A Wind-Solar-Pumped Hydro Storage Joint Power Generation Dispatch Method Based on an Improved Bat Algorithm

Abstract

This invention, belonging to the field of renewable energy power generation dispatching technology, proposes a wind-solar-pumped hydro storage joint power generation dispatching method based on an improved Bat algorithm. It constructs a dispatching model aimed at maximizing the expected revenue of the joint power generation system, considering the grid connection revenue and operating costs of various generator units, the pumping costs of pumped hydro storage, and the pollutant emission penalty costs of thermal power units. The goal is to optimize the joint dispatching of wind, solar, thermal, and pumped hydro storage units, improving the overall system revenue and stability. This invention proposes an improved Bat algorithm, employing chaotic mapping initialization, introducing adaptive step-size updates, and a Cauchy mutation factor, significantly improving the algorithm's convergence speed and global search capability, and more effectively avoiding local optima traps. The dispatching scheme based on the improved Bat algorithm fully releases the peak-shaving capacity of pumped hydro storage, enhances the acceptance capacity of renewable energy, and reduces the impact on conventional thermal power units.

Description

Technical Field

[0001] This invention belongs to the field of renewable energy power generation dispatching technology, and in particular relates to a method for joint power generation dispatching of wind power, photovoltaic power, thermal power and pumped storage based on an improved bat algorithm. Background Technology

[0002] With the increasing global energy shortage and air pollution, new energy power generation (such as wind and solar power) is gradually becoming a major component of the power system. However, the randomness and volatility of wind and solar power pose challenges to the safe and stable operation of the power system. To balance the volatility of wind and solar power, pumped hydro storage units, as an important energy storage method, can effectively regulate grid load, reduce wind and solar curtailment, and stabilize power supply.

[0003] Traditional scheduling methods suffer from slow convergence and susceptibility to local optima when optimizing integrated wind-solar-storage systems, failing to meet the demands of modern large-scale renewable energy system scheduling. Therefore, a new scheduling method is needed to improve the system's economics and stability. Summary of the Invention

[0004] The wind-solar-pumped hydro storage joint power generation dispatch method based on the improved bat algorithm includes the following steps, which are performed sequentially:

[0005] Step 1: Construct a combined power generation system that includes wind power, photovoltaic power, thermal power and pumped storage units, wherein the pumped storage units include water pumps and water turbines;

[0006] Step 2: Analyze and model the loss mechanism of pumped storage units;

[0007] Step 3: Perform uncertainty modeling, scenario generation, and stochastic optimization of the joint generation system to ensure the robustness of the scheduling and allocation process of the joint generation system under different uncertainty scenarios;

[0008] Step 4: The combined power generation system also includes a transient analysis module. The transient analysis module models the dynamic behavior of the power system using physical models and mathematical formulas, and is used to monitor and analyze the dynamic response of the combined power generation system in real time, and to provide a basis for dispatching.

[0009] Step 5: Establish a scheduling model with the goal of maximizing the total revenue of the combined power generation system. The scheduling model includes the power generation revenue and operating costs of various generator units, the pumping cost of pumped storage, and the power generation cost and pollutant emission penalty cost of thermal power units. Conditional constraints are imposed on the combined power generation system to ensure the rationality and feasibility of the scheduling process.

[0010] Step 6: Solve the scheduling model using the improved bat algorithm to optimize the output scheduling of each generator unit.

[0011] The improved bat algorithm includes the following steps:

[0012] a. Initialize the population using chaotic mapping to increase the coverage of the initial solution space;

[0013] The Bat Algorithm initially uses a random method for population initialization, which cannot cover the entire solution space. Therefore, a chaotic mapping method is used for population initialization to improve the coverage of the initial solution space. The calculation formula is as follows:

[0014] ;

[0015] In the formula, It is a chaotic sequence. ; For the rounds of the sequence; To initialize the population dimension; To initialize the population size, facing Perform the inversion operation to obtain the solution space initialization population pairs, calculated as follows:

[0016] ;

[0017] In the formula; and These are the minimum and maximum values ​​of the variable's range of values, respectively. Indicates the first Only bats Position of the round;

[0018] b. Introduce an adaptive step size update method to accelerate the convergence process of the algorithm;

[0019] The adaptive step size update method uses a large initial step size, which improves the convergence speed. In the later stages of the algorithm, the step size becomes smaller, and the search becomes more refined. The formula for the bat velocity update method based on adaptive step size is as follows:

[0020] ;

[0021] ;

[0022] In the formula, , The first Only bats and Flight speed per round; For adaptive step size; For individuals in the optimal position; For the minimum step size, The maximum number of iterations, As a regulating factor; It is a natural constant;

[0023] c. Add a Cauchy mutation factor during the search process to enhance the algorithm's ability to escape local optima;

[0024] right t The bat's flight speed variable at time t is modified by adding a Cauchy mutation factor to change the bat's flight speed and enhance its ability to escape local optima. The mutation formula is shown below:

[0025] ;

[0026] ;

[0027] In the formula, For the first i Only bats Position of the round; It is a uniformly distributed random number within the interval [0, 1]. For the coefficient vector, It is a vector that decreases linearly from 2 to 0; Let be a uniformly distributed random vector within the interval [0, 1].

[0028] The specific method for analyzing and modeling the loss mechanism of pumped storage units in step two is as follows:

[0029] The loss mechanism includes hydraulic losses and electrical losses. Hydraulic losses are represented by the efficiency coefficients of the water pumps and turbines, while electrical losses are represented by the efficiency coefficients of the generators and motors. Finally, the overall energy conversion efficiency of the pumped storage system is calculated using a comprehensive efficiency coefficient formula, which is as follows:

[0030] ;

[0031] in, The overall efficiency coefficient, This represents the efficiency coefficient of the water turbine. It is the efficiency coefficient of the generator. It is the efficiency coefficient of the electric motor;

[0032] ;

[0033] ;

[0034] ;

[0035] In the formula, It is the hydraulic power of a water pump after it converts electrical energy into the potential energy of water. It is the input power of the water pump. It is the generator's electrical power. It is the mechanical power of the generator. It is the electric power of the motor. It refers to the mechanical power of the electric motor.

[0036] The uncertainty modeling in step three includes using Weibull and Beta distributions to model the uncertainty of wind power and photovoltaic power generation, and using historical data statistical analysis to obtain the probability density functions of wind speed and light intensity.

[0037] The probability density function of wind speed is:

[0038] ;

[0039] in,

[0040] ;

[0041] ;

[0042] In the formula, Wind speed; For shape parameters; For scale parameters; It is a gamma function; Average wind speed; This represents the standard deviation of wind speed, reflecting the degree to which wind speed fluctuates around its average value.

[0043] The probability density function of solar radiation intensity is:

[0044] ;

[0045] In the formula, Normalized light intensity; and For shape parameters; This is a Beta parameter.

