and maximize the revenue of the energy storage power station, while also considering the cost of battery life degradation.

[0041] The operating costs include charging costs and discharging costs;

[0042] Charging costs and electricity prices and charging power Related to the cost of discharge and the price of electricity for discharge. and discharge power Related;

[0043] Battery life degradation costs are calculated based on the number of charge-discharge cycles and the depth of charge-discharge.

[0044] Model using the Arrhenius formula:

[0045] Battery life degradation cost is ,

[0046] The objective function is:

[0047] ;

[0048] in, The scheduling cycle is set according to specific needs, such as one day or one hour.

[0049] The calculation of charging and discharging costs can be determined based on actual electricity pricing policies and market rules. If time-of-use pricing is adopted, and It has different values ​​at different times;

[0050] The calculation of battery life degradation cost can refer to the formula of the battery life degradation model based on the Arrhenius formula, according to the relationship between the life loss of energy storage batteries and discharge power:

[0051] Battery capacity is The number of loops is The loop depth is The battery throughput table is as follows: ;

[0052] Then, lifespan loss:

[0053] in Pre-exponential factor, For the experimental activation energy, The molar gas constant, Absolute temperature It is the reaction rate constant;

[0054] Further calculations yielded the cost of battery life degradation (combining factors such as battery replacement cost).

[0055] S32: Determine the constraints:

[0056] The constraints include the charging and discharging power constraints of the energy storage power station, the capacity constraints of the energy storage power station, and the power balance constraints of the power grid.

[0057] The charging and discharging power constraint of the energy storage power station is determined based on the rated power limit of the energy storage power station equipment, which determines the maximum charging and discharging power.

[0058] Including maximum charging power Maximum discharge power ;

[0059] but ;

[0060] For a 5MW energy storage power station:

[0061] (Considering a certain power reserve or efficiency factor).

[0062] The capacity constraints of the energy storage power station ensure its safe and stable operation, and its state of charge... It needs to be kept within a certain range;

[0063] set up The state of charge limit is, Upper limit of state of charge

[0064] but ;

[0065] The calculations are related to the battery's charging and discharging process:

[0066] ;

[0067] in For energy storage self-discharge rate, For charging efficiency, For discharge efficiency, The scheduling time interval;

[0068] The power balance constraint refers to the requirement that at any given time, the power exchanged between the power grid and the energy storage station, plus the charging and discharging power of the energy storage station, should equal the power grid load.

[0069] ;

[0070] in The power exchanged between the power grid and the energy storage power station. For grid load;

[0071] This constraint ensures the power balance of the power grid operation and avoids grid instability caused by power imbalance.

[0072] S33: Solving the charging and discharging strategy using the GAPSO algorithm:

[0073] By using the GAPSO algorithm to solve the charging and discharging strategy, the optimization objective function and constraints are transformed into a form that the GAPSO algorithm can handle; each particle represents a set of possible charging and discharging strategies, that is, the position vector of the particle represents the charging and discharging power value at different times.

[0074] Let the position vector of the particle be: ;in They represent particles respectively At any moment The charging power and discharging power;

[0075] Next, calculate the fitness value of the particles, and calculate the fitness of each particle according to the optimization objective function; the fitness function is: That is, particles The objective function value corresponding to the charge / discharge strategy is obtained by substituting the particle's position vector into the objective function, and then calculating the corresponding fitness value.

[0076] The particle velocity and position are updated based on the fitness value. The formulas used in the particle swarm algorithm are the same as those used in step S2, including inertia weight, learning factor, velocity update formula and position update formula.

[0077] During the update process, ensure that the position of the particle meets the constraints. If the updated particle position exceeds the charging / discharging power constraint or capacity constraint range, correct it to the feasible region.

[0078] For example, if the calculated charging power is greater than Then set it to If the state of charge is lower than If so, adjust the charging and discharging power to bring the state of charge back to a feasible range.

[0079] The process is iteratively optimized until the termination condition is met to obtain the optimal charging and discharging strategy. The charging and discharging power sequence corresponding to the optimal particle position at this point is the optimal charging and discharging strategy of the energy storage power station at different times, which can be used to guide the actual operation of the energy storage power station.

[0080] Preferably, step S4, real-time scheduling and response, includes the following steps:

[0081] S41: Real-time monitoring of grid load and energy storage station status:

[0082] High-precision sensors installed at key nodes of the power grid and inside energy storage power stations continuously acquire real-time load data of the power grid and real-time state of charge data of the energy storage power stations. Information on charging and discharging power;

[0083] The data acquisition frequency can be set to collect data every few minutes (e.g., 3-5 minutes) to ensure timely capture of changes in grid load and energy storage station status.

[0084] The collected data is processed and transmitted in real time to ensure its accuracy and timeliness. Reliable data communication technologies (such as specific protocols in wired or wireless communication) are used to transmit the data to the monitoring center or control system. During transmission, the data is verified and corrected to prevent data transmission errors from affecting subsequent decision-making.

[0085] S42: Make scheduling decisions based on prediction results and optimization strategies.

