Multi-target regulation and control method based on cascade water-wind-light-pumped storage multi-energy complementation

Abstract

The invention discloses a multi-target regulation and control method based on cascade water-wind-light-pumped storage multi-energy complementation, and the method comprises the following steps: collecting, screening and analyzing the comprehensive data of the environmental factors of a wind power generation group and a photovoltaic power generation group, and obtaining key impact factors; building a short-term output prediction model, optimizing hyper-parameters of the short-term output prediction model, and inputting the key influence factors and historical output data of the wind power generation group and the photovoltaic power generation group into the optimized short-term output prediction model to obtain a short-term output prediction result of the wind power generation group and the photovoltaic power generation group; in combination with the prediction result, establishing a constraint system of the hybrid renewable energy system, and constructing an optimization model by taking improvement of the system stability as an optimization target; and solving a multi-objective optimization problem of the optimization model to obtain scheduling strategies under different scenes and system operation mechanisms and a coordinated operation mode of various energy sources in the hybrid renewable energy system.

Description

Technical Field

[0001] The present invention belongs to the technical field of optimal operation of power systems, and particularly relates to a multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity. Background Art

[0002] With the continuous growth of global energy demand and the exacerbation of environmental problems, the issue of energy security has become the focus of global attention. Against this background, countries have gradually regarded renewable energy as an important means to address energy crises and ecological environment pressures. However, some renewable energies are greatly affected by weather conditions, showing obvious intermittency and volatility, which brings a series of challenges to the dispatching and energy storage management of power grids; when the power generation of photovoltaic and wind power fluctuates greatly due to weather conditions, hydropower can quickly supplement the power supply gap of the power grid, thereby improving the stability and response speed of the system. However, with the rapid growth of the installed capacity of wind power and photovoltaic power, traditional hydropower also faces new challenges in meeting the regulation requirements of these intermittent energies. Especially when the large-scale wind and light output fluctuates significantly, the regulation capacity of traditional hydropower may be insufficient.

[0003] For the above reasons, it is necessary to integrate various renewable energies such as wind power, photovoltaic power, and hydropower, utilize the complementarity between different energies, reduce the volatility risk brought by a single energy source, improve the stability and reliability of the system. At the same time, gradually transform the cascade hydropower system and add pumping devices to improve the regulation capacity and enhance the flexibility of power dispatching. Summary of the Invention

[0004] In order to overcome the deficiencies of the prior art, the present invention provides a multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity. The multi-objective regulation method can alleviate the impact of load fluctuations on the power grid, ensure the security and stability of the power grid under the background of high proportion of renewable energy penetration, and lay a foundation for the grid connection of large-scale clean energy in the future.

[0005] The technical solution of the present invention to solve the above technical problems is:

[0006] A multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity, comprising the following steps:

[0007] Step S1: Collect comprehensive data on environmental factors affecting the wind power generation group and the photovoltaic power generation group, screen and analyze the collected environmental factors, and obtain the key influencing factors that have the greatest impact on the wind power generation efficiency and the photovoltaic power generation efficiency;

[0008] Step S2: Build a short-term output prediction model for the wind power generation group and the photovoltaic power generation group. Optimize the hyperparameters of the short-term output prediction model through an optimization algorithm. Input the key influencing factors and the historical output data of the wind power generation group and the photovoltaic power generation group into the optimized short-term output prediction model to obtain the prediction results of the short-term output of the wind power generation group and the photovoltaic power generation group;

[0009] Step S3: Combine the prediction results of the short-term output of the wind power generation group and the photovoltaic power generation group to establish a constraint system for the hybrid renewable energy system, and construct an optimization model for the hybrid renewable energy system with the goal of improving the stability of the hybrid renewable energy system;

[0010] Step S4: Solve the optimization model through an improved multi-objective optimization algorithm to obtain the scheduling strategies under different scenarios and the operation mechanisms of the hybrid renewable energy system and the coordinated operation methods of various types of energy in the hybrid renewable energy system.

[0011] In a preferred embodiment of the present invention, the specific steps of step S1 are as follows:

[0012] S11: Collect the environmental data of the areas where the wind power stations and photovoltaic power stations are located, with a time span of the recent 3 months; the types of the collected environmental data include: humidity, solar radiation intensity, wind speed, air pressure, and precipitation; the data is collected hourly;

[0013] S12: Process the collected environmental data, including removing missing values, dealing with outliers, standardizing timestamps, and unifying data;

[0014] S13: Based on the correlation analysis method, by setting the threshold of the correlation coefficient, screen out the environmental factors that are most relevant to the wind power generation efficiency or the photovoltaic power generation efficiency; by identifying the linear correlation between environmental factors, obtain the key influencing factors that have the greatest impact on the wind power generation efficiency or the photovoltaic power generation efficiency.