[0046] The objective function of the scheduling model considers maximizing the expected return for different scenario probability distributions, and the formula is as follows:

[0047] ;

[0048] In the formula, To maximize the expected revenue of the combined power generation system; For the first One scenario; For the scene The probability of; For the scene The following benefits; Total number of scenes; Revenue generated by a combined power generation system that includes wind power, photovoltaic power, thermal power units and pumped storage units; For the operating costs of wind power and photovoltaic units; The cost of pumping water for pumped storage units; The cost of generating electricity from thermal power units; The punitive costs of pollutant emissions;

[0049] Among them, the power generation revenue of the combined power generation system including wind power, photovoltaic, thermal power units and pumped storage units It is the sum of the following parts:

[0050] The feed-in tariff for solar power is multiplied by the output power of the solar power unit.

[0051] The on-grid electricity price of wind power multiplied by the output power of the wind turbine;

[0052] The grid-connected electricity price of pumped hydro storage is multiplied by the output power of the pumped hydro storage unit.

[0053] The on-grid electricity price of thermal power units multiplied by the output power of thermal power units;

[0054] The scenario involves using scenario generation technology to create multiple possible wind power and photovoltaic unit power generation output scenarios. Each scenario is randomly generated based on the probability distribution of wind speed and solar irradiance, reflecting power generation fluctuations under different weather conditions and time periods, thereby improving the robustness and accuracy of the scheduling model. A scenario As shown below:

[0055] ;

[0056] in, and The first The output power of wind turbines and photovoltaic units at time t in each scenario.

[0057] The transient analysis module establishes dynamic simulation models for the following components:

[0058] Based on the mechanical and electromagnetic characteristics of the generator, the transient model of the generator is represented as follows:

[0059] ;

[0060] In the formula, The moment of inertia of the generator; It is the mechanical angular velocity; and These are the mechanical input torque and the electromagnetic output torque, respectively. The damping coefficient;

[0061] ;

[0062] In the formula: It is the stator voltage; It is the stator resistance; It is the stator current; It is a stator inductor; It is the transient electromotive force of the generator; The stator-side electrical angular frequency;

[0063] To ensure system stability during sudden changes in power supply and demand, voltage response and frequency response modeling are used to address sudden fault events. Voltage response is used to describe the voltage recovery process after a fault. The model is as follows:

[0064] ;

[0065] In the formula, It is the voltage before the fault; It is the attenuation coefficient; It is the damped oscillation frequency; It is the phase angle; The time variable is the period after the fault occurs;

[0066] Frequency response The change in system frequency after a disturbance is described by the following equation:

[0067] ;

[0068] in, It is the maximum value of the frequency offset; It is the natural oscillation frequency.

[0069] The transient analysis module also includes time-domain simulation and frequency-domain analysis modules. It simulates the changes in voltage, frequency and phase angle of the system through differential equations and numerical integration methods, and uses Fast Fourier Transform (FFT) to perform frequency-domain analysis on the established dynamic simulation model to evaluate the stability and oscillation modes of the system.

[0070] Through the above design scheme, the present invention can bring the following beneficial effects:

[0071] This invention constructs a scheduling model aimed at maximizing the expected revenue of a combined power generation system. It considers the power generation revenue and operating costs of various generator units, the pumping costs of pumped hydro storage, and the pollutant emission penalty costs of thermal power units. The aim is to optimize the joint scheduling of wind power, photovoltaic, thermal power, and pumped hydro storage units, improving the overall system revenue and stability. To address the poor convergence and susceptibility to local optima in traditional scheduling algorithms when handling complex optimization problems, this invention proposes an improved Bat Algorithm. This algorithm employs chaotic mapping initialization, introduces adaptive step-size updates, and incorporates a Cauchy mutation factor, significantly improving the algorithm's convergence speed and global search capability. Case studies demonstrate that the improved algorithm not only enhances convergence speed but also effectively avoids local optima traps. The scheduling scheme based on the improved Bat Algorithm fully leverages the peak-shaving capacity of pumped hydro storage, significantly mitigating the impact of renewable energy output uncertainty on the power grid during large-scale distributed energy grid integration. This scheme enhances the acceptance capacity of renewable energy while reducing the impact on conventional thermal power units.

[0072] This invention significantly improves dispatch efficiency and system benefits through innovative algorithm improvements. The method optimizes the output dispatch of wind power, photovoltaic power, and pumped storage units, reducing wind and solar curtailment and enhancing the absorption capacity of renewable energy. Simultaneously, the system effectively regulates grid load, mitigates the impact of renewable energy fluctuations on the grid, reduces dispatch pressure on thermal power units, and decreases pollutant emissions. Furthermore, the improved bat algorithm exhibits strong robustness, adapting to complex grid environments and uncertain power generation output, ensuring stable system operation under various conditions. Through multi-objective optimization, this invention achieves a balance between economic and environmental benefits, providing an effective solution for the optimized dispatch of green power systems.

[0073] Furthermore, this invention deeply analyzes the loss mechanism of pumped storage systems, including hydraulic and electrical losses, and accurately calculates the overall energy conversion efficiency of the system using a comprehensive efficiency formula. Addressing the uncertainties of wind and solar power generation, scenario generation technology is employed to create multiple possible power output scenarios to maximize expected returns and improve the robustness and accuracy of the scheduling model. To further optimize the scheduling scheme, this invention introduces a highly integrated transient analysis module, employing time-domain simulation and frequency-domain analysis methods to evaluate the dynamic response of the power system in real time and provide a basis for scheduling strategies. Simulation results show that the improved bat algorithm, compared to traditional algorithms, reaches the optimal solution in fewer iterations, improves scheduling returns, optimizes the absorption capacity of new energy sources, and effectively reduces the scheduling burden on thermal power units. This invention provides an effective solution for the optimized scheduling of wind-solar-pumped storage combined power generation systems, possessing significant application value and promising prospects for widespread adoption. Attached Figure Description

[0074] The present invention will be further described below with reference to the accompanying drawings and specific embodiments:

[0075] Figure 1 This is a structural block diagram of a distributed combined power generation system containing pumped hydro storage in the wind-solar-pumped hydro storage combined power generation scheduling method based on the improved bat algorithm of the present invention.

[0076] Figure 2 This is a flowchart of the scheduling model solution based on the improved bat algorithm in the wind-solar-pumped hydro storage joint power generation scheduling method based on the improved bat algorithm of the present invention.

[0077] Figure 3 This is a wind power and photovoltaic output curve diagram in an embodiment of the wind-solar-pumped hydro storage joint power generation scheduling method based on the improved bat algorithm of the present invention.

[0078] Figure 4 This is a simulation iterative convergence curve in an embodiment of the wind-solar-pumped hydro storage joint power generation scheduling method based on the improved bat algorithm of the present invention;

[0079] Figure 5 This is a revenue curve of the joint power generation system at each time step in an embodiment of the wind-solar-pumped hydro storage joint power generation scheduling method based on the improved bat algorithm of the present invention, after using the bat algorithm, particle swarm algorithm and improved bat algorithm.