[0086] The real-time monitoring data in step S41 is input into the pre-built load forecasting model in step S2 to obtain short-term load forecast values; then, the neural network forecasting model optimized based on the GAPSO algorithm is used to predict the trend of power grid load change in the next 15 minutes or half an hour based on the current time and previous power grid load, time characteristics, and weather factors.

[0087] S43: Adjust the operating status of the energy storage power station:

[0088] After receiving a charge / discharge power command, the energy storage power station controls the output of the inverter to perform the charge / discharge operation. The inverter adjusts the output voltage, frequency, and phase according to the command to precisely control the charge / discharge power of the energy storage power station so that it matches the command value.

[0089] Preferably, step S5, multi-objective optimization adjustment, includes the following steps:

[0090] S51: Construction of Multi-Objective Optimization Functions

[0091] Taking into account key objectives such as grid stability, economic benefits of energy storage power stations, and lifespan loss of energy storage power stations, a multi-objective optimization function is constructed.

[0092] For power grid stability, load fluctuation index is used as a measure:

[0093] The variance of load fluctuations is as follows:

[0094] ;in No. Load value at any given time This is the average load value. The number of samples;

[0095] In terms of economic efficiency, the difference between operating costs and revenue is used as the indicator:

[0096] ;

[0097] in For energy storage power stations at all times Other revenues include revenues from participation in frequency regulation and the standby market;

[0098] The battery life degradation index is calculated based on the number of charge-discharge cycles and the depth of charge, consistent with step S31, using a relevant battery life degradation model, namely the Arrhenius formula model. The battery life degradation cost is assumed to be... ,

[0099] but ;

[0100] Construct a multi-objective optimization function:

[0101] ,in The weighting coefficients are assigned to the corresponding objectives; adjustments are made based on the actual power grid's emphasis on stability, economy, and energy storage lifespan.

[0102] When the power grid load fluctuates significantly and high stability requirements are needed, the load can be appropriately increased. The value; if the economic benefits of energy storage power stations are the primary consideration, then the value can be increased. The value; if you want to extend the battery life of the energy storage power station, you can increase it. value.

[0103] S52: GAPSO algorithm for solving multi-objective optimization problems:

[0104] The multi-objective optimization function and related constraints are integrated into the GAPSO algorithm framework. Each particle represents a set of parameter combinations that satisfy the constraints, including load prediction model parameters and energy storage power station charging and discharging strategy parameters.

[0105] The fitness value of the particles is calculated. Since it is a multi-objective optimization problem, non-dominated sorting and crowding calculation methods are used to evaluate the fitness of the particles.

[0106] Calculate the objective function value vector for each particle based on the multi-objective optimization function. ,in For particles The position vectors are used to sort the particles by non-dominated order, placing the non-dominated particles first to form different non-dominated levels. For particles within the same non-dominated level, their crowding degree is calculated. Crowding degree reflects the density of solutions around the particle; the higher the crowding degree, the sparser the solutions around the particle, and the better its diversity. The fitness value can be determined based on the non-dominated level and crowding degree; the smaller the non-dominated level and the higher the crowding degree, the higher the fitness value of the particle.

[0107] The particle velocity and position are updated based on the fitness value. The formulas used in the particle swarm algorithm are the same as those used in step S2, including inertia weight, learning factor, velocity update formula and position update formula.

[0108] During the update process, it is necessary to ensure that the position of the particles meets the constraints, including the charging and discharging power constraints, capacity constraints, and grid power balance constraints of the energy storage power station.

[0109] Iterative optimization continues until the termination condition is met (such as reaching the maximum number of iterations or the convergence of the non-dominated solution set; the change in the non-dominated solution set can be set to a threshold after several consecutive iterations, or the preset maximum number of iterations can be reached, such as 150 times), to obtain a multi-objective solution.

[0110] During the iterative optimization process, the velocity and position of the particles are continuously adjusted to allow the particles to search for better parameter combinations in the solution space.

[0111] After each iteration, the objective function value vector, non-dominated sorting, and crowding degree are recalculated based on the new particle positions, and the individual extrema and global extrema of the particles are updated.

[0112] Individual extrema are the parameter combinations that each particle finds at its optimal position during the iteration process, while global extrema are the optimal parameter combinations in the entire particle swarm.

[0113] S53: Implement optimization measures and continuously improve:

[0114] Based on the optimal parameter combination obtained by the GAPSO algorithm, a new load forecasting model and energy storage power station charging and discharging strategy are implemented.

[0115] The optimized load forecasting model parameters are applied to the load forecasting module to update the forecasting model and improve the accuracy of load forecasting.

[0116] The charging and discharging strategy parameters of the energy storage power station are applied to the control system of the energy storage power station to adjust the charging and discharging operation of the energy storage power station.

[0117] The weighting coefficients are dynamically adjusted based on actual operating conditions. If the power grid stability is found to be poor in actual operation, the following can be appropriately increased: The value is then evaluated, and the GAPSO algorithm is run again for optimization to achieve better overall performance. Through continuous monitoring, evaluation, and adjustment, the operation of the energy storage power station is made to adapt to the dynamic changes of the power grid, improving the overall performance and adaptability of the system. Simultaneously, the optimization process and results are recorded and analyzed to provide data support and experience for subsequent system optimization and decision-making.

