New energy station emergency control method and system under extreme weather

Abstract

The invention belongs to the technical field of power system optimization control, and relates to a new energy station emergency control method and system under extreme meteorology, and the method comprises the steps: determining a wind-light-storage combined output model and an EV active output model, building a target function based on the wind-light-storage combined output model and the EV active output model, and carrying out the optimization control of a new energy station. The objective function takes the charging active power of an EV set participating in emergency control, the power deviation of a wind power station set participating in emergency control, the power deviation of a photovoltaic station set participating in emergency control and the charging power of an energy storage station set as decision variables; and optimizing the target function by adopting an improved whale algorithm, and distributing and scheduling the active power of the new energy station according to the obtained active power distribution scheme. The method is simple and easy to operate, high in search capability and high in fitness, can realize reasonable scheduling of new energy resources, and improves the stability of the new energy station under the extreme meteorological condition.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system optimization control, and specifically relates to a new energy station emergency control method and system under extreme weather conditions. Background Art

[0002] With the increasing penetration of renewable energy into the power grid and the growing reliance on electricity across various sectors, extreme weather events pose an increasingly serious threat to power systems. Once extreme weather occurs, it can easily lead to significant fluctuations in renewable energy output, disrupting the grid's power balance and even impacting the entire power system's ability to supply power. However, current research on the resilience of renewable energy clusters under extreme conditions is relatively limited. This research primarily focuses on risk reliability assessment, supporting grid operations, the impact of renewable energy characteristics on the safe and stable operation of power systems, and damage to power systems during natural disasters.

[0003] Currently, the main approaches to addressing the emergency control of renewable energy stations during extreme weather conditions include classical algorithms and intelligent algorithms. Classical algorithms, such as graph theory and heuristic algorithms, are computationally intensive, inefficient, and unsuitable for large-scale systems. Intelligent algorithms, on the other hand, are also not suitable due to their numerous parameters, difficulty in operation, and tendency to fall into local optimal solutions. Furthermore, existing scheduling schemes for extreme weather conditions often fail to fully utilize all resources at renewable energy stations, limiting the emergency control strategy's ability to withstand extreme weather conditions and potentially leading to other adverse consequences. Summary of the Invention

[0004] The purpose of the present invention is to provide an emergency control method and system for new energy stations under extreme weather conditions. The method is simple and easy to operate, has strong search capabilities, and high adaptability. It can achieve reasonable scheduling of new energy resources and improve the stability of new energy stations under extreme weather conditions.

[0005] To achieve the above objectives, the present invention adopts the following technical solutions: According to one aspect of the present invention, a method and system for emergency control of a new energy station under extreme weather conditions is provided, comprising the following steps: Determine the wind, solar, and energy storage combined output model and the EV (Electric Vehicle) active power output model; An objective function is constructed based on the wind-solar-storage combined output model and the EV active output model. The objective function uses the charging active power of the set of EVs participating in emergency control, the power deviation of the set of wind farms participating in emergency control, the power deviation of the set of photovoltaic farms participating in emergency control, and the charging power of the set of energy storage farms as decision variables to output an active power allocation plan. The improved whale algorithm is used to optimize the objective function, and the active power of the new energy stations is allocated and dispatched according to the obtained active power allocation scheme.

[0006] The steps of determining the wind, solar and energy storage combined output model and the EV active power output model include the following steps: Collect data, including meteorological data, remaining power of energy storage stations, and EV active output power; and construct a wind, solar, and energy storage combined output model and an EV active output model based on the collected data.

[0007] In the process of constructing the objective function, the goal is to minimize the power change, and the constraints are the EV charging power range, the EV SOC (State of Charge) safety range, and the EV charging power per unit time period.

[0008] In the step of optimizing the objective function using the improved whale algorithm, the goal is to minimize the power change.

[0009] The above technical solution uses an improved whale algorithm to optimize the objective function and obtain an active power allocation plan. The algorithm is simple and easy to find the optimal solution. It also implements a global search to avoid local optimal solutions, improves the rationality of the scheduling of new energy stations under extreme weather conditions, and ensures the stability of new energy stations. In addition, the above technical solution incorporates EVs into the power allocation and scheduling plan under extreme weather conditions, adding representative distributed resources. EVs are used to absorb the additional power generated by the power fluctuations of new energy stations under extreme weather conditions, making the scheduling strategy more diverse and enabling more rational utilization of the resources of new energy stations.

[0010] According to one embodiment of the present invention, the meteorological data includes air density, actual wind speed, light intensity, etc.