[0080] Figure 6 This is an example of the wind-solar-pumped hydro storage joint power generation and dispatching method based on the improved bat algorithm of the present invention. A schematic diagram of the Pareto optimal solution for the constraint method. Detailed Implementation

[0081] This invention relates to a wind-solar-pumped hydro storage joint power generation scheduling method based on an improved bat algorithm, which is implemented through the following technical solutions:

[0082] 1. Design of a distributed combined generation system including pumped hydro storage

[0083] Pumped hydro storage utilizes surplus electricity from the power grid to drive pumps that draw water from a lower reservoir to an upper reservoir, storing the potential energy of the water in the upper reservoir. When the power grid's demand is insufficient, the potential energy of the water is converted into electrical energy through a turbine generator. Pumped hydro storage's ability to absorb low-voltage electricity and generate high-voltage electricity effectively stores electrical energy, effectively regulating the production, supply, and use of the power system and maintaining a dynamic balance among the three.

[0084] The integration of pumped hydro storage units allows for control of the charging and discharging process based on grid demand, thus forming an integrated wind-solar-storage system. The platform control mode in this system is more convenient and flexible, and is now trending towards intelligence. The platform control mode can adjust the energy storage system's capacity according to power conditions. A typical platform mainly includes a wind farm, photovoltaic system, energy storage system, control center, power transmission lines, and signal feedback lines. The structure of a distributed combined generation system including pumped hydro storage is as follows: Figure 1 As shown.

[0085] The main components of a pumped-storage hydroelectric power unit (PSH) are pumps and turbines. During off-peak electricity demand, the pumps draw water from the lower reservoir to the upper reservoir, converting excess electrical energy in the grid into potential energy for storage. During peak electricity demand, the PSH draws water from the upper reservoir to the lower reservoir, driving the turbines in the main powerhouse to generate electricity, converting potential energy into electrical energy. The mathematical model for the energy storage and release process can be expressed as:

[0086] (1);

[0087] (2);

[0088] In the formula, The power of the water pump; The density of water; It is the acceleration due to gravity; The pump head; The volumetric flow rate of water pumped into the reservoir; This is the overall efficiency coefficient; The power generated by the water turbine; This refers to the volumetric flow rate of water from the reservoir to the turbine. This represents the efficiency coefficient of the water turbine.

[0089] A detailed analysis of the loss mechanisms in pumped hydro storage units is crucial for accurately evaluating system performance. This invention further deepens the discussion of various loss mechanisms in pumped hydro storage units and details how these losses are modeled in efficiency calculations, as well as their impact on overall system performance. The process is as follows:

[0090] The efficiency of pumped hydro storage is affected by various loss mechanisms. The heat loss of pumped hydro storage units mainly originates from the energy conversion process within the hydraulic system. When water flows through the pump and turbine, due to friction, turbulence, and flow resistance, some mechanical energy is converted into heat energy, causing the liquid temperature to rise. This heat energy cannot be converted back into mechanical energy, thus directly reducing the system's efficiency. In the model of this invention, heat loss is modeled using the efficiency coefficient of the hydraulic system, specifically the efficiency reduction factor of the pump and turbine. In actual operation, the efficiency of the pump is typically between 80% and 90%, while the efficiency of the turbine may be slightly higher. In short, heat loss mainly occurs in the hydraulic system; when water flows through the pump and turbine, due to friction and turbulence, some energy is lost as heat. These losses can be mitigated by the efficiency coefficient of the turbine. To express:

[0091] (3);

[0092] in, It is the hydraulic power of a water pump after it converts electrical energy into the potential energy of water. It is the input power of the water pump.

[0093] Electrical losses in pumped-storage units refer to the energy losses incurred by generators and motors during energy conversion due to factors such as resistance, hysteresis, and eddy currents. Specifically, electrical losses occur when the motor converts electrical energy into mechanical energy to drive the pump, and when the generator converts hydraulic mechanical energy into electrical energy. These losses not only affect the system's power generation efficiency but also negatively impact energy conversion during storage and release. In the model of this invention, electrical losses are represented by the efficiency coefficients of the generator and motor. The generator's efficiency coefficient... and the efficiency coefficient of the electric motor It can be represented as:

[0094] (4);

[0095] (5);

[0096] In the formula, It is the generator's electrical power. It is the mechanical power of the generator. It is the electric power of the motor. It refers to the mechanical power of the electric motor.

[0097] After detailed modeling of the various losses mentioned above, this invention uses a comprehensive efficiency formula to calculate the overall efficiency of the pumped storage unit. (Comprehensive efficiency coefficient) It is the product of the efficiency coefficients of each component, representing the overall energy conversion efficiency of the entire pumped storage unit, i.e.:

[0098] (6);

[0099] The impact of various losses on the overall system performance manifests as a decrease in energy conversion efficiency and adverse economic effects. The existence of loss mechanisms means that the actual energy recovery rate is lower than the theoretical value, which directly affects the system's profitability and economic feasibility. Furthermore, higher losses can lead to decreased system stability under high load conditions, especially in scenarios with significant fluctuations in wind and solar power generation. Therefore, this invention places particular emphasis on accurate modeling of losses when optimizing scheduling strategies to ensure that the system can maintain high efficiency and stability under various operating conditions.

[0100] Uncertainty surrounding renewable energy is one of the major challenges facing modern power systems. To effectively address this uncertainty, the model in this invention employs multiple methods, including uncertainty modeling, scenario generation, and stochastic optimization, to ensure the robustness of scheduling and allocation processes under different uncertainty scenarios. This invention uses historical data to statistically analyze wind speed and solar irradiance, establishing corresponding probability distribution models. For wind speed, it is typically assumed to follow a Weibull distribution; for solar irradiance, it is assumed to follow a Beta distribution. The probability density function of wind speed is calculated as follows:

[0101] (7);

[0102] in,

[0103] (8);

[0104] (9);

[0105] In the formula, Wind speed; For shape parameters; For scale parameters; It is a gamma function; Average wind speed; This represents the standard deviation of wind speed, reflecting the degree of fluctuation in wind speed around its average value.

[0106] The probability density function of solar radiation intensity is calculated as follows:

[0107] (10);

[0108] In the formula, Normalized light intensity; and For shape parameters; This is a Beta parameter.

[0109] To capture the volatility of power generation, this invention uses scenario generation technology to create multiple possible wind and photovoltaic power output scenarios. Each scenario is randomly generated according to a different probability distribution, reflecting different weather conditions and time periods.

[0110] (11);

[0111] in, For the first One scenario; and The first The output power of wind turbines and photovoltaic units at time t in each scenario; This represents the total number of scenes.

[0112] This invention maximizes expected return by considering the probability distribution of different scenarios. The goal of stochastic optimization is to maximize expected return, as shown in the following formula:

[0113] (12);

[0114] in, For expected returns; For the scene The probability of; For the scene The resulting profits.