[0118] Compared with the prior art, the beneficial effects of the present invention are as follows: First, the present invention achieves accurate prediction of grid load by introducing the GAPSO algorithm. The GAPSO algorithm combines the advantages of genetic algorithms and particle swarm optimization algorithms, and has the characteristics of strong global search capability and fast convergence speed. It can significantly improve the accuracy and stability of load prediction, and provide reliable data support for the optimization of charging and discharging strategies of energy storage power stations.

[0119] Secondly, this invention achieves a breakthrough in optimizing the charging and discharging strategies of energy storage power stations. By establishing an optimization objective function and imposing constraints on the capacity of the energy storage power station, the optimal charging and discharging strategy is obtained using the GAPSO algorithm. This strategy not only minimizes the operating costs and maximizes the revenue of the energy storage power station, but also fully considers the cost of battery life degradation, thus extending the service life of the energy storage power station.

[0120] Furthermore, this invention enables real-time scheduling and response. By monitoring the grid load and the status of energy storage power stations in real time, and making scheduling decisions based on prediction results and optimization strategies, the operating status of energy storage power stations can be adjusted in a timely manner, ensuring the stable operation of the grid.

[0121] Finally, this invention also considers multi-objective optimization adjustments. Taking into account multiple key objectives such as grid stability, economic benefits of energy storage power stations, and energy storage power station lifespan losses, a multi-objective optimization function was constructed, and the optimal parameter combination was obtained by solving the problem using the GAPSO algorithm. This scheme improves the accuracy of load forecasting and optimizes the charging and discharging strategy of energy storage power stations while also taking into account requirements for grid stability, economy, and energy storage lifespan. Attached Figure Description

[0122] Figure 1 This is a schematic diagram of the method framework in this invention;

[0123] Figure 2 This is a schematic diagram of the method flow in this invention. Detailed Implementation

[0124] To provide a further understanding of the purpose, structure, features, and functions of the present invention, detailed descriptions are provided below with reference to specific embodiments.

[0125] Please refer to the reference. Figure 1 as well as Figure 2 This invention provides a load tracking method for energy storage power stations based on the GAPSO algorithm, which includes the following steps: S1, data acquisition and preprocessing; S2, construction of a power grid load prediction model based on the GAPSO algorithm; S3, optimization of charging and discharging strategies for energy storage power stations; S4, real-time scheduling and response; S5, multi-objective optimization adjustment.

[0126] Preferably, step S1, data acquisition and preprocessing, includes the following steps:

[0127] S11. Collect data;

[0128] The data collected includes historical load data and real-time load data of the power grid, operating parameters of energy storage power stations (including battery capacity, charging and discharging efficiency, etc.), electricity price information, and other relevant influencing factors (such as weather data, time factors, etc.). Load data for different time periods are obtained from the power grid monitoring system, battery status information is obtained from the energy storage power station management system, and electricity price change data is obtained from the electricity market.

[0129] S12, Data preprocessing;

[0130] The collected data is cleaned to remove outliers and erroneous data;

[0131] The Raida criterion was used to identify and correct data points that significantly deviated from the normal range.

[0132] Missing data can be supplemented using methods such as mean interpolation and linear interpolation. If load data is missing at a certain moment, it can be supplemented based on the average value or linear relationship of load data before and after that moment. Data normalization can be performed to map the data to a specific interval (such as [0,1] or [-1,1]) to improve the efficiency of subsequent calculations.

[0133] Let the original data be Normalized data (For mapping to the interval [0,1]), where and These are the minimum and maximum values ​​of the data, respectively.

[0134] Preferably, step S2, the construction of the power grid load forecasting model based on the GAPSO algorithm, includes the following steps:

[0135] S21: Determine the input and output variables;

[0136] Constructing input vectors This includes load values ​​at six different times. arrive The current hour is The indicator variable for whether the day is a workday is: ,in It is 0 or 1; the temperature value is Humidity value Special event flag variables The output variable is That is, the load forecast value at a future time;

[0137] S22: Construct a neural network prediction model optimized by the GAPSO algorithm;

[0138] Let the particle The position vector is ,in Indicates the neural network's first... The weights of each connection, Indicates the first The threshold of a neuron;

[0139] Calculate the fitness value for each particle, using the mean squared error as the fitness function. Let the number of predicted samples be... The actual load value is The predicted load value is Then the fitness function ;

[0140] Forecast load value To pass the input vector The input is fed into the neural network and calculated through forward propagation, i.e. ,in The mapping function representing the neural network;

[0141] The velocity and position of the particles are updated based on the fitness value. During the update process, dynamic crossover and mutation probability are used for the genetic algorithm part.

[0142] Crossover probability:

[0143] ,

[0144] Mutation probability:

[0145] ,

[0146] in For the number of iterations, and These are the minimum crossover probability and the maximum mutation probability, respectively.