[0011] According to one embodiment of the present invention, a Monte Carlo sampling method is used to predict the time when EVs arrive at and leave a wind farm or a photovoltaic farm, and to determine a set of EVs that participate in emergency control.

[0012] According to one embodiment of the present invention, a pre-trained LSTM (Long Short-Term Memory) neural network model is used to predict the power deviation of wind farms and photovoltaic stations based on real-time meteorological data and predicted meteorological data. The set of wind farms and photovoltaic stations participating in emergency control is determined based on whether the predicted power deviation exceeds a preset value.

[0013] Since wind and solar power have strong temporal and nonlinear characteristics, the use of LSTM neural network models can improve the stability and reliability of power deviation predictions for wind farms and photovoltaic stations, especially the robustness of predictions under extreme weather conditions; thereby improving the rationality of the resulting emergency control scheme, achieving reasonable scheduling of new energy resources, and improving the stability of new energy stations under extreme weather conditions.

[0014] According to one embodiment of the present invention, the power deviation of the wind farm set and the power deviation of the photovoltaic farm set are respectively related to the priority power consumption coefficients of the corresponding wind farms and photovoltaic farms, and the priority power consumption coefficients are time-dependent piecewise functions; The priority electricity consumption coefficient is the product of the level importance coefficient of the wind farm and the photovoltaic farm and the electricity consumption coefficient under the time scale.

[0015] According to one embodiment of the present invention, the steps of optimizing the objective function using the improved whale algorithm include: Determine the population size of the whale algorithm, randomly generate whale positions, use the optimized bubble attack method to update the whale positions and search for the optimal solution.

[0016] According to one embodiment of the present invention, in the optimized bubble attack method, the convergence factor a Calculated according to the following formula: ; Where: t is the current iteration number; M is the maximum number of iterations.

[0017] The convergence factor a Introducing an exponential function into the formula makes step size changes smoother, reduces algorithm oscillation during optimization, and improves stability. This improvement enables the Whale Algorithm to achieve a better balance between global search capability and local convergence speed, ensuring the optimal solution. a Slowly decreasing the factor in the early stages of an iteration (to enhance global exploration) and rapidly decreasing it in the later stages (to enhance local exploitation) better aligns with the "exploration before exploitation" principle of intelligent optimization algorithms. By optimizing the convergence factor, we balance exploration and exploitation during the whale search process, avoiding the rigidity of linear decay and potentially improving the algorithm's convergence speed and solution accuracy.

[0018] According to one aspect of the present invention, there is provided a new energy station emergency control device under extreme weather conditions, comprising: Model building module, used to determine the wind, solar and energy storage combined output model and the EV active power output model; an objective function construction module for constructing an objective function based on the wind-solar-storage combined output model and the EV active output model, wherein the objective function uses the charging active power of the set of EVs participating in emergency control, the power deviation of the set of wind farms participating in emergency control, the power deviation of the set of photovoltaic farms participating in emergency control, and the charging power of the set of energy storage farms as decision variables to output an active power allocation plan; The scheme generation module is used to optimize the objective function using the improved whale algorithm, and to allocate and dispatch the active power of the new energy stations according to the obtained active power allocation scheme.

[0019] According to another aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for emergency control of a new energy station under any of the above-mentioned extreme weather conditions is implemented.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, it implements any of the above-mentioned new energy station emergency control methods under extreme weather conditions.

[0021] Compared with the prior art, the present invention has at least the following beneficial effects: 1. Utilize the improved Whale Algorithm to obtain active power allocation solutions. This simple algorithm easily finds the optimal solution and implements a global search to avoid local optimal solutions. This improves the rationality of new energy station scheduling under extreme weather conditions and ensures the stability of new energy stations. By incorporating EVs into power allocation and scheduling under extreme weather conditions, representative distributed resources are added, making the scheduling strategy more diverse and enabling more rational utilization of new energy station resources.

[0022] 2. Use the improved whale algorithm for optimization and convergence factor a An exponential function is introduced into the formula to reduce the oscillation of the algorithm during the optimization process, improve stability, and enhance the convergence speed and solution accuracy of the algorithm. This improvement enables the whale algorithm to achieve a better balance between global search capability and local convergence speed, ensuring the optimal solution.