[0115] Probability of each scenario This data, estimated using historical data, reflects the probability of power generation under different weather conditions. By considering these probabilities, this invention can prioritize high-probability scenarios in scheduling schemes, thereby improving the actual operational performance of the system.

[0116] The rapid fluctuations in wind and solar power generation pose a significant challenge to renewable energy power systems, especially under conditions of drastic changes in load demand and power generation. To address these rapid fluctuations, this invention considers a system incorporating a highly integrated transient analysis module. This module employs various physical models and mathematical formulas to provide a detailed model of the power system's dynamic behavior. It not only monitors and analyzes the system's dynamic response in real time but also provides a reliable basis for optimizing scheduling and control strategies.

[0117] Time-domain simulation and frequency-domain analysis enable real-time prediction and evaluation of the dynamic response of power systems under sudden events, including voltage and frequency changes, power oscillations, and system stability. The modular architecture includes a data acquisition module, a dynamic simulation module, a frequency domain analysis module, and a decision support module. The data acquisition module acquires real-time system operating data, including power generation, load demand, voltage, current, and frequency. This data is acquired at high frequencies through phasor measurement units (PMUs) and a supervisory control and data acquisition (SCADA) system to ensure timeliness and accuracy. The dynamic simulation module, based on the system's electrical parameters and topology, simulates the power system's response behavior under disturbances in real time. This module uses differential equations and numerical integration methods to describe in detail the dynamic changes in voltage, frequency, and phase angle. The frequency domain analysis module performs frequency domain analysis on the system's power output and voltage signals using Fast Fourier Transform (FFT) to identify potential resonant frequencies and low-frequency oscillations, and evaluate the system's oscillation modes. Based on a decision support module, this invention provides automated scheduling and control suggestions according to transient analysis results, including starting standby generator sets, adjusting reactive power compensation equipment, and implementing load shedding measures.

[0118] The dynamic simulation model is the core component of the transient analysis module, and it primarily employs the following physical model for system modeling. Considering the mechanical and electromagnetic characteristics of the generator, the transient model of the generator can be expressed as:

[0119] (13);

[0120] In the formula, The moment of inertia of the generator; It is the mechanical angular velocity; and These are the mechanical input torque and the electromagnetic output torque, respectively. is the damping coefficient.

[0121] (14);

[0122] In the formula, It is the stator voltage; It is the stator resistance; It is the stator current; It is a stator inductor; It is the transient electromotive force of the generator; This is the stator-side electrical angular frequency.

[0123] To ensure system stability during periods of sudden power supply and demand fluctuations, the system of this invention utilizes voltage and frequency response modeling to demonstrate its ability to respond to sudden events (such as short-circuit faults and sudden load changes) within a short timeframe. To describe the voltage recovery process after a fault, the voltage response... It can be modeled as:

[0124] (15);

[0125] In the formula, It is the voltage before the fault; It is the attenuation coefficient; It is the damped oscillation frequency; It is the phase angle. This is a time variable representing the period following the occurrence of the fault, and it is a continuous time interval.

[0126] Frequency response is used to describe the change in system frequency after a disturbance. Modeled by the following equation:

[0127] (16);

[0128] in, It is the maximum value of the frequency offset; It is the natural oscillation frequency.

[0129] To improve the system's robustness in the face of extreme conditions, this invention, based on a decision support module, provides automated scheduling and control recommendations according to transient analysis results, including measures such as starting standby generators, adjusting reactive power compensation equipment, and implementing load shedding. Through voltage and frequency response modeling, it possesses the ability to respond to sudden events in a short period. When extreme weather (such as storms, strong winds, or intense sunlight) is predicted, standby generators are started in advance or the reserve capacity of the energy storage system is increased. These strategies enhance the system's regulation margin, ensuring that system balance can still be maintained in the event of emergencies. Through the above detailed modeling and analysis, the transient analysis module of this invention can effectively handle the rapid changes in load demand and power generation processes of wind and photovoltaic power generation, ensuring the dynamic stability of the power system. These technical details are fully described in the revised draft.

[0130] 2. Construction of Optimal Scheduling Model for Combined Generation System

[0131] Pumped storage units can serve as a special type of electrical load and store excess electricity in the power system, significantly reducing wind and solar power curtailment, ensuring economic benefits, and playing a crucial role in stabilizing the power grid. This invention considers the power generation revenue and pollutant emission penalty costs of wind power, photovoltaic power, thermal power units, and pumped storage units, establishing a scheduling model with the goal of maximizing the total revenue of the combined power generation system. The model's objective function includes the power generation revenue and operating costs of various power sources, the pumping cost of pumped storage, and the power generation cost and pollutant emission penalty costs of thermal power units. Combining equation (12), the objective revenue function that maximizes the expected revenue can be expressed as:

[0132] (17);

[0133] In the formula, Revenue generated by a combined power generation system that includes wind power, photovoltaic power, thermal power units and pumped storage units; For the operating costs of wind power and photovoltaic units; The cost of pumping water for pumped storage units; The cost of generating electricity from thermal power units; The punitive costs of pollutant emissions.

[0134] Revenue from combined power generation systems including wind power, solar power, thermal power units, and pumped hydro storage It can be represented as:

[0135] (18);

[0136] In the formula, For photovoltaic feed-in tariffs; For wind power grid connection price; Grid connection price for pumped hydro storage power generation: The on-grid electricity price for thermal power units; Let t be the power transmitted from the photovoltaic unit to the grid. Let t be the power delivered to the grid by the wind turbine. for t The power that is constantly pumped and stored in water and transmitted to the power grid; for t The power transmitted from thermal power units to the power grid at all times.

[0137] Operating costs of wind power and solar power It can be represented as:

[0138] (19);

[0139] In the formula, and These are the power generation cost coefficients for wind turbines and photovoltaics, respectively; For a moment t Inner The actual power generation of the typhoon turbines; For a moment t Inner The actual power generation of the photovoltaic system in Taiwan; The number of wind turbine units; This refers to the number of photovoltaic (PV) units.

[0140] Pumping cost of pumped storage unit It can be represented as:

[0141] (20);

[0142] In the formula, The pumping electricity price for pumped storage; Pumping power for pumped storage.

[0143] Power generation cost of thermal power units It can be represented as:

[0144] (twenty one);

[0145] In the formula: For the first t In the period of the first The power output of the thermal power unit; The total time of the scheduling cycle; This refers to the number of thermal power units. , , For the first The power generation cost coefficient of a thermal power unit; and For the first The startup cost and startup status of a thermal power unit.

[0146] During system scheduling, the planned output of thermal power units with high pollutant emission characteristic coefficients should be reduced as much as possible, while the output of thermal power units with low pollutant emission characteristic coefficients should be increased. Therefore, the penalty cost of the pollutant emission coefficient of thermal power units should be considered. It can be represented as:

[0147] (twenty two);

[0148] In the formula, , and For the first Pollutant emission coefficients of thermal power units in Taiwan.