[0147] Initial crossover probability The initial mutation probability is set between 0.6 and 0.9, the minimum crossover probability is set between 0.3 and 0.5, and the initial mutation probability is... Set between 0.01 and 0.1 for the maximum mutation probability. Set between 0.1 and 0.3;

[0148] In the particle swarm optimization algorithm, inertia weights... , As the initial weights, It is a constant. The number of iterations, the learning factor Initial weights Set between 0.7 and 0.9, with the constant determined by the maximum number of iterations. calculate, ,in, ;

[0149] The speed update formula is:

[0150] ,

[0151] The position update formula is:

[0152] ,

[0153] in For particles In the During the nth iteration Dimensional speed, For particles In the During the nth iteration The position of the dimension For particles In the The individual extreme value at the nth iteration is... Dimension value, The global extremum is at the th Dimension value, and It is a random number that is uniformly distributed within the interval [0,1].

[0154] Preferably, step S3, optimizing the charging and discharging strategy of the energy storage power station, includes the following steps:

[0155] S31: Establish the objective function for optimization:

[0156] The objective function is established to minimize the operating cost and maximize the revenue of the energy storage power station, while also considering the cost of battery life degradation.

[0157] The operating costs include charging costs and discharging costs;

[0158] Charging costs and electricity prices and charging power Related to the cost of discharge and the price of electricity for discharge. and discharge power Related;

[0159] Battery life degradation costs are calculated based on the number of charge-discharge cycles and the depth of charge-discharge.

[0160] Model using the Arrhenius formula:

[0161] Battery life degradation cost is ,

[0162] The objective function is:

[0163] ;

[0164] in, The scheduling cycle is set according to specific needs, such as one day or one hour.

[0165] The calculation of charging and discharging costs can be determined based on actual electricity pricing policies and market rules. If time-of-use pricing is adopted, and It has different values ​​at different times;

[0166] The calculation of battery life degradation cost can refer to the formula of the battery life degradation model based on the Arrhenius formula, according to the relationship between the life loss of energy storage batteries and discharge power:

[0167] Battery capacity is The number of loops is The loop depth is The battery throughput table is as follows: ;

[0168] Then, lifespan loss:

[0169] in Pre-exponential factor, For the experimental activation energy, The molar gas constant, Absolute temperature It is the reaction rate constant;

[0170] Further calculations yielded the cost of battery life degradation (combining factors such as battery replacement cost).

[0171] S32: Determine the constraints:

[0172] The constraints include the charging and discharging power constraints of the energy storage power station, the capacity constraints of the energy storage power station, and the power balance constraints of the power grid.

[0173] The charging and discharging power constraint of the energy storage power station is determined based on the rated power limit of the energy storage power station equipment, which determines the maximum charging and discharging power.

[0174] Including maximum charging power Maximum discharge power ;

[0175] but ;

[0176] For a 5MW energy storage power station:

[0177] (Considering a certain power reserve or efficiency factor).

[0178] The capacity constraints of the energy storage power station ensure its safe and stable operation, and its state of charge... It needs to be kept within a certain range;

[0179] set up The state of charge limit is, Upper limit of state of charge

[0180] but ;

[0181] The calculations are related to the battery's charging and discharging process:

[0182] ;

[0183] in For energy storage self-discharge rate, For charging efficiency, For discharge efficiency, The scheduling time interval;

[0184] The power balance constraint refers to the requirement that at any given time, the power exchanged between the power grid and the energy storage station, plus the charging and discharging power of the energy storage station, should equal the power grid load.

[0185] ;

[0186] in The power exchanged between the power grid and the energy storage power station. For grid load;

[0187] This constraint ensures the power balance of the power grid operation and avoids grid instability caused by power imbalance.

[0188] S33: Solving the charging and discharging strategy using the GAPSO algorithm:

[0189] By using the GAPSO algorithm to solve the charging and discharging strategy, the objective function and constraints are transformed into a form that the GAPSO algorithm can handle; each particle represents a set of possible charging and discharging strategies, that is, the position vector of the particle represents the charging and discharging power value at different times.

[0190] Let the particle's position vector be: ;in They represent particles respectively At any moment The charging power and discharging power;

[0191] Next, calculate the fitness value of the particles, and calculate the fitness of each particle according to the optimization objective function; the fitness function is: That is, particles The objective function value corresponding to the charge / discharge strategy is obtained by substituting the particle's position vector into the objective function, and then calculating the corresponding fitness value.

[0192] The particle velocity and position are updated based on the fitness value. The formulas used in the particle swarm algorithm are the same as those used in step S2, including inertia weight, learning factor, velocity update formula and position update formula.

[0193] During the update process, ensure that the position of the particle meets the constraints. If the updated particle position exceeds the charging / discharging power constraint or capacity constraint range, correct it to the feasible region.

[0194] For example, if the calculated charging power is greater than Then set it to If the state of charge is lower than If so, adjust the charging and discharging power to bring the state of charge back to a feasible range.