[0023] 3. The Monte Carlo sampling method can be used to predict the arrival and departure time of EVs at wind farms or photovoltaic stations without relying on strict analytical models. As historical data accumulates, the sampling distribution can be dynamically adjusted to improve the real-time nature of the prediction. This method is particularly suitable for regional wind and photovoltaic power plant coordinated scheduling, helps to improve the rationality of emergency control methods under extreme weather conditions, and realizes efficient and rational distribution and scheduling of active power at new energy stations.

[0024] 4. The LSTM neural network model is used to predict the power deviation of wind farms and photovoltaic stations, and then determine the set of wind farms and photovoltaic stations participating in emergency control. It can effectively capture the historical changes in meteorological factors such as wind speed, light, and temperature, as well as the dynamic impact of these factors on power deviation, eliminate the impact of time lag on emergency control plans under extreme meteorological conditions, improve the stability and reliability of prediction results, and thus improve the rationality of the obtained emergency control plan, which can realize the rational scheduling of new energy resources and improve the stability of new energy stations under extreme meteorological conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The accompanying drawings, which constitute part of the present invention, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings: Figure 1 This is a flow chart of the emergency control method for a new energy station under extreme weather conditions according to Example 1 of the present invention; Figure 2 This is a flowchart of predicting the output fluctuation of a new energy station in the next 1 to 6 hours using an LSTM neural network in Example 1 of the present invention. DETAILED DESCRIPTION

[0026] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention can be combined with each other.

[0027] The following detailed description is an exemplary description and is intended to provide further detailed description of the present invention. Unless otherwise indicated, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention.

[0028] Example 1 An emergency control method for new energy stations under extreme weather conditions, such as Figure 1 and Figure 2 As shown, the following steps are included: Collecting data, including meteorological data, remaining power of energy storage stations, and EV active output power; and constructing a wind, solar, and energy storage combined output model and an EV active output model based on the collected data; An objective function is constructed based on the wind-solar-storage combined output model and the EV active output model. The objective function uses the charging active power of the set of EVs participating in emergency control, the power deviation of the set of wind farms participating in emergency control, the power deviation of the set of photovoltaic farms participating in emergency control, and the charging power of the set of energy storage farms as decision variables to output an active power allocation plan. With the goal of minimizing power variation, the improved whale algorithm is used to optimize the objective function, and the active power of new energy stations (including wind farms and photovoltaic stations within a certain range) is allocated and dispatched according to the obtained active power allocation scheme.

[0029] When constructing the objective function, the goal is to minimize the power change, and the EV charging power range, EV SOC safety range, and EV charging power per unit time period are used as constraints.

[0030] The above technical solution uses an improved whale algorithm to optimize the objective function and obtain an active power allocation plan. The algorithm is simple and easy to find the optimal solution. It also implements a global search to avoid local optimal solutions, improving the rationality of new energy station scheduling under extreme weather conditions and ensuring the stability of new energy stations. In addition, the above technical solution adds representative distributed resources to the power allocation and scheduling scheme for EVs under extreme weather conditions, making the scheduling strategy more diverse and enabling more rational utilization of new energy station resources.

[0031] Specifically, the emergency control methods for new energy stations under extreme weather conditions are as follows: Collect meteorological data, the remaining power of energy storage stations, and the active output power of EVs; build a wind, solar, and storage combined output model and an EV active output model based on the collected data.

[0032] Among them, meteorological data include the air density, wind speed, wind direction, etc. of the wind farm station during each sampling period, which are used to combine the relevant parameters of the power generation units in the wind farm station (such as the swept area of ​​the wind turbine, wind energy utilization coefficient, tip speed ratio, pitch angle, wind turbine blade radius and blade rotation angular velocity, etc.) to obtain the output power of the wind farm station during the period; meteorological data also include the light intensity, temperature, cloud cover, etc. of the photovoltaic station during each sampling period, which are used to combine the relevant parameters of the power generation units in the photovoltaic station (such as the efficiency of converting rated light energy into electrical energy, the area of ​​photovoltaic modules receiving light, etc.) to obtain the output power of the photovoltaic station during the period.

[0033] The construction of the objective function includes the following steps: S1. Determine the wind, solar and storage combined output model The output model of the wind farm is: ; Where:P wt ( t ) is the t The output power of the wind farm in each period; is the air density (kg / m 3 ); S w is the swept area of ​​the wind turbine (m 2 ); v is the actual wind speed (m / s); C p is the wind energy utilization coefficient; λ is the tip speed ratio, λi is an intermediate variable; is the pitch angle (°); R t is the radius of the wind turbine blade (m); is the blade rotation angular velocity (rad / s).