[0149] Constraints:

[0150] Taking into account the output limitations of various units in the combined power generation system, power balance constraints, thermal power unit ramping rate, and pumped storage unit output constraints, the rationality and feasibility of the dispatching process are ensured.

[0151] (1) Power balance constraint

[0152] The mathematical relationship between the output of thermal power units, wind power clusters, photovoltaic power, and energy storage charging and discharging power in a combined power generation system can be expressed as follows:

[0153] (twenty three);

[0154] In the formula, for t In the moment The power output of the thermal power unit; for t Within a certain time period The actual power generation of the typhoon turbines; for t Within a certain time period The actual power generation of the photovoltaic system in Taiwan; Pumped storage pumping power; for t The power that is constantly pumped and stored in water and transmitted to the power grid; For load power; The number of wind turbine units, The number of photovoltaic units.

[0155] (2) Upper and lower limits of wind turbine output constraints

[0156] (twenty four);

[0157] In the formula: , These represent the upper and lower limits of wind turbine output, respectively.

[0158] (3) Upper and lower limits of photovoltaic output constraints

[0159] (25);

[0160] In the formula: , These represent the upper and lower limits of photovoltaic unit output, respectively.

[0161] (4) Upper and lower limits of thermal power unit output constraints

[0162] (26);

[0163] In the formula: , These represent the upper and lower limits of the output of thermal power units, respectively.

[0164] (5) Climbing constraints of thermal power units

[0165] The amount of power that different thermal power units can increase or decrease per unit time depends on their generator operating parameters. Therefore, the output ramp-up rate of thermal power units must be considered when scheduling their output. Different thermal power units have different ramp-up rates, and it is impossible to significantly increase or decrease output per unit time. Therefore, the ramp-up constraints of each thermal power unit must be considered in the scheduling plan, resulting in:

[0166] (27);

[0167] (28);

[0168] (29);

[0169] (30);

[0170] In the formula: For the first t-1 In the moment The power output of the thermal power unit; for t Within a certain time period Start-up status of the thermal power unit; for t Within a certain time period The thermal power unit is in the off state; for t Within a certain time period Operating status of the thermal power unit; for t-1 Within a certain time period Operating status of the thermal power unit; and They are respectively t Within a certain time period The upward and downward ramp rates of the thermal power units during the dispatch cycle.

[0171] (6) Output constraints of pumped storage units

[0172] (31);

[0173] In the formula, and These are the maximum and minimum outputs of the pumped storage unit, respectively.

[0174] (7) Grid acceptance capacity constraints:

[0175] (32);

[0176] in , These are the upper and lower limits of the power that the power grid can accept, respectively. The power supplied to the grid by the photovoltaic system; The power transmitted from wind turbines to the power grid; The power supplied to the grid by the pumped storage unit. The power transmitted from thermal power units to the power grid.

[0177] As an effective energy storage method, pumped hydro storage can effectively regulate the dynamic balance between power generation, supply, and consumption in the power system. The integration of pumped hydro storage units can, to a certain extent, control the charging and discharging process according to the needs of the power grid, thus forming an integrated wind-solar-storage system. In this integrated system, the platform control mode is more convenient and flexible, and tends to become more intelligent with technological advancements. This platform integrates a model predictive control (MPC) algorithm, using the system's mathematical model to predict future system behavior and generating the optimal control strategy based on the prediction results. In this invention, MPC focuses on optimizing the scheduling and energy management of the energy storage system to maximize economic benefits and ensure grid stability. In each control cycle, the MPC optimization objective function can be expressed as:

[0178] (33);

[0179] In the formula, Indicates the system status; For control input; As a reference trajectory, it is in the first... The system reference state vector at each discrete control moment; and This is the weight matrix; To predict the length of the time domain; This is the time step number.

[0180] The state-space model of a system is typically expressed as the following linear or nonlinear discrete-time system:

[0181] (34);

[0182] (35);

[0183] in, , , and These are system matrix parameters; For external disturbances (such as load changes or weather changes), it indicates the first time... The disturbance vector acting on the system at each discrete control moment; For system output, indicating the number of times... The system output vector at each discrete control moment; In the first The system state vector at the nth discrete control time represents the state vector at the nth discrete control time. The control input vector applied at each discrete control moment; This is the perturbation input matrix.

[0184] By accessing real-time data from the power grid and weather forecasts, the MPC (Magnetic Phasor Control) system can dynamically adjust its control strategy to cope with changes in external conditions. The MPC algorithm acquires key grid parameters in real time, such as current load levels, generation output, voltage, and frequency. This data is obtained through a SCADA system or a phasor measurement unit (PMU) and used to adjust the state variables in the MPC. and control variables Weather forecast data (such as wind speed and solar irradiance) has a direct impact on renewable energy power generation. Therefore, in MPC, this invention estimates future power generation based on predicted weather data and adjusts the operating strategy of the energy storage system accordingly. The impact of weather data on the energy storage system can be expressed as:

[0185] (36);

[0186] In the formula, To predict power generation; and It is divided into wind speed prediction and solar irradiance (also known as light intensity) prediction at time t; To include meteorological forecasts , A nonlinear function mapped to the predicted power generation value.

[0187] Based on this predictive data, MPC can pre-schedule energy storage systems to cope with potential power fluctuations. Adaptive control is one of the key features of MPC, playing a crucial role, especially in responding to sudden system changes. The system platform's MPC algorithm is designed to immediately adjust the control strategy upon detecting sudden system events (such as load spikes or power generation drops) to ensure stable system operation. Within each control cycle, the MPC algorithm monitors the system state and detects any disturbances exceeding normal limits (such as frequency shifts or voltage fluctuations). Once a disturbance is detected, MPC recalculates the optimal control strategy based on the new state. The disturbance model and adaptive control strategy can be represented as:

[0188] (37);

[0189] (38);

[0190] in, For the first State deviation at discrete control moments; For the first The control adjustment amount at each discrete control moment; For the first Disturbance at each discrete control moment; For the first Adaptive control input at discrete control moments; An adaptive gain matrix is ​​used to ensure that the system can quickly recover to normal operating conditions after a disturbance occurs; For the first The adaptive control input adjustment parameters are set at each discrete control moment.

[0191] To cope with sudden changes, the MPC algorithm employs a real-time optimization and iterative update strategy. By solving the optimization problem online, MPC can generate new control strategies at each time step to address the latest system state and external conditions. The iterative optimization formula is:

[0192] (39);

[0193] in, For the first Control input at discrete control moments; In the first The system state vector at each discrete control moment; In the first The system reference state vector at each discrete control moment.

[0194] In each control cycle, MPC iteratively optimizes and calculates new control inputs to ensure optimal system operation under dynamically changing conditions.