[0015] In a preferred embodiment of the present invention, the specific steps of step S13 are as follows:

[0016] S131: Calculate the Pearson correlation coefficient to measure the linear relationship between each environmental factor and the wind power generation efficiency or the photovoltaic power generation efficiency, generate a complete correlation matrix, and view the correlation between all environmental factors and the wind power generation efficiency or the photovoltaic power generation efficiency;

[0017] S132: Set a preset threshold range to screen out the environmental factors that have a strong correlation with the wind power generation efficiency or the photovoltaic power generation efficiency; when the Pearson correlation coefficient is within the preset threshold range, it is considered that the environmental factor has a significant linear relationship with the wind power generation efficiency or the photovoltaic power generation efficiency;

[0018] S133: Identify the multicollinearity among various environmental factors, and use the variance inflation factor VIF to check for the existence of multicollinearity; when the VIF of a certain environmental factor is greater than the set value, it indicates that there is a strong linear correlation between this environmental factor and other environmental factors, so this environmental factor needs to be deleted.

[0019] In a preferred embodiment of the present invention, in step S2, the steps for constructing the short-term output prediction model are as follows:

[0020] S21: Construct an LSTM model, and optimize the key hyperparameters in the LSTM model through the particle swarm optimization algorithm to obtain the short-term output prediction model;

[0021] S22: Input the key influencing factors and the historical output data of the wind power group and the photovoltaic power generation group into the short-term output prediction model to obtain the prediction results of the short-term output of the wind power group and the photovoltaic power generation group;

[0022] S23: Judge the prediction results. If the prediction results meet the accuracy requirements, use these prediction results as the final prediction results and end the prediction process; if the prediction results do not meet the accuracy requirements, update the particles in the particle swarm optimization algorithm, and optimize the key hyperparameters in the LSTM model through the particle swarm optimization algorithm until the final prediction results meet the accuracy requirements.

[0023] In a preferred embodiment of the present invention, in step S21, the key hyperparameters of the LSTM model include the learning rate, the number of training times, and the number of neurons.

[0024] In a preferred embodiment of the present invention, in step S21, the steps for the particle swarm optimization algorithm to optimize the key hyperparameters of the LSTM model are as follows:

[0025] S211: Initialize the particle swarm, set the initial parameters of the particle swarm, including the position and velocity of the particles, and at the same time define the inertia weight factor, the learning factor, and the maximum number of iterations;

[0026] S212: Update the particles. According to the current position and velocity of the particles, and in combination with the velocity update formula and the position update formula of the particle swarm optimization algorithm, gradually iterate and update the particles;

[0027] S213: Use the optimal solution obtained by the particle swarm optimization algorithm to adjust the key hyperparameters of the LSTM model.

[0028] In a preferred embodiment of the present invention, in step S3, the hybrid renewable energy system includes a wind power generation group, a photovoltaic power generation group, a three - stage cascade hydropower station, a pumping device between each level of hydropower stations, and a central control system; the interaction mode of this hybrid renewable energy system is as follows: the pumping device between each level of hydropower stations is powered by the wind power generation group and the photovoltaic power generation group; the wind power generation group and the photovoltaic power generation group can be optimally selected to supply the pumping device between each level of hydropower stations or be connected to the power grid; the power generated by the three - stage cascade hydropower station is directly connected to the power grid; the pumping device between each level of hydropower stations pumps the remaining water volume in the downstream reservoir to the upstream reservoir.

[0029] In a preferred embodiment of the present invention, in step S3, the constraint system of the hybrid renewable energy system includes power balance constraints, pumping device constraints, system ecological constraints, interaction constraints, and cascade hydropower and unit constraints. Among them,

[0030] The power balance constraints include the total balance constraint of the power generation module;

[0031] The pumping device constraints include power - water volume conversion constraints;

[0032] The system ecological constraints include instantaneous discharge constraints, daily average discharge constraints, and the third - type flow constraints;

[0033] The interaction constraints include the interaction balance constraints of the power generation module;

[0034] The cascade hydropower and unit constraints include climbing constraints, vibration zone constraints, start - stop constraints, start - up and shut - down time constraints, water level constraints, head constraints, power generation constraints, water volume balance constraints, discharge volume constraints, water level - storage capacity constraints, and discharge volume - tail water level constraints.

[0035] In a preferred embodiment of the present invention, the optimization objective is to minimize the mean square error of the remaining load and maximize the renewable energy access capacity.

[0036] In a preferred embodiment of the present invention, in step S4, the multi - objective optimization algorithm is a particle swarm optimization algorithm. The steps for improving this particle swarm optimization algorithm are as follows:

[0037] S41: Initialize the particle swarm, set the initial parameters of the particle swarm, including the position and velocity of the particles, and define the particle swarm size, the maximum number of iterations, the objective function, the initial value and range of the inertia weight factor, and the learning factors;

[0038] S42: According to the current number of iterations and the maximum number of iterations, adjust the inertia weight factor using the linear decreasing formula strategy;

[0039] S43: According to the current number of iterations and the maximum number of iterations, adjust the individual learning factor and the group learning factor using the linear decreasing formula strategy;

[0040] S44: Calculate the fitness value. According to the current position of the particle, substitute it into the objective function of the optimization model to calculate the fitness value of each particle on multiple objectives;

[0041] S45: Adopt the Pareto front method. According to the obtained fitness values, construct the non-dominated solution set of the current particle, screen out the Pareto front optimal solutions, and store them in the external archive;

[0042] S46: Update the particle velocity and position;

[0043] S47: Check whether the particle meets the stop condition. If the maximum number of iterations is not reached or the accuracy requirement is not met, return to step S42 to continue adjusting the inertia weight factor, individual learning factor, and swarm learning factor, and perform iterative optimization; when the stop condition is met, output the Pareto front solution set and the corresponding scheduling strategy to complete the solution of the multi-objective optimization problem.