[0034] The output model of the photovoltaic station is: ; Where: For the t The output power of photovoltaic power generation in each period; η The efficiency of converting rated light energy into electrical energy; A is the area of ​​the photovoltaic module receiving light; S is the intensity of sunlight on the photovoltaic module.

[0035] The charging model of the energy storage station is: ; The discharge model of the energy storage station is: ; Where: t The selected time period can be adjusted according to the needs. Generally, each time period is 1 hour, and an average of 24 time periods are selected per day. E ( t ) is the t The remaining power of the energy storage device in each period, E ( t +1) is the t +1 moment remaining power of the energy storage device; For the t Charge and discharge power in each period; h Is the charge and discharge time. E ( t )and E ( t +1) can be directly read and obtained using E ( t )andE ( t +1) can be calculated .

[0036] Considering the uncertainty of wind power generation and photovoltaic power generation, the energy storage system is added for dynamic balance. The wind, photovoltaic and storage combined output model is shown as follows: ; In the above formula, for t The combined output of wind, solar and storage systems at all times; for t The output power of wind power at any moment; for t The output power of photovoltaic power generation at all times; for t Charging power at all times.

[0037] In the process of determining the set of wind farms and photovoltaic stations participating in emergency control, based on real-time meteorological data (such as wind speed, wind direction, and air density for wind farms, and light intensity, temperature, cloud cover, etc. for photovoltaic stations), a pre-trained LSTM neural network model is used to predict the output fluctuation range of wind farms and photovoltaic stations in the next 1 to 6 hours (the prediction period can be adjusted according to actual conditions) to obtain the predicted power deviation reference value; the predicted power deviation reference value is judged to determine whether the preset conditions are met to determine whether the corresponding wind farm or photovoltaic station participates in emergency control.

[0038] When using the trained LSTM neural network model to predict the power deviation reference value, the input data is real-time meteorological data (such as wind speed, wind direction, air density of wind farms, light intensity, temperature, cloud cover, etc. of photovoltaic plants) and the weather forecast corresponding to the real-time meteorological data for the next 1 to 6 hours; the output is the predicted power deviation reference value of the wind farm and photovoltaic plant for the next 1 to 6 hours. Generally, if the predicted power deviation of a wind farm or photovoltaic plant is greater than the preset value, it is considered that the power fluctuation of the wind farm or photovoltaic plant is too large and needs to participate in emergency control, thereby obtaining the set of wind farms for emergency control. N w and photovoltaic station collection N p .

[0039] The LSTM neural network model has a dual-channel structure, processing wind power and photovoltaic time series data respectively. The training method of the LSTM neural network model is as follows: (1) Data collection: Obtain a historical meteorological data set, including a historical meteorological data set F1 of a wind farm station, where F1 includes wind speed, wind direction, and air density of the wind farm station; a historical meteorological data set G1 of a photovoltaic station, where G1 includes light intensity, temperature, and cloud cover of the photovoltaic station; a historical forecast meteorological data set F2 for the next 1 to 6 hours corresponding to the weather forecast of the historical meteorological data set F1, and a historical forecast meteorological data set G2 for the next 1 to 6 hours corresponding to the weather forecast of the historical meteorological data set G1; Obtain a historical output data set M1 of the wind farm corresponding to F1 and a historical output data set N1 of the photovoltaic farm corresponding to G1, obtain a historical output data set M2 of the wind farm corresponding to F1 for the next 1 to 6 hours and a historical output data set N2 of the photovoltaic farm corresponding to G1 for the next 1 to 6 hours; both historical output data sets M1 and M2 include the power deviation of the wind farm, and both historical output data sets N1 and N2 include the power deviation of the photovoltaic farm; the power deviation of the wind farm is obtained by calculating the difference between the actual power and the rated power of the wind farm; the power deviation of the photovoltaic farm is obtained by calculating the difference between the actual power and the rated power of the photovoltaic farm; The above meteorological data sets F1, G1, F2, and G2 are standardized according to the historical output data sets M1, N1, M2, and N2 to obtain the corresponding time series.

[0040] (2) Model construction: A dual-channel LSTM neural network model is designed, including a wind power channel and a photovoltaic channel, which process the time series of wind farms and photovoltaic stations respectively; the wind power channel inputs the time series of the historical meteorological data set F1 of the wind farm, the time series of the historical forecast meteorological data set F2, and the historical output data set M1; the photovoltaic channel inputs the time series of the historical meteorological data set G1 of the photovoltaic power station, the corresponding time series of the historical forecast meteorological data set G2, and the historical output data set N1.