[0195] The process is iteratively optimized until the termination condition is met to obtain the optimal charging and discharging strategy. The charging and discharging power sequence corresponding to the optimal particle position at this point is the optimal charging and discharging strategy of the energy storage power station at different times, which can be used to guide the actual operation of the energy storage power station.

[0196] Preferably, step S4, real-time scheduling and response, includes the following steps:

[0197] S41: Real-time monitoring of grid load and energy storage station status:

[0198] High-precision sensors installed at key nodes of the power grid and inside energy storage power stations continuously acquire real-time load data of the power grid and real-time state of charge data of the energy storage power stations. Information on charging and discharging power;

[0199] The data acquisition frequency can be set to collect data every few minutes (e.g., 3-5 minutes) to ensure timely capture of changes in grid load and energy storage station status.

[0200] The collected data is processed and transmitted in real time to ensure its accuracy and timeliness. Reliable data communication technologies (such as specific protocols in wired or wireless communication) are used to transmit the data to the monitoring center or control system. During transmission, the data is verified and corrected to prevent data transmission errors from affecting subsequent decision-making.

[0201] S42: Make scheduling decisions based on prediction results and optimization strategies.

[0202] The real-time monitoring data in step S41 is input into the pre-built load forecasting model in step S2 to obtain short-term load forecast values; then, the neural network forecasting model optimized based on the GAPSO algorithm is used to predict the trend of power grid load change in the next 15 minutes or half an hour based on the current time and previous power grid load, time characteristics, and weather factors.

[0203] S43: Adjust the operating status of the energy storage power station:

[0204] After receiving a charge / discharge power command, the energy storage power station controls the output of the inverter to perform the charge / discharge operation. The inverter adjusts the output voltage, frequency, and phase according to the command to precisely control the charge / discharge power of the energy storage power station so that it matches the command value.

[0205] An efficient data acquisition system was established, which uses high-precision sensors to acquire real-time information on grid load and energy storage power station status, and employs reliable data communication technology for real-time processing and transmission, ensuring the accuracy and timeliness of the data.

[0206] Based on real-time monitoring data and optimized prediction models, timely and accurate scheduling decisions can be made to adjust the charging and discharging power of energy storage power stations, enabling them to respond quickly to changes in grid load, achieve precise load tracking and power regulation, effectively ensure the stable operation of the grid, and reduce the risk of grid failures caused by load fluctuations.

[0207] Preferably, step S5, multi-objective optimization adjustment, includes the following steps:

[0208] S51: Construction of Multi-Objective Optimization Functions

[0209] Taking into account key objectives such as grid stability, economic benefits of energy storage power stations, and lifespan loss of energy storage power stations, a multi-objective optimization function is constructed.

[0210] For power grid stability, load fluctuation index is used as a measure:

[0211] The variance of load fluctuations is as follows:

[0212] ;in No. Load value at any given time This is the average load value. The number of samples;

[0213] In terms of economic efficiency, the difference between operating costs and revenue is used as the indicator:

[0214] ;

[0215] in For energy storage power stations at all times Other revenues include revenues from participation in frequency regulation and the standby market;

[0216] The battery life degradation index is calculated based on the number of charge-discharge cycles and the depth of charge, consistent with step S31, using a relevant battery life degradation model, namely the Arrhenius formula model. The battery life degradation cost is assumed to be... ,

[0217] but ;

[0218] Construct a multi-objective optimization function:

[0219] ,in The weighting coefficients are assigned to the corresponding objectives; adjustments are made based on the actual power grid's emphasis on stability, economy, and energy storage lifespan.

[0220] When the power grid load fluctuates significantly and high stability requirements are needed, the load can be appropriately increased. The value; if the economic benefits of energy storage power stations are the primary consideration, then the value can be increased. The value; if you want to extend the battery life of the energy storage power station, you can increase it. value.

[0221] S52: GAPSO algorithm for solving multi-objective optimization problems:

[0222] The multi-objective optimization function and related constraints are integrated into the GAPSO algorithm framework. Each particle represents a set of parameter combinations that satisfy the constraints, including load prediction model parameters and energy storage power station charging and discharging strategy parameters.

[0223] The fitness value of the particles is calculated. Since it is a multi-objective optimization problem, non-dominated sorting and crowding calculation methods are used to evaluate the fitness of the particles.

[0224] Calculate the objective function value vector for each particle based on the multi-objective optimization function. ,in For particles The position vectors are used to sort the particles by non-dominated order, placing the non-dominated particles first to form different non-dominated levels. For particles within the same non-dominated level, their crowding degree is calculated. Crowding degree reflects the density of solutions around the particle; the higher the crowding degree, the sparser the solutions around the particle, and the better its diversity. The fitness value can be determined based on the non-dominated level and crowding degree; the smaller the non-dominated level and the higher the crowding degree, the higher the fitness value of the particle.

[0225] The particle velocity and position are updated based on the fitness value. The formulas used in the particle swarm algorithm are the same as those used in step S2, including inertia weight, learning factor, velocity update formula and position update formula.