[0044] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0045] 1. The multi-objective regulation method based on cascaded water-wind-solar-pumped storage multi-energy complementarity of the present invention integrates the pumping device into the hybrid system of cascaded hydropower, photovoltaic power generation, and wind power generation, providing stronger flexibility and regulation ability for the power grid; different from traditional hydropower generation, the addition of the pumping device can further improve the stability of the system under the condition of large fluctuations in wind power generation and photovoltaic power generation, providing more regulation space for energy scheduling.

[0046] 2. The multi-objective regulation method based on cascaded water-wind-solar-pumped storage multi-energy complementarity of the present invention introduces ecological constraints and unit information in multi-objective optimization. Considering that the hybrid renewable energy system should not only optimize stability but also take into account the ecological environment and the operating characteristics of power units, while minimizing the negative impact on the environment to the greatest extent, combining unit information can more accurately reflect the technical characteristics and operating constraints of various generating units, avoiding unrealistic scheduling schemes.

[0047] 3. The present invention accurately predicts short-term wind power generation and photovoltaic power generation through a short-term output prediction model, and substitutes the prediction data into the optimization model to provide a reliable data source for the optimization model.

[0048] 4. The short-term output prediction model in the present invention provides accurate future data, and the optimization model makes scientific decisions based on these data, avoiding errors caused by subjective judgment. Through the prediction data, the optimization model can formulate power generation plans, energy storage scheduling, and load distribution strategies in advance to achieve efficient utilization of resources; the prediction data can be updated in real time, and the optimization model can dynamically adjust the strategy according to the latest data to adapt to changes in weather, load, and other conditions. Description of the Drawings

[0049] Figure 1 It is a principle flow chart of the multi - objective regulation method based on cascade water - wind - light - pumped storage multi - energy complementarity of the present invention.

[0050] Figure 2 It is a principle flow chart of the short - term output prediction model.

[0051] Figure 3 It is a prediction result graph of the short - term output prediction model.

[0052] Figure 4 It is a structure diagram of the hybrid renewable energy system constructed by the present invention.

[0053] Figure 5 It is a fitting result graph of the reservoir capacity curve and the discharge - tail water level curve.

[0054] Figure 6 It is a result graph of the Pareto - front solution set solved by the improved particle swarm optimization algorithm.

[0055] Figure 7 It is a result graph of the working strategy of the pumping device and the water level change of the three - stage hydropower station.

[0056] Figure 8 It is a result graph of the output summary of the hybrid renewable energy system.

[0057] Figure 9 It is a start - stop strategy graph of each stage of units in the three - stage hydropower station.

[0058] Figure 10 It is a result graph of the output of each stage of units in the three - stage hydropower station. Detailed Embodiments

[0059] The present invention will be further described in detail below in conjunction with embodiments and the drawings, but the embodiments of the present invention are not limited thereto.

[0060] See Figures 1 - 10 , the multi - objective regulation method based on cascade water - wind - light - pumped storage multi - energy complementarity of the present invention includes the following steps:

[0061] Step S1: Collect comprehensive data on environmental factors affecting the wind power generation group and the photovoltaic power generation group, screen and analyze the collected environmental factors, and obtain the key influencing factors that have the greatest impact on wind power generation efficiency and photovoltaic power generation efficiency; specifically:

[0062] S11: Collect environmental data in the areas where the wind power stations and photovoltaic power stations are located, with a time span of nearly 3 months; the types of the collected environmental data include: humidity, solar radiation intensity, wind speed, air pressure, and precipitation; the data is collected hourly; among them,

[0063] Humidity: The amount of water vapor in the air, expressed as a percentage;

[0064] Solar radiation intensity: The solar radiation energy received per unit area, expressed in watts per square meter (W / m 2 )

[0065] Wind speed: The wind speed per unit time, measured in meters per second (m / s);

[0066] Air pressure: The air pressure, with the unit of hectopascal (hPa);

[0067] Precipitation: The amount of precipitation per unit time, measured in millimeters (mm).