[0041] (3) Model training: By sharing hidden layers, the spatiotemporal correlation of wind and solar power output is captured, and the wind power deviation and photovoltaic power deviation for the next 1 to 6 hours are output. A joint loss function is used for model training, and adversarial training is introduced to enhance the generalization ability in extreme scenarios.

[0042] Among them, the joint loss function is: ; is the total loss value, which represents the overall error of the model prediction and is used to optimize the parameters of the LSTM neural network model, that is, the mean square error of the power deviation of the wind farm and the photovoltaic farm; and is the weight coefficient, which is used to balance the importance of wind farm power deviation and photovoltaic farm power deviation in the total loss, and is usually set equal to 0.5; , are the mean square errors of power deviations of photovoltaic stations and wind farms, respectively.

[0043] Based on the historical output data set M2 for wind farms and N2 for photovoltaic plants, the mean squared difference between the actual deviations (M2 and N2) and the median of the prediction interval is calculated. If the mean squared difference meets the preset conditions, the LSTM neural network model is trained. If the mean squared difference does not meet the preset conditions, the data collection range is expanded and training is repeated.

[0044] The specific steps of adversarial training are as follows: (1) Random noise and meteorological labels (such as typhoons, sandstorms, etc.) are input into the generator of the LSTM neural network model to generate simulated extreme meteorological data (such as abnormal wind speed and sudden drop in irradiance); the discriminator of the LSTM neural network model distinguishes between real data and generated data, forcing the generator to approach the characteristics of real extreme scenes; (2) The generated data and real data are mixed and input into the LSTM neural network model to jointly optimize the prediction error and improve the model's sensitivity to extreme data; (3) Periodically inject sudden meteorological events simulated by the generator (such as a sudden increase in wind speed, a sudden change in cloud cover, etc.) to test the model's prediction robustness; if the predicted value exceeds the limit, adjust the generator parameters to generate more stringent scenarios and iterate the training; (4) Based on the performance of the model on the generated data, the extreme parameters of the generator (such as maximum wind speed, cloud coverage, etc.) are dynamically enhanced, and the training data boundaries are continuously expanded to ensure that the model adapts to unknown meteorological conditions.

[0045] S2. Obtain EV active power output model The Monte Carlo sampling method is used to construct the EV output model. The probability density function type and parameter value settings for the probability parameters of EVs in wind farms and photovoltaic farms are shown in Table 1: Table 1EV probability density function values

[0046] Based on the corresponding probability density function and the corresponding parameter values, the Monte Carlo method is used to predict the probability parameters of the total number of EVs arriving at charging piles (including wind farms and photovoltaic farms), arrival time, departure time, and initial SOC. The predicted values ​​of the number of EVs arriving and leaving the charging station, the arrival time of the charging station, the departure time of the charging station, and the initial SOC are obtained respectively; by increasing the number of sampling times, the expected results are obtained.

[0047] Based on the predicted values ​​of the time when the EV arrives at and leaves the charging station, as well as the extreme weather period predicted by the weather forecast, it is determined whether the EV can participate in the emergency control for the weather conditions at that stage, thereby obtaining the EV set participating in the emergency control. .

[0048] The set of EVs involved in emergency control obtained based on the Monte Carlo sampling results of EVs , and the constraints during EV charging and discharging, the EV active output can be derived as follows: ; Where, For the t Period i Active power output of EVs participating in emergency control; 、 For the i Active charging and discharging output of EVs participating in emergency control; is the set of EVs that can participate in emergency control.

[0049] Among them, the constraints during EV charging and discharging are: ; Where: For the i EVs in t Charging power during the time period; For the i Minimum charging power for EVs; For the i Maximum charging power for EVs; For the i The minimum value of the EV's SOC safety range (generally 20%), For the i The maximum value of the EV's SOC safety range (generally 90%); For the i The battery capacity of the EV; For the i The charging and discharging efficiency of EVs; To control the time step; for t Wind and solar power generation during the period; for t Energy storage output during the time period; for t Grid loss power within the time period; for t Load power during the time period.

[0050] S3. Based on the actual output of new energy stations under extreme weather conditions, establish an emergency control allocation model involving EVs.