[0195] 3. Transformation of multi-objective problems

[0196] To address the trade-off between maximizing economic benefits and minimizing environmental emissions, this invention is based on... The constraint method transforms the multi-objective problem of pollutant emission costs and economic benefits into a single-objective problem. It optimizes the overall objective function by maximizing economic benefits, with pollutant emission levels as the constraint condition. Mathematically, this is expressed as:

[0197] (40);

[0198] In the interval [0,1], a fuzzy decision-maker is used to assign fuzzy membership degrees to each point, and the membership function is:

[0199] (41);

[0200] In the formula, The membership value of the o-th objective function. and For the first The minimum and maximum values ​​that an objective function can reach in the feasible solution space. For the first The actual value of the objective function under the current solution. For the first The upper limit of satisfaction (i.e., the expected value) of an objective function. For the first The upper limit of tolerance (i.e. the maximum allowed value) for each objective function.

[0201] The optimal compromise solution is obtained through the min-max method, and the minimum solution is selected. The maximum value is taken as the best compromise. This represents the membership degree or satisfaction level of the first objective function under the current solution. This represents the membership degree or satisfaction level of the second objective function under the current solution.

[0202] 4. Solution method for scheduling model based on improved bat algorithm

[0203] This invention employs an improved Bat Algorithm (BA) to solve the scheduling model transformed into a single-objective problem. Traditional Bat Algorithms suffer from slow convergence and a tendency to get trapped in local optima when dealing with large-scale optimization problems. Therefore, this invention improves upon the Bat Algorithm as follows:

[0204] (1) Chaotic mapping initialization: Population initialization is performed by chaotic mapping method to improve the coverage of the initial solution space.

[0205] (2) Adaptive step size update: An adaptive step size update strategy is introduced to accelerate the convergence speed.

[0206] (3) Cauchy mutation factor: Cauchy mutation factor is added during the search process to enhance the algorithm’s ability to escape local optima.

[0207] Based on the biomimetic principle of bats, the bat algorithm comprises three elements: the search pulse frequency, the pulse intensity, and the emitted pulse frequency. The flight speed is determined by the search pulse frequency, while the probability of receiving a position update is determined by the pulse intensity and frequency. The steps of the bat algorithm are as follows:

[0208] Step 1: Set the bat population size Maximum impulse intensity Maximum pulse frequency Upper limit of pulse frequency Pulse frequency lower limit Sound intensity attenuation coefficient Frequency increase coefficient Set the dimension of the position vector and the maximum number of iterations.

[0209] Step 2: Initialize the bat position Simultaneously, based on the objective function, the fitness value of the bat's location is calculated, and the individual in the optimal location is found. The bat position is initialized as shown in equation (42).

[0210] (42);

[0211] In the formula: , Indicates the upper and lower limits of the area where the bat is located; Indicates the random generation of a one-dimensional element. A vector matrix with values ​​in the range (0, 1).

[0212] Step 3: Initialize the pulse frequency of the unit bat, calculate the flight speed of the unit bat, update the position of the bat, and update the formulas as shown in equations (43) to (45).

[0213] (43);

[0214] (44);

[0215] (45);

[0216] In the formula, It is a random factor that is uniformly distributed in the interval (0, 1); , For the first bat in and Flight speed per round; , They represent the first Only bats and The position of the round.

[0217] Step 4: In each iteration, generate random numbers for the unit bat. ,like If so, then the current best solution is selected, and a local perturbation is performed. Wherein, For the first Only bats The pulse frequency of each round. The local perturbation is shown in equation (46).

[0218] (46);

[0219] Step 5: Calculate the bat's new fitness after the perturbation. If the new fitness is better than its optimal fitness or... The new position after the disturbance is used to replace the old position and is saved. At the same time, the pulse frequency and intensity are updated, as shown in equations (47) and (48).

[0220] (47);

[0221] (48);

[0222] In the formula, For the first Only bats The pulse frequency of each round; For the first Only bats The intensity of the pulse sound in each round; For the first Only bats The intensity of the pulse sound in each round.

[0223] Step 6: Stop the search when the maximum number of iterations is reached, and output the position and fitness value of the bat corresponding to the global optimal solution; otherwise, jump back to step 3 to continue the search.

[0224] As the number of iterations increases in the basic Bat Algorithm, the individual differences within the population become increasingly pronounced in later iterations, leading to a decrease in population diversity, which eventually approaches zero. Therefore, this invention employs chaotic mapping for population initialization to improve the coverage of the initial solution space. The Bat Algorithm is improved through adaptive step-size updates and the introduction of the Cauchy mutation inverse cumulative distribution function. The improvement strategy is as follows:

[0225] Population initialization operation using chaotic mapping

[0226] The Bat Algorithm initially uses a random method for population initialization, which cannot cover the entire solution space. Therefore, this invention employs a chaotic mapping method for population initialization to improve the coverage of the initial solution space. The calculation formula is as follows:

[0227] (49);

[0228] In the formula, , It is a chaotic sequence; To initialize the population dimension; To initialize the population size. Facing Perform the inversion operation to obtain the solution space initialization population pairs, calculated as follows:

[0229] (50);

[0230] In the formula; Indicates the first Only bats The position of the round, and These represent the minimum and maximum values ​​of the variable's range of values, respectively.

[0231] Adaptive step size update

[0232] Based on equations (47) and (48) to update bat velocity, a bat velocity update method based on adaptive step size is proposed. The algorithm has a relatively large initial step size, which improves the convergence speed. In the later stages of the algorithm, the step size becomes smaller, and the search becomes more refined.

[0233] (51);

[0234] (52);

[0235] In the formula, For the minimum step size, The maximum number of iterations, It is a regulating factor.

[0236] Introducing the Cauchy inverse cumulative distribution function

[0237] Cauchy mutation originates from the Cauchy distribution of a continuous probability distribution. Appropriate random variables are used to generate the Cauchy mutation factor, which is then applied in step 3 of the Bat Algorithm. The speed variable in each round is modified by adding a Cauchy mutation factor to change the bat's flight speed and enhance its ability to escape local optima. The mutation formulas are shown in equations (53)-(54).

[0238] (53);

[0239] (54);

[0240] In the formula, It is a uniformly distributed random number within the interval [0, 1]. For the coefficient vector, It is a vector that decreases linearly from 2 to 0; Let be a uniformly distributed random vector within the interval [0, 1]. The solution process of the scheduling model based on the improved bat algorithm is as follows: Figure 2 As shown.

[0241] 5. Examples

[0242] The combined power generation system consists of three thermal power units, one pumped storage unit, one wind farm, and one photovoltaic unit. The installed capacity of each power source is shown in Table 1. The feed-in tariff is shown in Table 2. The parameters of the thermal power units are shown in Table 3. The wind and photovoltaic power output curves are shown below. Figure 2 As shown.