[0068] S12: Process the collected environmental data, including removing missing values, handling outliers, standardizing timestamps, and unifying data; specifically:

[0069] S121: By checking for null or NaN values in each column of data and using the Pandas library in the Python programming language for operations, directly delete rows or columns with a large number of missing values; fill in missing values using interpolation, and for columns with a small number of missing values, fill them with the mean or median;

[0070] S122: Based on the standard deviation, consider data greater than or less than 2 times the standard deviation of the mean in the data as outliers and correct the outliers to the upper and lower limit values;

[0071] S123: Ensure that all timestamp formats are consistent and convert strings to the standard time format;

[0072] S123: Ensure that the units of all meteorological data are unified. If the dimensional differences of different variables are large, standardize the data.

[0073] S13: Based on the correlation analysis method, by setting the threshold of the correlation coefficient, screen out the environmental factors most relevant to wind power generation efficiency or photovoltaic power generation efficiency; by identifying the linear correlation between environmental factors, obtain the key influencing factors that have the greatest impact on wind power generation efficiency or photovoltaic power generation efficiency; specifically:

[0074] S131: Measure the linear relationship between each environmental factor and wind power generation efficiency or photovoltaic power generation efficiency by calculating the Pearson correlation coefficient, generate a complete correlation matrix, and view the correlation between all environmental factors and wind power generation efficiency or photovoltaic power generation efficiency;

[0075]

[0076] Where: cov(X,Y) is the covariance, σX and σ Y are the standard deviations of variable X and variable Y, respectively;

[0077] S132: Set a preset threshold range to screen out environmental factors with strong correlations with wind power generation efficiency or photovoltaic power generation efficiency. When the Pearson correlation coefficient is within the preset threshold range, it is considered that there is a significant linear relationship between the environmental factor and wind power generation efficiency or photovoltaic power generation efficiency;

[0078] In this embodiment, the set threshold is 0.6. When the Pearson correlation coefficient is greater than 0.6 or less than -0.6, it is considered that there is a significant linear relationship between the environmental factor and wind power generation efficiency or photovoltaic power generation efficiency;

[0079] S133: Identify the multicollinearity among environmental factors and use the variance inflation factor VIF to check for multicollinearity. When the VIF of a certain environmental factor is greater than 10, it indicates that there is a strong linear correlation between this environmental factor and other environmental factors, so this environmental factor needs to be deleted.

[0080] Step S2: Build a short-term output prediction model for the wind power generation group and the photovoltaic power generation group (see the principle flowchart in Figure 2 , and see the prediction result graph in Figure 3 ). Optimize the hyperparameters of this short-term output prediction model through an optimization algorithm, and input the key influencing factors, historical output data of the wind power generation group and the photovoltaic power generation group into the optimized short-term output prediction model to obtain the prediction results of the short-term output of the wind power generation group and the photovoltaic power generation group;

[0081] In this embodiment, the construction steps of the short-term output prediction model are as follows:

[0082] S21: Initialize the particle swarm, determine the size of the particle swarm, that is, the number of particles. Among them, each particle represents a set of key hyperparameters of the LSTM model, including the learning rate, the number of training times, and the number of neurons; set the initial parameters of the particle swarm, including the position and velocity of the particles, and at the same time define the inertia weight factor, the learning factor, and the maximum number of iterations;

[0083] S22: Update the particles. According to the current position and velocity of the particles, and in combination with the velocity update formula and position update formula of the particle swarm optimization algorithm, gradually iterate and update the particles;

[0084] S23: Use the optimal solution obtained by the particle swarm optimization algorithm to adjust the key hyperparameters of the LSTM model, thereby obtaining the short-term output prediction model;

[0085] S24: Input the key influencing factors and the historical output data of the wind power generation group and the photovoltaic power generation group into the short-term output prediction model to obtain the prediction results of the short-term output of the wind power generation group and the photovoltaic power generation group;

[0086] S25: Judge the prediction results. If the prediction results meet the accuracy requirements, use the prediction results as the final prediction results and end the prediction process; if the prediction results do not meet the accuracy requirements, return to step S22 to continue iterative optimization until the final prediction results meet the accuracy requirements.

[0087] Step S3: Combine the prediction results of the short-term output of the wind power generation group and the photovoltaic power generation group, establish a constraint system for the hybrid renewable energy system, and construct an optimization model for the hybrid renewable energy system with the goal of improving the stability of the hybrid renewable energy system;

[0088] In this embodiment, the hybrid renewable energy system includes a wind power generation group, a photovoltaic power generation group, a three-stage cascade hydropower station, a pumping device between each hydropower station, and a central control system; the interaction mode of the hybrid renewable energy system is reflected in that: the pumping device between each hydropower station is powered by the wind power generation group and the photovoltaic power generation group; the wind power generation group and the photovoltaic power generation group can be optimally selected to supply the pumping station or be connected to the power grid; the power generation of the three-stage cascade hydropower station is directly connected to the power grid; the pumping device between each hydropower station pumps the remaining water volume in the downstream reservoir to the upstream reservoir.