[0051] In the allocation and dispatch of new energy stations considering EV recovery characteristics, the importance of new energy stations can be divided into different levels according to the impact of extreme weather on new energy stations and the size of the economic losses caused. The impact of different stations at different times during extreme weather is also different. Important and serious new energy stations should be prioritized for emergency optimization. Therefore, new energy stations should be ranked according to their importance. t Priority electricity consumption coefficient for time period for: ; In the above formula, is the priority electricity coefficient; The importance coefficient of the new energy station. Generally, it can be divided into three levels: low, medium and high according to the frequency deviation and power shortage of the new energy station. The values ​​are 1, 5 and 10 respectively from low to high according to the importance of the new energy station; The power consumption coefficient of the new energy station under the time scale is in the range of 1-10. If a certain moment is the most severe period of extreme weather impact, Set to 10, the remaining periods of this time scale will be Gradually decrease.

[0052] In the scheduling of new energy stations that consider the recovery characteristics of EVs, the output of distributed power sources varies in different time periods under extreme weather conditions. Furthermore, the charging and discharging power of EVs available at charging stations (wind farms and photovoltaic stations) also varies significantly during different time periods. To determine the impact of EVs at charging stations on the emergency control of new energy stations under extreme weather conditions, a time-based function is used to establish an allocation model for EV participation in emergency control: ; ; In the above formula, For the t Power deviation of wind farms in each period; For the t Power deviation of PV stations in each period; t The selected time period can be adjusted according to the needs. Generally, each time period is 1 hour, and 24 time periods are selected per day. n w is the number of wind farm stations; n p is the number of photovoltaic stations; For the i Power deviation of each wind farm; ω wiFor the i The priority electricity consumption coefficient of each wind farm station, ω pi For the i The priority electricity consumption coefficient of each photovoltaic station; For the l Power deviation of each photovoltaic station; y wi For the i The output power status of each wind farm station is 1 or 0, 1 is overload status, 0 is full load or under load status, y pl For the l The output power status of a photovoltaic station, the value is 1 or 0, 1 is overload status, 0 is full load or underload status.

[0053] With the goal of minimizing power variation, the constraints of EV (Electric Vehicle) charging power range, EV SOC (State of Charge) safety range, and EV charging power per unit time period are used. The decision variables are the charging active power of the EV set participating in emergency control, the power deviation of the wind farm set participating in emergency control, the power deviation of the photovoltaic set participating in emergency control, and the charging power of the energy storage set. An objective function is constructed. The obtained objective function is: ; Where: n ev is the number of EVs that can participate in emergency control, For the d The charging active power allocated to each EV; W t for t The total difference between the active power of the wind farms and the rated active power in the time period; P t for t The total difference between the active power of the photovoltaic stations and the rated active power during the time period; n c is the number of energy storage stations, P ck For the k The charging power of each energy storage station; c 1. c 2. c 3. c 4 are the weight values ​​in the objective function, which are usually determined according to the actual needs of emergency control.

[0054] The above objective function is the emergency control allocation model with EV participation.

[0055] With the goal of minimizing the power change of the objective function, the improved whale algorithm is used to find the optimal solution and obtain the active power allocation scheme.

[0056] Input basic data on the original composition of new energy stations into the aforementioned EV-involved emergency control allocation model, including wind farms, photovoltaic stations, and energy storage stations, as well as their output values ​​under extreme weather conditions; The wind farm station set of emergency control obtained by LSTM neural network model N w and photovoltaic station collection N p , the EV set participating in emergency control obtained by Monte Carlo sampling method and energy storage station collection N c Input the emergency control allocation model with EV participation, update the relevant parameters in the wind, solar and storage combined output model, and further update the relevant parameters of the objective function; The population position is initialized according to the updated objective function definition, with the goal of minimizing power change. The objective function is optimized iteratively using the improved whale algorithm. i The position of the whale is calculated, and the fitness of the individual whale is calculated using the objective function; then the position of the population is disturbed, the population position is updated and iterated again to obtain the updated first i The position of the whale is calculated and the fitness is calculated; it is iterated continuously. If the obtained fitness does not meet the preset iteration stop threshold, the population is disturbed again and iterated again until the fitness reaches the preset iteration stop threshold. The algorithm is terminated and the optimal fitness value is obtained.

[0057] Generally, the preset iteration stop threshold is the power variation range of the objective function. The power variation can be set to not exceed a certain value according to actual conditions. This is the preset iteration stop threshold.

[0058] Based on the optimal solution obtained by the whale algorithm, the active power allocation plan of the new energy station under extreme weather conditions is determined, and the active power of the new energy station is allocated and dispatched according to the obtained active power allocation plan.