[0226] During the update process, it is necessary to ensure that the position of the particles meets the constraints, including the charging and discharging power constraints, capacity constraints, and grid power balance constraints of the energy storage power station.

[0227] Iterative optimization continues until the termination condition is met (such as reaching the maximum number of iterations or the convergence of the non-dominated solution set; the change in the non-dominated solution set can be set to a threshold after several consecutive iterations, or the preset maximum number of iterations can be reached, such as 150 times), to obtain a multi-objective solution.

[0228] During the iterative optimization process, the velocity and position of the particles are continuously adjusted to allow the particles to search for better parameter combinations in the solution space.

[0229] After each iteration, the objective function value vector, non-dominated sorting, and crowding degree are recalculated based on the new particle positions, and the individual extrema and global extrema of the particles are updated.

[0230] Individual extrema are the parameter combinations that each particle finds at its optimal position during the iteration process, while global extrema are the optimal parameter combinations in the entire particle swarm.

[0231] S53: Implement optimization measures and continuously improve:

[0232] Based on the optimal parameter combination obtained by the GAPSO algorithm, a new load forecasting model and energy storage power station charging and discharging strategy are implemented.

[0233] The optimized load forecasting model parameters are applied to the load forecasting module to update the forecasting model and improve the accuracy of load forecasting.

[0234] The charging and discharging strategy parameters of the energy storage power station are applied to the control system of the energy storage power station to adjust the charging and discharging operation of the energy storage power station.

[0235] The weighting coefficients are dynamically adjusted based on actual operating conditions. If the power grid stability is found to be poor in actual operation, the following can be appropriately increased: The value is then evaluated, and the GAPSO algorithm is run again for optimization to achieve better overall performance. Through continuous monitoring, evaluation, and adjustment, the operation of the energy storage power station is made to adapt to the dynamic changes of the power grid, improving the overall performance and adaptability of the system. Simultaneously, the optimization process and results are recorded and analyzed to provide data support and experience for subsequent system optimization and decision-making.

[0236] First, this invention achieves accurate load forecasting by introducing the GAPSO algorithm. The GAPSO algorithm combines the advantages of genetic algorithms and particle swarm optimization, featuring strong global search capabilities and fast convergence speed. It significantly improves the accuracy and stability of load forecasting, providing reliable data support for optimizing the charging and discharging strategies of energy storage power stations.

[0237] Secondly, this invention achieves a breakthrough in optimizing the charging and discharging strategies of energy storage power stations. By establishing an optimization objective function and imposing constraints on the capacity of the energy storage power station, the optimal charging and discharging strategy is obtained using the GAPSO algorithm. This strategy not only minimizes the operating costs and maximizes the revenue of the energy storage power station, but also fully considers the cost of battery life degradation, thus extending the service life of the energy storage power station.

[0238] Furthermore, this invention enables real-time scheduling and response. By monitoring the grid load and the status of energy storage power stations in real time, and making scheduling decisions based on prediction results and optimization strategies, the operating status of energy storage power stations can be adjusted in a timely manner, ensuring the stable operation of the grid.

[0239] An optimization function was constructed that comprehensively considers multiple key objectives, including grid stability (measured by load fluctuation index), economic benefits of energy storage power stations (measured by the difference between operating costs and revenues), and lifespan loss of energy storage power stations (calculated based on the number of battery charge-discharge cycles and depth). The function also allows for flexible emphasis on different objectives by adjusting the weighting coefficients.

[0240] The GAPSO algorithm is used to solve a multi-objective optimization problem, yielding a set of non-dominated solutions that clearly demonstrate the trade-offs between different objectives, facilitating the selection of appropriate optimization schemes based on actual needs. By continuously monitoring and dynamically adjusting the weight coefficients, the operation of the energy storage power station can adapt to dynamic changes in the power grid, achieving comprehensive optimization of multiple objectives and improving the overall performance and adaptability of the power system.

[0241] The present invention has been described in the above-described embodiments; however, these embodiments are merely examples for implementing the present invention. It must be noted that the disclosed embodiments do not limit the scope of the present invention. Conversely, any modifications and refinements made without departing from the spirit and scope of the present invention are within the scope of patent protection of the present invention.