[0089] The constraint system of the hybrid renewable energy system includes power balance constraints, pumping device constraints, system ecological constraints, interaction constraints, and cascade hydropower and unit constraints. Among them, the power balance constraints include: total power generation module balance constraints; the pumping device constraints include: power-water volume conversion constraints; the system ecological constraints include: instantaneous discharge constraints, daily average discharge constraints, and the third type of flow constraints; the interaction constraints include: power generation module interaction balance constraints; the cascade hydropower and unit constraints include: ramp constraints, vibration zone constraints, start-stop constraints, start-stop time constraints, water level constraints, head constraints, hydropower generation constraints, water volume balance constraints, discharge volume constraints, water level-storage capacity constraints, discharge volume-tail water level constraints, storage capacity constraints; specifically,

[0090] The storage capacity constraint is:

[0091]

[0092] In the formula: V m and are the minimum and maximum storage capacities of the mth reservoir respectively, and V t,m is the storage capacity of the mth reservoir at time t.

[0093]

[0094] Where: δ represents the allowable fluctuation range of the reservoir capacity at the end of the scheduling period; is the target capacity of the m-th reservoir at the end of the scheduling period; V T,m is the capacity of the m-th reservoir at the end of the scheduling period.

[0095] The water discharge constraint is:

[0096]

[0097] Where: and q m,i are the maximum and minimum water discharge rates of the i-th unit of the m-th hydropower station respectively, Q t,m is the total water discharge of the m-th hydropower station at time t, q t,m,i is the water discharge rate of the i-th unit of the m-th hydropower station at time t.

[0098]

[0099] Where: I is the total number of hydropower units, is the power generation flow rate of the i-th unit of the m-th hydropower station at time t, and are the maximum and minimum power generation flow rates of the i-th unit of the m-th hydropower station respectively, is the total power generation flow rate of the m-th hydropower station at time t.

[0100] The hydropower generation constraint is:

[0101]

[0102] Where: and p m,i are the upper and lower limits of the power generation of the i-th unit of the m-th hydropower station respectively, p t,m,i is the power generation of the i-th unit of the m-th hydropower station at time t.

[0103] The water level constraint is:

[0104]

[0105] Where: and Z m are the upper and lower limits of the water level of the m-th reservoir respectively, Z t,m is the water level of the m-th reservoir at time t.

[0106] The water volume balance constraint is:

[0107]

[0108]

[0109] Wherein: is the amount of water discharged from the m-th hydropower station in the t-th period, I t,m is the inflow of the m-th reservoir in the t-th period, τ m represents the delay time for water to flow from the m-th reservoir to the (m + 1)-th reservoir, represents the delay time for water to be pumped from the (m + 1)-th reservoir to the m-th reservoir, is the floor function, R t,m is the inflow from the interval of the m-th reservoir in the t-th period, represents the amount of water pumped from reservoir (m + 1) to reservoir m in the t-th period (sorting the hydropower stations from upstream to downstream).

[0110] The head constraint is:

[0111]

[0112] Wherein: H t,m represents the net head of hydropower station m in the t-th period, is the tail water level without head loss of hydropower station m in the t-th period.

[0113] The water level - storage capacity constraint is:

[0114] Z t,m = f zv,m (V t,m );

[0115] The discharge - tail water level constraint is:

[0116]

[0117] The power generation constraint is:

[0118]

[0119] Wherein: f zv,m (·) is a function of the water level in front of the reservoir and the reservoir storage capacity, f zq,m (·) is a function of the discharge of the reservoir and the tail water level, represents the power generation efficiency coefficient of the i-th unit of hydropower station m, f zv,m (·) is a third-order polynomial non-linear function, which is fitted by the least squares method according to many relevant data points corresponding to the water level and the reservoir capacity, f zq,m (·) Similarly, for the fitting result graph, see Figure 5 .

[0120] The power - water conversion constraint is:

[0121]

[0122] where χ is the unit conversion coefficient, represents the efficiency coefficient of pumping water from the (m + 1)-th reservoir to the m-th reservoir, is the power consumption of pumping water from the (m + 1)-th reservoir to the m-th reservoir during the t-th period, ρ represents the water density, and h m represents the vertical height difference from the (m + 1)-th reservoir to the m-th reservoir.

[0123] The interaction balance constraint of the power generation module is:

[0124]

[0125] where: and are the predicted power generations of the wind power station and the photovoltaic power station during the t-th period, which are predicted by the PSO-LSTM prediction model in step 3.

[0126] The ramp constraint is:

[0127] |p t,m,i -p t-1,m,i |≤Δp m,i ;

[0128] where p t,m,i is the power generation of the i-th unit of the m-th hydropower station during the t-th period, and Δp m,i is the allowable ramp range of the i-th unit of the m-th hydropower station.

[0129] The vibration zone constraint is:

[0130]

[0131] where: and are the upper and lower limits of the j-th vibration zone of the i-th unit of the m-th hydropower station.

[0132] The start-stop constraint and the start-stop time constraint are:

[0133]

[0134] where: is the start-up state variable of the i-th unit of the m-th hydropower station, is the shutdown state variable of the i-th unit of the m-th hydropower station, is the minimum start-up time of the i-th unit of the m-th hydropower station, is the minimum shutdown time of the i-th unit of the m-th hydropower station.