[0059] The steps of iteratively optimizing the objective function using the improved whale algorithm with the goal of minimizing power change are as follows: According to the objective function, a whale population is randomly generated; the position coordinates of the whales correspond to the charging active power of the EV set participating in emergency control, the power deviation of the wind farm set participating in emergency control, the power deviation of the photovoltaic station set participating in emergency control, and the charging power of the energy storage station set in the objective function; the optimized bubble attack method is used to attack the whale position, and the attacked whale updates its position according to a spiral route. The optimized prey search method is used to conduct a global search for the whale position.

[0060] Specifically, in order to make the convergence factor a As the number of iterations gradually decreases from 1 to 0, the bubble attack method iterations are optimized: ; Where: t is the current iteration number; M is the maximum number of iterations; the convergence factor of the bubble attack method in this embodiment is a The exponential function is introduced to optimize a The decay rate changes adaptively, without the need to manually adjust the decay curve, making it more universal.

[0061] In the initial stage of iteration, a The decay rate of the convergence factor is lower than that of the original whale algorithm. The algorithm can maintain a larger search step for a longer time, which helps to jump out of the local optimum and find the global optimal solution more easily. In the subsequent iterations, a It approaches 0 quickly, allowing the algorithm to focus on local fine search and accelerate convergence to the optimal solution.

[0062] The spiral optimization update position is: ; In the above formula: is the distance between the i-th whale and the target position; is the optimal agent position at the current moment; For the i The individual has a number of iterations t The iteration position when ; For the i The individual has a number of iterations t+ Iteration position at 1: b is the constant of the spiral equation; l is a random number in the range [-1,1]; Update the value of the dimension of the current whale's position; Optimize the method of searching for prey, i The first whale position is updated as follows: ; In the above formula:D For the i The distance between the whale and the target location; C is the vector coefficient; It represents the position vector of a randomly selected whale; is the position vector of the current individual; For the i The probability of the search location of a whale; exp is the Logistic function; A For variables i Control of the global search and local development capabilities of whales; a is the convergence factor, ranging from [2,0]; Is to introduce variables A The probability after for No. j dimensions; Update the value of the dimension of the current whale's position; The current optimal agent j The value of the dimension; r is a random number in [0,1]; rand A random number in [0,1].

[0063] In addition, in the process of optimizing using the whale algorithm, the constraints of the objective function are: ; In the above formula: For the i EVs in t Charging power during the time period; For the i Minimum charging power for EVs; For the i Maximum charging power for EVs; For the i The minimum value of the EV's SOC safety range (generally 20%), For the i The maximum value of the EV's SOC safety range (generally 90%); For the i The battery capacity of the EV; For the i The charging and discharging efficiency of EVs; To control the time step; for t Wind and solar power generation during the period; for t Energy storage output during the time period; for t Grid loss power within the time period; for t Load power during the time period.

[0064] The convergence factor is optimized in the process of optimizing the objective function using the improved whale algorithm. a , reducing the oscillation of the algorithm during the optimization process, improving stability, and ensuring convergence speed and solution accuracy; this improvement enables the whale algorithm to achieve a better balance between global search capability and local convergence speed, ensuring the optimal solution.

[0065] In the process of obtaining the set of EVs that can participate in emergency control, the Monte Carlo sampling method is used to accurately predict the time when EVs arrive at and leave wind farms or photovoltaic stations without relying on strict analytical models, thereby improving the real-time nature of the prediction, which is particularly suitable for regional wind and photovoltaic power plant coordinated scheduling; in the process of obtaining the set of wind farms and photovoltaic stations that can participate in emergency control, A dual-channel LSTM neural network model is adopted, and a joint loss function is used for model training. Adversarial training is introduced to enhance the generalization ability of extreme scenarios. The resulting LSTM neural network model can effectively capture the historical changes in meteorological factors such as wind speed, light, and temperature, and is highly sensitive to extreme weather data. It can effectively improve the stability and reliability of the prediction results, thereby improving the rationality of the obtained emergency control plan, realizing the rational scheduling of new energy resources, and improving the stability of new energy stations under extreme weather conditions.

[0066] The emergency control method for new energy stations under extreme weather conditions in this embodiment is simple and easy to operate, has strong search capabilities, and is highly adaptable. It can achieve reasonable scheduling of new energy resources and improve the stability of new energy stations under extreme weather conditions. The improved whale algorithm is used to obtain the active power distribution scheme. The algorithm is simple and easy to find the optimal solution, and it implements global search to avoid local optimal solutions, thereby improving the rationality of new energy station scheduling under extreme weather conditions and ensuring the stability of new energy stations. In the scheme of incorporating EV into power distribution scheduling under extreme weather conditions, representative distributed resources are added, making the scheduling strategy more diverse and enabling more reasonable utilization of the resources of new energy stations.