Claims

1. A load tracking method for pumped storage substations based on the GAPSO algorithm, characterized in that: The specific steps of the method are as follows: S1: Data acquisition and preprocessing; During data acquisition, historical load data and real-time load data of the power grid, operating parameters of energy storage power stations, electricity price information, and other relevant influencing factors data are collected, including weather data and time factors. It also obtains load data for different time periods from the power grid monitoring system, battery status information from the energy storage power station management system, and electricity price change data from the electricity market; The collected and acquired data is then cleaned to remove outliers and erroneous data. S2: Construction of a power grid load forecasting model based on the GAPSO algorithm; S21: Determine the input and output variables; Constructing input vectors This includes load values ​​at six different times. arrive ; The current hour is The indicator variable for whether the day is a workday is: ,in It is 0 or 1; the temperature value is Humidity value ; Special event flag variables ; The output variable is That is, the load forecast value at a future time; S22: Construct a neural network prediction model optimized by the GAPSO algorithm; Let the particle The position vector is ,in Indicates the neural network's first... The weights of each connection, Indicates the first The threshold of a neuron; Calculate the fitness value for each particle, using the mean squared error as the fitness function. Let the number of predicted samples be... The actual load value is The predicted load value is Then the fitness function ; Forecast load value To pass the input vector The input is fed into the neural network and calculated through forward propagation, i.e. ,in The mapping function representing the neural network; The velocity and position of the particles are updated based on the fitness value. During the update process, dynamic crossover and mutation probability are used for the genetic algorithm part. Crossover probability: ， Mutation probability: ， in For the number of iterations, and These are the minimum crossover probability and the maximum mutation probability, respectively. Initial crossover probability The initial mutation probability is set between 0.6 and 0.9, the minimum crossover probability is set between 0.3 and 0.5, and the initial mutation probability is... Set between 0.01 and 0.1 for the maximum mutation probability. Set between 0.1 and 0.3; In the particle swarm optimization algorithm, inertia weights... , As the initial weights, It is a constant. The number of iterations, the learning factor Initial weights Set between 0.7 and 0.9, with the constant determined by the maximum number of iterations. calculate, ,in, ; The speed update formula is: ， The position update formula is: ， in For particles In the During the nth iteration Dimensional speed, For particles In the During the nth iteration The position of the dimension For particles In the The individual extreme value at the nth iteration is... Dimension value, The global extremum is at the th Dimension value, and It is a random number uniformly distributed within the interval [0,1]. When the termination condition is met, the optimal weights and values ​​are output to construct the final load prediction model. The combination of weights and thresholds corresponding to the optimal particle position at this time is the desired result, which is then applied to the neural network for subsequent load prediction. S3: Optimization of charging and discharging strategies for energy storage power stations; By establishing an optimization objective function and imposing constraints on the capacity of the energy storage power station, and using the GAPSO algorithm in step S2 to solve the charging and discharging strategy, the optimization objective function and constraints are transformed into a form that can be handled by the GAPSO algorithm. The fitness value of each particle is recalculated according to the optimization objective function, and the velocity and position of the particles are updated according to the recalculated fitness value. The optimization is iterated until the termination condition is met to obtain the optimal charging and discharging strategy. S4: Real-time scheduling and response; By monitoring the grid load and the status of the energy storage station in real time, scheduling decisions are made based on the prediction results of step S2 and the optimization strategy of step S3. After receiving the charging and discharging power command, the energy storage station controls the output of the inverter equipment to realize the charging and discharging operation. S5: Multi-objective optimization adjustment; Taking into account multiple key objectives such as grid stability, economic benefits of energy storage power stations, and lifespan loss of energy storage power stations, a multi-objective optimization function is constructed. The multi-objective optimization function and related constraints are integrated into the GAPSO algorithm framework. Based on the optimal parameter combination obtained by the GAPSO algorithm, a new load forecasting model and energy storage power station charging and discharging strategy are implemented.

2. The load tracking method for pumped storage substations based on the GAPSO algorithm as described in claim 1, characterized in that: In step S1, the data acquisition and preprocessing are as follows: S11: Data Collection During data acquisition, historical load data and real-time load data of the power grid, operating parameters of energy storage power stations, electricity price information, and other relevant influencing factors data are collected, including weather data and time factors. It also obtains load data for different time periods from the power grid monitoring system, battery status information from the energy storage power station management system, and electricity price change data from the electricity market; S12: Data Preprocessing The collected data is cleaned to remove outliers and erroneous data; The Laida criterion was used to identify and correct data points that significantly deviated from the normal range; missing data were supplemented using mean interpolation and linear interpolation methods. Normalize the data to map it to a specific interval, [0,1] or [-1,1].