[0135]

[0136] where: is the conventional power demand load during period t.

[0137] The instantaneous discharge constraint is:

[0138]

[0139] The daily average discharge constraint is:

[0140]

[0141] The third type of flow constraint is:

[0142]

[0143] In the formula: is the ecological flow of the m-th hydropower station, is the critical value of the water level above the dam of the hydropower station. These are the three types of ecological flow constraints, and one of them is selected for constraint according to the different requirements of each hydropower station.

[0144] This embodiment is based on actual cascade hydropower stations in a certain area of southern China, considering the installation of pumping devices between three cascade hydropower stations. The key parameters of the cascade hydropower stations are shown in Table 1.

[0145] Table 1 Parameters of cascade hydropower stations

[0146]

[0147]

[0148] The optimization objective (i.e., the objective function) is to minimize the mean square error of the remaining load and maximize the renewable energy access capacity; where

[0149] Minimize the mean square error of the remaining load:

[0150]

[0151] Maximize the renewable energy access capacity:

[0152]

[0153] In the formula: T is the total scheduling period; t is the time index; is the remaining load during period t; M is the total number of hydropower stations; m is the hydropower station index; is the power generation of the m-th hydropower station during period t; is the grid-connected power of wind power stations and photovoltaic power stations during period t.

[0154] Step S4: Solve the multi-objective optimization problem of the optimization model through an improved multi-objective optimization algorithm to obtain the scheduling strategy under different scenarios and the operation mechanism of the hybrid renewable energy system and the coordinated operation method of various types of energy in the hybrid renewable energy system.

[0155] Among them, the multi-objective optimization algorithm is the particle swarm optimization algorithm, and the steps for improving this particle swarm optimization algorithm are as follows:

[0156] S41: Initialize the particle swarm, set the initial parameters of the particle swarm, including the position and velocity of the particles, and define the particle swarm size, the maximum number of iterations, the objective function, the initial value and range of the inertia weight factor, and the learning factor;

[0157] S42: Dynamically adjust the inertia weight factor. According to the current iteration number and the maximum iteration number, use the linear decreasing formula strategy to adjust the inertia weight factor. The inertia weight factor maintains a relatively high value in the initial stage to enhance the global search ability and gradually decreases in the later stage to improve the local search accuracy. Specifically:

[0158]

[0159] In the formula: IT is the current iteration number; MI is the total number of iterations; w s and w e are the initial value and the termination value of the inertia weight factor. Among them, in the initial stage of iteration, a larger inertia weight factor will prevent the algorithm from falling into local minima easily and is conducive to global search; in the later stage of iteration, a smaller inertia weight factor is beneficial to local search and is conducive to the convergence of the algorithm;

[0160] S43: According to the current iteration number and the maximum iteration number, use the linear decreasing formula strategy to adjust the individual learning factor and the population learning factor to balance the influence of the individual optimal solution and the population optimal solution. Specifically:

[0161]

[0162] In the formula: c1 is the individual learning factor; c2 is the population learning factor; c 1s and c 1e are the initial value and the stop value of c1, and c 1s is greater than c 1e ; c 2s and c 2e are the initial value and the stop value of c2, and c 2s is less than c 2e;At the initial stage of iteration, larger c1 and smaller c2 endow the particles with better self-learning ability and poorer social learning ability, which is conducive to global search; at the later stage of iteration, smaller c1 and larger c2 endow the particles with stronger social learning ability and poorer self-learning ability, which is conducive to the convergence of the algorithm.

[0163] S44: Calculate the fitness value. According to the current position of the particle, substitute it into the objective function of the optimization model to calculate the fitness value of each particle on multiple objectives;

[0164] S45: Adopt the Pareto front method. According to the obtained fitness values, construct the non-dominated solution set of the current particle, screen out the Pareto front optimal solutions, and store them in the external archive;

[0165] S46: Update the particle velocity and position;

[0166] Assume that the total number of particles is M. The position and velocity of the nth particle in dimension d are expressed as follows:

[0167]

[0168] Each particle adjusts its velocity and position by tracking its previous personal best position and the group best position. These two best positions are expressed as:

[0169]

[0170] In the formula: P n ' is the personal best position of the nth particle; P g ' is the group best position obtained from all particles in the previous iteration;

[0171] The velocity and position update formulas of the particle swarm optimization algorithm are expressed as:

[0172]

[0173] In the formula: w is the inertia weight factor; among them, by increasing the inertia weight factor, it is not easy to fall into local minima, which is convenient for global search; by reducing the inertia weight factor, it is conducive to local search, so it is conducive to the convergence of the algorithm; c1 is the individual learning factor, c2 is the group learning factor, and the two respectively reflect the self-learning ability and social learning ability of the particles; r1' and r2' are random numbers uniformly distributed in [0,1];

[0174] S47: Check whether the particle meets the stop condition. If the maximum number of iterations is not reached or the accuracy requirement is not met, return to step S42, continue to adjust the inertia weight factor, individual learning factor and group learning factor, and perform iterative optimization; when the stop condition is met, output the Pareto front solution set and the corresponding scheduling strategy to complete the solution of the multi-objective optimization problem.