[0243] Table 1 Model Parameters

[0244]

[0245] Table 2 Grid Connection Price

[0246]

[0247] Table 3 Parameters of Thermal Power Units

[0248]

[0249] To verify the robustness of the Improved Bat algorithm, the Bat algorithm, PSO algorithm, and the improved Bat algorithm were used in three typical test functions. The results of each algorithm were tested and compared in MATLAB software. These six test functions are:

[0250] Schaffer function: (55);

[0251] Ackley function: (56);

[0252] Salomon function: (57);

[0253] For vectors The Each component, when considering the two-dimensional case. , These represent variables in the first and second dimensions, respectively. Pi is a constant. , used to control the periodicity of the cosine term; is the dimension of the Salomon function, i.e., the number of decision variables.

[0254] Different values ​​were set for the maximum number of iterations and the population size for the three different test functions, as shown in Table 4.

[0255]

[0256] For the optimal and average optimal value analysis, the specific parameters were set as follows: 200 iterations and a population size of 40. To avoid errors caused by the randomness of the algorithm, the three functions were run independently 30 times on different dimensions, and the worst, best, and average values ​​were taken. The results are shown in Table 5.

[0257] As can be seen from Table 5, compared with the BA and PSO algorithms, the ImprovedBat algorithm has better overall optimization performance when solving low-dimensional test functions. The optimal value obtained is close to the theoretical value, and the algorithm has good robustness, which further illustrates that the ImprovedBat algorithm has good adaptability and robustness.

[0258] We will now conduct simulations to compare the Bat Algorithm, Particle Swarm Optimization (PSO) Algorithm, and an improved Bat Algorithm. The maximum number of iterations for each algorithm is set to 200, and the population size is set to 100. The simulation convergence curves are shown below. Figure 4 As shown.

[0259] from Figure 4 It can be seen that the improved bat algorithm used in this invention can approach the optimal solution with fewer iterations during the optimization process. Its optimization efficiency is significantly better than the traditional bat algorithm and particle swarm optimization algorithm. This verifies the necessity and effectiveness of the improved bat algorithm for optimizing scheduling solutions. The revenue curves of the combined power generation system at each time step after using the bat algorithm, particle swarm optimization algorithm, and improved bat algorithm are shown below. Figure 5 As shown.

[0260] from Figure 5 It can be seen that before 3 PM, the improved Bat Algorithm generally outperformed both the original and improved Bat Algorithms in terms of overall return. While its overall return was lower than the other two algorithms at 3 PM and later, its overall return was still slightly higher than the other two algorithms. After optimization, the Bat Algorithm improved the overall return by 2.7% compared to the Particle Swarm Optimization algorithm, and the improved Bat Algorithm improved the overall return by 21.9% compared to the original Bat Algorithm. The scheduling results of thermal power units based on the method of this invention are shown in Table 6.

[0261] As shown in Table 6, the output changes of Unit 3 are highly sensitive to the introduction of pollution discharge costs because this unit has a high pollutant emission coefficient. Introducing penalty costs will reduce the scheduling of Unit 3, which has the highest pollutant emission coefficient. During this process, the thermal power units do not need to frequently adjust their output plans, and their equivalent system load remains essentially unchanged. This not only significantly reduces the additional impact of traditional large-scale renewable energy grid connection on the system but also fully utilizes pumped storage units to participate in the system's peak shaving and valley filling tasks in a more rational manner.

[0262] pass The Pareto optimal solution of the constraint method is as follows Figure 6As shown, after 20 iterations, the Pareto optimal solution is obtained, with the maximum membership function being 0.750, corresponding to the 13th solution. Therefore, considering the min-maximum fuzzy satisfaction method, the 13th solution is chosen as the optimal compromise solution.

[0263] Simulation results verify that the improved bat algorithm proposed in this invention can approach the optimal solution with fewer iterations during the optimization process, significantly improving scheduling benefits compared to the traditional bat algorithm and particle swarm optimization (PSO). Simulations show that the optimized scheduling scheme not only improves the capacity for renewable energy absorption but also effectively reduces the scheduling burden on thermal power units.