[0067] Example 2 An emergency control device for a new energy station under extreme weather conditions, comprising: Model building module, used to determine the wind, solar and energy storage combined output model and the EV active power output model; an objective function construction module for constructing an objective function based on the wind-solar-storage combined output model and the EV active output model, wherein the objective function uses the charging active power of the set of EVs participating in emergency control, the power deviation of the set of wind farms participating in emergency control, the power deviation of the set of photovoltaic farms participating in emergency control, and the charging power of the set of energy storage farms as decision variables to output an active power allocation plan; The scheme generation module is used to optimize the objective function using the improved whale algorithm, and to allocate and dispatch the active power of the new energy stations according to the obtained active power allocation scheme.

[0068] Example 3 A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for emergency control of a new energy station under extreme weather conditions of Example 1 is implemented.

[0069] Example 4 A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the emergency control method for a new energy station under extreme weather conditions of Example 1.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A new energy station emergency control method under extreme weather conditions, characterized in that: The following steps are involved: Determine the wind, solar and energy storage combined output model and the EV active power output model; An objective function is constructed based on the wind-solar-storage combined output model and the EV active output model. The objective function uses the charging active power of the set of EVs participating in emergency control, the power deviation of the set of wind farms participating in emergency control, the power deviation of the set of photovoltaic farms participating in emergency control, and the charging power of the set of energy storage farms as decision variables to output an active power allocation plan. The improved whale algorithm is used to optimize the objective function, and the active power of the new energy stations is allocated and dispatched according to the obtained active power allocation scheme.

2. The method for emergency control of new energy stations under extreme weather conditions according to claim 1, characterized in that: The meteorological data includes air density, actual wind speed, and light intensity.

3. The method for emergency control of new energy stations under extreme weather conditions according to claim 1, characterized in that: The Monte Carlo sampling method is used to predict the arrival and departure time of EVs at wind farms or photovoltaic stations, and to determine the set of EVs participating in emergency control.

4. The method for emergency control of new energy stations under extreme weather conditions according to claim 1, characterized in that: A pre-trained LSTM neural network model is used to predict the power deviation of wind farms and photovoltaic stations based on real-time meteorological data and predicted meteorological data. The set of wind farms and photovoltaic stations participating in emergency control is determined based on whether the predicted power deviation exceeds the preset value.

5. The method for emergency control of new energy stations under extreme weather conditions according to claim 1, characterized in that: The power deviation of the wind farm set and the power deviation of the photovoltaic farm set are respectively related to the priority power consumption coefficients of the corresponding wind farms and photovoltaic farms, and the priority power consumption coefficients are time-dependent piecewise functions; The priority electricity consumption coefficient is the product of the level importance coefficient of the wind farm or photovoltaic farm and the electricity consumption coefficient under the time scale.

6. The method for emergency control of new energy stations under extreme weather conditions according to claim 1, characterized in that: The steps of optimizing the objective function using the improved whale algorithm include: Determine the population size of the whale algorithm, randomly generate whale positions, use the optimized bubble attack method to update the whale positions and search for the optimal solution.

7. The method for emergency control of new energy stations under extreme weather conditions according to claim 6, characterized in that: In the optimized bubble attack method, the convergence factor a Calculated according to the following formula: ； Where: t is the current iteration number; M is the maximum number of iterations.

8. An emergency control device for a new energy station under extreme weather conditions, characterized in that: include: Model building module, used to determine the wind, solar and energy storage combined output model and the EV active power output model; an objective function construction module for constructing an objective function based on the wind-solar-storage combined output model and the EV active output model, wherein the objective function uses the charging active power of the set of EVs participating in emergency control, the power deviation of the set of wind farms participating in emergency control, the power deviation of the set of photovoltaic farms participating in emergency control, and the charging power of the set of energy storage farms as decision variables to output an active power allocation plan; The scheme generation module is used to optimize the objective function using the improved whale algorithm, and to allocate and dispatch the active power of the new energy stations according to the obtained active power allocation scheme.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for emergency control of a new energy station under extreme weather conditions as claimed in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the method for emergency control of a new energy station under extreme weather conditions as claimed in any one of claims 1 to 7 is implemented.

Images