[0175] Finally, the obtained Pareto front solution set results are as Figure 6 shown. The objective function values of the two are normalized and added, and the particle with the smallest objective value after addition is taken, that is, the compromise solution is searched for analysis;

[0176] The working strategy of the pumping device and the water level change results of the three-stage hydropower station are as Figure 7 shown. It can be seen from this that the pumping device can flexibly adjust the working mode according to the demand to achieve efficient utilization of water energy; through reasonable scheduling, the water level changes smoothly, avoiding the impact of too high or too low water level on the operation of the hydropower station; the pumping device pumps and stores energy when the power is surplus and discharges water to generate electricity when the power is insufficient, improving the energy storage capacity and stability of the system.

[0177] The output summary results of the hybrid renewable energy system are as Figure 8 shown; it can be seen from this that the output characteristics of wind power, photovoltaic power and hydropower are complementary, smoothing the volatility of renewable energy; the total output of the hybrid system is relatively stable, reducing the impact of the fluctuation of a single energy source on the power system; through optimized scheduling, the maximum utilization of renewable energy is achieved, reducing the phenomena of wind curtailment and PV curtailment.

[0178] The start-stop strategies of each stage of the three-stage hydropower station are as Figure 9 shown. It can be seen from this that according to the load demand and the output of renewable energy, the start-stop states of each stage of the unit are dynamically adjusted to achieve refined scheduling; through optimizing the start-stop strategy, the idling and inefficient operation of the unit are reduced, and the energy loss is lowered.

[0179] The output results of each stage of the three-stage hydropower station are as Figure 10 shown. It can be seen from this that the output of each stage of the unit can be flexibly adjusted, highly matching the load demand of the power system; through optimizing the output distribution of each stage of the unit, the efficient utilization of water energy resources is achieved.

[0180] The above are the preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the above content. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent replacement methods and are all included in the protection scope of the present invention.

Claims

1. A multi-objective regulation method based on cascaded water-wind-solar-pumped storage multi-energy complementarity, characterized in that, It includes the following steps: Step S1: Collect comprehensive data on environmental factors affecting the wind power generation group and the photovoltaic power generation group, screen and analyze the collected environmental factors, and obtain the key influencing factors that have the greatest impact on the wind power generation efficiency and the photovoltaic power generation efficiency; Step S2: Build a short-term output prediction model for the wind power generation group and the photovoltaic power generation group, optimize the hyperparameters of the short-term output prediction model through an optimization algorithm, and input the key influencing factors, the historical output data of the wind power generation group and the photovoltaic power generation group into the optimized short-term output prediction model to obtain the prediction results of the short-term output of the wind power generation group and the photovoltaic power generation group; Step S3: Combine the prediction results of the short-term output of the wind power generation group and the photovoltaic power generation group, establish a constraint system for the hybrid renewable energy system, and construct an optimization model for the hybrid renewable energy system with the goal of improving the stability of the hybrid renewable energy system; Step S4: Solve the optimization model through an improved multi-objective optimization algorithm to obtain the scheduling strategy under different scenarios and the hybrid renewable energy system operation mechanism and the coordinated operation mode of various types of energy in the hybrid renewable energy system.

2. The multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity according to claim 1, characterized in that, The specific steps of the said Step S1 are: S11: Collect the environmental data of the areas where the wind power stations and the photovoltaic power stations are located, with a time span of nearly 3 months; the types of the collected environmental data include: humidity, solar radiation intensity, wind speed, air pressure and precipitation; the data is collected hourly; S12: Process the collected environmental data, including removing missing values, processing outliers, standardizing timestamps and unifying data; S13: Based on the correlation analysis method, by setting the threshold of the correlation coefficient, screen out the environmental factors that are most relevant to the wind power generation efficiency or the photovoltaic power generation efficiency; by identifying the linear correlation between environmental factors, obtain the key influencing factors that have the greatest impact on the wind power generation efficiency or the photovoltaic power generation efficiency.

3. The multi-objective regulation method based on cascaded water-wind-light-pumped storage multi-energy complementarity according to claim 2, wherein The specific steps of the said Step S13 are: S131: Calculate the Pearson correlation coefficient to measure the linear relationship between each environmental factor and the wind power generation efficiency or the photovoltaic power generation efficiency, generate a complete correlation matrix, and view the correlation between all environmental factors and the wind power generation efficiency or the photovoltaic power generation efficiency; S132: Set a preset threshold range to screen out the environmental factors that have a strong correlation with the wind power generation efficiency or the photovoltaic power generation efficiency; when the Pearson correlation coefficient is within the preset threshold range, it is considered that the environmental factor has a significant linear relationship with the wind power generation efficiency or the photovoltaic power generation efficiency; S133: Identify the multicollinearity between environmental factors, and use the variance inflation factor VIF to check whether there is multicollinearity; when the VIF of an environmental factor is greater than the set value, it means that there is a strong linear correlation between this environmental factor and other environmental factors, so this environmental factor needs to be deleted.