Claims

1. A wind-solar-pumped hydro storage joint power generation scheduling method based on an improved bat algorithm, characterized in that: The steps include the following steps, and the following steps are performed in sequence: Step 1: Construct a combined power generation system that includes wind power, photovoltaic power, thermal power and pumped storage units, wherein the pumped storage units include water pumps and water turbines; Step 2: Analyze and model the loss mechanism of pumped storage units; Step 3: Perform uncertainty modeling, scenario generation, and stochastic optimization of the joint generation system to ensure the robustness of the scheduling and allocation process of the joint generation system under different uncertainty scenarios; Step 4: The combined power generation system also includes a transient analysis module. The transient analysis module models the dynamic behavior of the power system using physical models and mathematical formulas, and is used to monitor and analyze the dynamic response of the combined power generation system in real time, and to provide a basis for dispatching. Step 5: Establish a scheduling model with the goal of maximizing the total revenue of the combined power generation system. The scheduling model includes the power generation revenue and operating costs of various generator units, the pumping cost of pumped storage, and the power generation cost and pollutant emission penalty cost of thermal power units. Conditional constraints are imposed on the combined power generation system to ensure the rationality and feasibility of the scheduling process. Step Six: Solve the scheduling model using the improved bat algorithm to optimize the output scheduling of each generator unit; The objective function of the scheduling model considers maximizing the expected return for different scenario probability distributions, and the formula is as follows: ； In the formula, To maximize the expected revenue of the combined power generation system; For the first One scenario; For the scene The probability of; For the scene The following benefits; Total number of scenes; Revenue generated by a combined power generation system that includes wind power, photovoltaic power, thermal power units and pumped storage units; For the operating costs of wind power and photovoltaic units; The cost of pumping water for pumped storage units; The cost of generating electricity from thermal power units; The punitive costs of pollutant emissions; Among them, the power generation revenue of the combined power generation system including wind power, photovoltaic, thermal power units and pumped storage units It is the sum of the following parts: The feed-in tariff for solar power is multiplied by the output power of the solar power unit. The on-grid electricity price of wind power multiplied by the output power of the wind turbine unit; The grid-connected electricity price of pumped hydro storage is multiplied by the output power of the pumped hydro storage unit. The on-grid electricity price of thermal power units multiplied by the output power of thermal power units; The scenario involves using scenario generation technology to create multiple possible wind power and photovoltaic unit power generation output scenarios. Each scenario is randomly generated based on the probability distribution of wind speed and solar irradiance, reflecting power generation fluctuations under different weather conditions and time periods, thereby improving the robustness and accuracy of the scheduling model. A scenario As shown below: ； in, and The first The output power of wind turbines and photovoltaic units at time t in each scenario; The transient analysis module establishes dynamic simulation models for the following components: Based on the mechanical and electromagnetic characteristics of the generator, the transient model of the generator is represented as follows: ； In the formula, The moment of inertia of the generator; It is the mechanical angular velocity; and These are the mechanical input torque and the electromagnetic output torque, respectively. The damping coefficient; ； In the formula: It is the stator voltage; It is the stator resistance; It is the stator current; It is a stator inductor; It is the transient electromotive force of the generator; The stator-side electrical angular frequency; To ensure system stability during sudden changes in power supply and demand, voltage response and frequency response modeling are used to address sudden fault events. Voltage response is used to describe the voltage recovery process after a fault. The model is as follows: ； In the formula, It is the voltage before the fault; It is the attenuation coefficient; It is the damped oscillation frequency; It is the phase angle; The time variable is the period after the fault occurs; Frequency response The change in system frequency after a disturbance is described by the following equation: ； in, It is the maximum value of the frequency offset; It is the natural oscillation frequency; Pumped hydro storage can effectively regulate the dynamic balance between power generation, supply, and consumption in the power system. The integration of pumped hydro storage units allows for control of the charging and discharging process according to the grid's needs, thus forming an integrated wind-solar-storage system. This integrated system utilizes a platform control mode, which integrates a Model Predictive Control (MPC) algorithm. This platform uses the system's mathematical model to predict future system behavior and generates the optimal control strategy based on the prediction results. MPC focuses on optimizing the scheduling and energy management of the energy storage system to maximize economic benefits and ensure grid stability. In each control cycle, the MPC objective function is expressed as: ； In the formula, Indicates the system status; For control input; As a reference trajectory, it is in the first... The system reference state vector at each discrete control moment; and This is the weight matrix; To predict the length of the time domain; The time step number; The state-space model of a system is typically expressed as the following linear or nonlinear discrete-time system: ； ； in, , , and These are system matrix parameters; For external disturbances (such as load changes or weather changes), it indicates the first time... The disturbance vector acting on the system at each discrete control moment; For system output, indicating the number of times... The system output vector at each discrete control moment; In the first The system state vector at the nth discrete control time represents the state vector at the nth discrete control time. The control input vector applied at each discrete control moment; The perturbation input matrix; These represent the state variables and control variables in MPC, respectively. In MPC, future power generation is estimated using predicted weather data, and the operating strategy of the energy storage system is adjusted accordingly. The impact of weather data on the energy storage system is represented as follows: ； In the formula, To predict power generation; and It is divided into wind speed prediction and solar irradiance (also known as light intensity) prediction at time t; To include meteorological forecasts , A nonlinear function mapped to the predicted power generation value; During each control cycle, the MPC algorithm monitors the system state and detects any disturbances exceeding a set range. Once a disturbance is detected, MPC recalculates the optimal control strategy based on the new state. The disturbance model and adaptive control strategy are represented as follows: ； ； in, For the first State deviation at discrete control moments; For the first The control adjustment amount at each discrete control moment; For the first Disturbance at each discrete control moment; For the first Adaptive control input at discrete control moments; An adaptive gain matrix is ​​used to ensure that the system can quickly recover to normal operating conditions after a disturbance occurs; For the first Adaptive control input adjustment parameters at discrete control moments; To cope with sudden changes, the MPC algorithm employs a real-time optimization and iterative update strategy. By solving the optimization problem online, MPC can generate new control strategies at each time step to address the latest system state and external conditions. The iterative optimization formula is as follows: ； in, For the first Control input at discrete control moments; In the first The system state vector at each discrete control moment; In the first The system reference state vector at each discrete control moment; The improved bat algorithm includes the following steps: a. Initialize the population using chaotic mapping to increase the coverage of the initial solution space; The Bat Algorithm initially uses a random method for population initialization, which cannot cover the entire solution space. Therefore, a chaotic mapping method is used for population initialization to improve the coverage of the initial solution space. The calculation formula is as follows: ； In the formula, It is a chaotic sequence. ; For the rounds of the sequence; To initialize the population dimension; To initialize the population size, facing Perform the inversion operation to obtain the solution space initialization population pairs, calculated as follows: ； In the formula; and These are the minimum and maximum values ​​of the variable's range of values, respectively. Indicates the first Only bats Position of the round; b. Introduce an adaptive step size update method to accelerate the convergence process of the algorithm; The adaptive step size update method uses a large initial step size, which improves the convergence speed. In the later stages of the algorithm, the step size becomes smaller, and the search becomes more refined. The formula for the bat velocity update method based on adaptive step size is as follows: ； ； In the formula, , The first Only bats and Flight speed per round; For adaptive step size; For individuals in the optimal position; For the minimum step size, The maximum number of iterations, As a regulating factor; It is a natural constant; c. Add a Cauchy mutation factor during the search process to enhance the algorithm's ability to escape local optima; right t The bat's flight speed variable at time t is modified by adding a Cauchy mutation factor to change the bat's flight speed and enhance its ability to escape local optima. The mutation formula is shown below: ； ； In the formula, For the first i Only bats Position of the round; It is a uniformly distributed random number within the interval [0, 1]. For the coefficient vector, It is a vector that decreases linearly from 2 to 0; Let be a uniformly distributed random vector within the interval [0, 1].

2. The wind-solar-pumped hydro storage joint power generation scheduling method based on the improved bat algorithm according to claim 1, characterized in that: The specific method for analyzing and modeling the loss mechanism of pumped storage units in step two is as follows: The loss mechanism includes hydraulic losses and electrical losses. Hydraulic losses are represented by the efficiency coefficients of the water pumps and turbines, while electrical losses are represented by the efficiency coefficients of the generators and motors. Finally, the overall energy conversion efficiency of the pumped storage system is calculated using a comprehensive efficiency coefficient formula, which is as follows: ； in, The overall efficiency coefficient, This represents the efficiency coefficient of the water turbine. It is the efficiency coefficient of the generator. It is the efficiency coefficient of the electric motor; ； ； ； In the formula, It is the hydraulic power of a water pump after converting electrical energy into the potential energy of water. It is the input power of the water pump. It is the generator's electrical power. It is the mechanical power of the generator. It is the electric power of the motor. It refers to the mechanical power of the electric motor.

3. The wind-solar-pumped hydro storage joint power generation scheduling method based on the improved bat algorithm according to claim 1, characterized in that: The uncertainty modeling in step three includes using Weibull and Beta distributions to model the uncertainty of wind power and photovoltaic power generation, and using historical data statistical analysis to obtain the probability density functions of wind speed and light intensity. The probability density function of wind speed is: ； in, ； ； In the formula, Wind speed; For shape parameters; For scale parameters; It is a gamma function; Average wind speed; This represents the standard deviation of wind speed, reflecting the degree to which wind speed fluctuates around its average value. The probability density function of solar radiation intensity is: ； In the formula, Normalized light intensity; and For shape parameters; These are Beta parameters.

4. The wind-solar-pumped hydro storage joint power generation scheduling method based on the improved bat algorithm according to claim 1, characterized in that: The transient analysis module also includes time-domain simulation and frequency-domain analysis modules. It simulates the changes in voltage, frequency and phase angle of the system through differential equations and numerical integration methods, and uses Fast Fourier Transform (FFT) to perform frequency-domain analysis on the established dynamic simulation model to evaluate the stability and oscillation modes of the system.

Images