3. The load tracking method for pumped storage substations based on the GAPSO algorithm as described in claim 1, characterized in that: In step S3, the energy storage power station's charging and discharging strategy is optimized, and the specific steps are as follows: S31: Establish the objective function for optimization: The objective function is established to minimize the operating cost and maximize the revenue of the energy storage power station, while also considering the cost of battery life degradation. The operating costs include charging costs and discharging costs; Charging costs and electricity prices and charging power Related to the cost of discharge and the price of electricity for discharge. and discharge power Related; Battery life degradation costs are calculated based on the number of charge-discharge cycles and the depth of charge-discharge. Model using the Arrhenius formula: Battery life degradation cost is , The objective function is then: ； in, The scheduling period; S32: Determine the constraints: The constraints include the charging and discharging power constraints of the energy storage power station, the capacity constraints of the energy storage power station, and the power balance constraints of the power grid. The charging and discharging power constraint of the energy storage power station is determined based on the rated power limit of the energy storage power station equipment, which determines the maximum charging and discharging power. Including maximum charging power Maximum discharge power ; but ; The capacity constraints of the energy storage power station ensure its safe and stable operation, and its state of charge... It needs to be kept within a certain range; set up The state of charge limit is, Upper limit of state of charge but ; The calculations are related to the battery's charging and discharging process: ； in For energy storage self-discharge rate, For charging efficiency, For discharge efficiency, The scheduling time interval; The power balance constraint refers to the requirement that at any given time, the power exchanged between the power grid and the energy storage station, plus the charging and discharging power of the energy storage station, should equal the power grid load. ； in The power exchanged between the power grid and the energy storage power station. For grid load; S33: Solving the charging and discharging strategy using the GAPSO algorithm: By using the GAPSO algorithm to solve the charging and discharging strategy, the objective function and constraints are transformed into a form that the GAPSO algorithm can handle; each particle represents a set of possible charging and discharging strategies, that is, the position vector of the particle represents the charging and discharging power value at different times. Let the particle's position vector be: ;in They represent particles respectively At any moment The charging power and discharging power; Next, calculate the fitness value of the particles, and calculate the fitness of each particle according to the optimization objective function; the fitness function is: That is, particles The objective function value corresponding to the charge / discharge strategy is obtained by substituting the particle's position vector into the objective function, and then calculating the corresponding fitness value. The particle velocity and position are updated based on the fitness value. The formulas used in the particle swarm algorithm are the same as those used in step S2, including inertia weight, learning factor, velocity update formula and position update formula. During the update process, ensure that the position of the particle meets the constraints. If the updated particle position exceeds the charging / discharging power constraint or capacity constraint range, correct it to the feasible region.

4. The load tracking method for pumped storage substations based on the GAPSO algorithm as described in claim 1, characterized in that: In step S4, real-time scheduling and response include the following steps: S41: Real-time monitoring of grid load and energy storage station status: High-precision sensors installed at key nodes of the power grid and inside energy storage power stations continuously acquire real-time load data of the power grid and real-time state of charge data of the energy storage power stations. Information on charging and discharging power; S42: Make scheduling decisions based on prediction results and optimization strategies. The real-time monitoring data in step S41 is input into the pre-built load forecasting model in step S2 to obtain short-term load forecast values; then, the neural network forecasting model optimized based on the GAPSO algorithm is used to predict the future trend of power grid load changes based on the current and previous power grid load, time characteristics, and weather factors. S43: Adjust the operating status of the energy storage power station: After receiving a charge / discharge power command, the energy storage power station controls the output of the inverter to perform the charge / discharge operation. The inverter adjusts the output voltage, frequency, and phase according to the command to precisely control the charge / discharge power of the energy storage power station so that it matches the command value.

5. The load tracking method for pumped storage substations based on the GAPSO algorithm as described in claim 3, characterized in that: The specific steps for multi-objective optimization adjustment in step S5 are as follows: S51: Construction of Multi-Objective Optimization Functions Taking into account key objectives such as grid stability, economic benefits of energy storage power stations, and lifespan loss of energy storage power stations, a multi-objective optimization function is constructed. For power grid stability, load fluctuation index is used as a measure: The variance of load fluctuations is as follows: ;in No. Load value at any given time This is the average load value. The number of samples; In terms of economic efficiency, the difference between operating costs and revenue is used as the indicator: ； in For energy storage power stations at all times Other revenues include revenues from participation in frequency regulation and the standby market; The battery life degradation index is calculated based on the number of charge-discharge cycles and the depth of charge, consistent with step S31, using a relevant battery life degradation model, namely the Arrhenius formula model. The battery life degradation cost is assumed to be... , but ; Construct a multi-objective optimization function: ,in The weighting coefficients are assigned to the corresponding objectives; adjustments are made based on the actual power grid's emphasis on stability, economy, and energy storage lifespan. S52: GAPSO algorithm for solving multi-objective optimization problems: The multi-objective optimization function and related constraints are integrated into the GAPSO algorithm framework. Each particle represents a set of parameter combinations that satisfy the constraints, including load prediction model parameters and energy storage power station charging and discharging strategy parameters. The fitness value of the particles is calculated. Since it is a multi-objective optimization problem, non-dominated sorting and crowding calculation methods are used to evaluate the fitness of the particles. Calculate the objective function value vector for each particle based on the multi-objective optimization function. ,in For particles The position vector is then used to sort the particles by non-dominated order, placing the non-dominated particles at the front to form different non-dominated levels. For particles within the same non-dominated level, their crowding degree is calculated. The particle velocity and position are updated based on the fitness value. The formulas used in the particle swarm algorithm are the same as those used in step S2, including inertia weight, learning factor, velocity update formula and position update formula. During the update process, it is necessary to ensure that the position of the particles meets the constraints, including the charging and discharging power constraints, capacity constraints, and grid power balance constraints of the energy storage power station. S53: Implement optimization measures and continuously improve: Based on the optimal parameter combination obtained by the GAPSO algorithm, a new load forecasting model and energy storage power station charging and discharging strategy are implemented. The optimized load forecasting model parameters are applied to the load forecasting module to update the forecasting model and improve the accuracy of load forecasting. The charging and discharging strategy parameters of the energy storage power station are applied to the control system of the energy storage power station to adjust the charging and discharging operation of the energy storage power station.

Images