4. The multi-objective regulation method based on cascaded water-wind-light-pumped storage multi-energy complementarity according to claim 1, wherein In Step S2, the construction steps of the said short-term output prediction model are: S21: Build an LSTM model, optimize the key hyperparameters in the LSTM model through the particle swarm optimization algorithm to obtain a short-term output prediction model; S22: Input the key influencing factors and the historical output data of the wind power group and the photovoltaic power generation group into the short-term output prediction model to obtain the prediction results of the short-term output of the wind power group and the photovoltaic power generation group; S23: Judge the prediction results. If the prediction results meet the accuracy requirements, use the prediction results as the final prediction results and end the prediction process; if the prediction results do not meet the accuracy requirements, update the particles in the particle swarm optimization algorithm, and optimize the key hyperparameters in the LSTM model through the particle swarm optimization algorithm until the final prediction results meet the accuracy requirements.

5. The multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity according to claim 4, wherein, In step S21, the key hyperparameters of the LSTM model include the learning rate, the number of training times, and the number of neurons.

6. The multi-objective regulation method based on cascaded water-wind-solar-pumped storage multi-energy complementarity according to claim 5, characterized in that, In step S21, the steps for the particle swarm optimization algorithm to optimize the key hyperparameters of the LSTM model are as follows: S211: Initialize the particle swarm, set the initial parameters of the particle swarm, including the position and velocity of the particles, and at the same time define the inertia weight factor, the learning factor, and the maximum number of iterations; S212: Update the particles. According to the current position and velocity of the particles, combined with the velocity update formula and the position update formula of the particle swarm optimization algorithm, gradually iterate and update the particles; S213: Use the optimal solution obtained by the particle swarm optimization algorithm to adjust the key hyperparameters of the LSTM model.

7. The multi-objective regulation method based on cascaded water-wind-light-pumped storage multi-energy complementarity according to claim 1, wherein In step S3, the hybrid renewable energy system includes a wind power group, a photovoltaic power generation group, a three-stage cascade hydropower station, a pumping device between each hydropower station, and a central control system; the interaction mode of the hybrid renewable energy system is reflected as follows: the pumping device between each hydropower station is powered by the wind power group and the photovoltaic power generation group; the wind power group and the photovoltaic power generation group can be optimally selected to supply the pumping device between each hydropower station or be connected to the power grid; the power generation of the three-stage cascade hydropower station is directly connected to the power grid; the pumping device between each hydropower station pumps the remaining water volume in the downstream reservoir to the upstream reservoir.

8. The multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity according to claim 7, wherein In step S3, the constraint system of the hybrid renewable energy system includes power balance constraints, pumping device constraints, system ecological constraints, interaction constraints, and cascade hydropower and unit constraints, where the power balance constraints include the total balance constraint of the power generation module; the pumping device constraints include the power-water volume conversion constraint; the system ecological constraints include the instantaneous discharge constraint, the daily average discharge constraint, and the third type of flow constraint; the interaction constraints include the interaction balance constraint of the power generation module; the cascade hydropower and unit constraints include the ramp constraint, the vibration zone constraint, the start-stop constraint, the start-stop time constraint, the water level constraint, the head constraint, the power generation constraint, the water volume balance constraint, the discharge volume constraint, the water level-storage capacity constraint, and the discharge volume-tail water level constraint.

9. The multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity according to claim 8, characterized in that, The optimization objective is to minimize the mean square error of the remaining load and maximize the renewable energy access capacity.

10. The multi-objective regulation method based on cascade water-wind-light-pumped storage multi-energy complementarity according to claim 1, characterized in that In step S4, the multi-objective optimization algorithm is the particle swarm optimization algorithm, and the steps for improving the particle swarm optimization algorithm are as follows: S41: Initialize the particle swarm, set the initial parameters of the particle swarm, including the position and velocity of the particles, define the particle swarm size, the maximum number of iterations, the objective function, the initial value and range of the inertia weight factor, and the learning factor; S42: Adjust the inertia weight factor according to the current iteration number and the maximum iteration number using the linear decreasing formula strategy; S43: Adjust the individual learning factor and the swarm learning factor according to the current iteration number and the maximum iteration number using the linear decreasing formula strategy; S44: Calculate the fitness value. Substitute the current position of the particle into the objective function of the optimization model to calculate the fitness value of each particle on multiple objectives; S45: Adopt the Pareto front method to construct the non-dominated solution set of the current particle according to the obtained fitness value, screen out the Pareto front optimal solution, and store it in the external archive; S46: Update the particle velocity and position; S47: Check whether the particle meets the stop condition. If the maximum iteration number is not reached or the accuracy requirement is not met, return to step S42 to continue adjusting the inertia weight factor, the individual learning factor and the swarm learning factor, and perform iterative optimization; when the stop condition is met, output the Pareto front solution set and the corresponding scheduling strategy to complete the solution of the multi-objective optimization problem.

Images