Source-grid-load-storage multidirectional cooperation method and system in new energy high-permeability power grid

Abstract

The invention discloses a source-grid-load-storage multidirectional cooperation method and system in a new energy high-permeability power grid, and belongs to the technical field of source-grid-load-storage collaboration.The method comprises the steps that a historical data set and social economic data of a power system in a region are obtained, and a prediction model is constructed to obtain a load prediction result; acquiring solar radiation quantity, wind energy resource distribution data and historical meteorological data in the region to obtain a power supply prediction result; obtaining a load prediction correction value based on the power supply prediction result; constructing a power balance model based on the load prediction correction value and the power supply prediction result; solving the established power balance model based on a genetic algorithm, and determining an operation scheme; and evaluating the determined operation scheme, and if the operation scheme meets requirements, outputting the operation scheme for source-network-load-storage multidirectional collaboration. According to the method, the load change rule can be comprehensively captured, and the problem of inaccurate load prediction caused by single data in the prior art is effectively solved.

Description

Technical Field

[0001] This invention belongs to the field of source-grid-load-storage coordination technology, specifically relating to a multi-directional coordination method and system for source-grid-load-storage in a new energy high-penetration power grid. Background Technology

[0002] Under the overarching trends of sustainable development and addressing climate change, the environmental and resource problems caused by traditional energy sources are driving countries to accelerate the transition of their energy structures towards clean and low-carbon energy, with the proportion of new energy sources such as solar and wind power in the power grid continuously increasing. However, the intermittent and volatile nature of new energy sources poses numerous challenges to the power system. It disrupts the original balance between power supply and demand, making the interrelationships between power generation, grid, load, and storage more complex, and traditional operation and management models are insufficient to cope. Simultaneously, the instability of new energy sources can cause voltage and frequency fluctuations in the power grid, threatening grid stability and reliability and increasing the risk of power outages. Moreover, the cost of new energy power generation is volatile and uncertain; improper planning and operation of the power generation, grid, load, and storage systems can increase grid operating costs.

[0003] Therefore, developing multi-directional collaborative technologies for power generation, grid, load, and storage in high-penetration power grids to achieve coordinated interaction among all links is crucial for ensuring stable and reliable power supply, improving system operating efficiency, reducing costs, and promoting the sustainable development of the power industry.

[0004] However, existing technologies have problems such as not fully integrating socio-economic data and various historical power system data, making it difficult to accurately capture the patterns of load and power supply changes. Furthermore, the prediction methods are simplistic, often employing simple models or single algorithms, which cannot adapt to the complex and ever-changing power system environment. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multi-directional collaborative method and system for energy sources, grids, loads, and storage in a high-penetration power grid, which can comprehensively capture load change patterns and effectively solve the problem of inaccurate load forecasting caused by single data in existing technologies.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A multi-directional collaborative method for energy sources, grids, loads, and energy storage in a high-penetration renewable energy grid includes the following steps: The system acquires historical power system data and socioeconomic data within the region, constructs a prediction model to obtain load forecast results, and the historical power system data includes historical load data, power generation, voltage, current, historical meteorological data, and historical electricity consumption data. The socioeconomic data includes total population, population density, and GDP. Data on solar radiation, wind energy resource distribution, and historical meteorological data within the region are acquired to obtain power generation forecasts, including predicted solar power output, predicted solar power generation, predicted wind power output, and predicted wind power generation. Based on the power supply forecast results, the load forecast correction value is obtained; A power balance model is constructed based on the load forecast correction value and the power source forecast result; The established power balance model is solved using a genetic algorithm to determine the operating scheme. The determined operation plan is evaluated. If the operation plan meets the requirements, the operation plan is output for multi-directional coordination of source, grid, load and storage.

[0007] Preferably, constructing a prediction model to obtain load prediction results includes the following steps: Obtain load-related parameters, combine them with historical load data, construct the first prediction model, and obtain the first load prediction value; Based on historical load data and load-related parameters, a second prediction model is constructed to obtain the second load prediction value; The combined load forecast value is obtained by combining the first load forecast value and the second load forecast value. The formula for calculating the comprehensive load forecast is as follows: ; In the formula, This is the comprehensive load forecast value. This is the first load forecast value. This is the second load forecast value. for Weighting factors for Weighting factors; Obtaining load-related parameters includes the following steps: The load-related parameters include multiple related parameters. Pearson correlation analysis is used to analyze the correlation coefficients between each related parameter and the load power. The correlation coefficients obtained from the analysis are compared with the correlation thresholds stored in the database. If the correlation coefficient is greater than the correlation threshold, the relevant parameter corresponding to the correlation coefficient is selected and recorded as the key parameter. If the correlation coefficient is not greater than the correlation threshold, the relevant parameter corresponding to that correlation coefficient will not be selected. The obtained key parameters are recorded as relevant key data, including temperature, temperature change rate, humidity, and solar radiation intensity.

[0008] Preferably, constructing the first prediction model includes the following steps: The relevant key data is imported into the neural network model to obtain the first prediction model; The least squares method is used to estimate the model parameters, calculate the coefficient of determination, and perform model testing. If the test statistic is greater than the critical value of the test statistic in the database, then the first prediction model is considered to be significant overall. Obtain relevant key data for the future, including predicted temperature, predicted rate of temperature change, predicted humidity, and predicted solar radiation intensity, import them into the first prediction model, and obtain the first load prediction. The formula for calculating the first load forecast value is: ; In the formula, T is the temperature. For reference temperature, For the rate of temperature change, Let S be the reference temperature change rate, and S be the solar radiation intensity. For reference to solar radiation intensity, H represents humidity. For reference humidity, The intercept is... for The slope, for The slope, for The slope, for The slope.

[0009] Preferably, the second prediction model is constructed, including the following steps: The relevant key data is imported into the neural network model to obtain the second prediction model; In the second prediction model: the input layer nodes include historical load data, temperature data, temperature change rate data, solar radiation intensity data, and humidity data. The first hidden layer has 10-15 neurons, the second hidden layer has 5-8 neurons, and the ReLU activation function is used. The output layer nodes are the second load prediction values. Based on historical load data and load-related parameters, a training set and a validation set are divided. The second prediction model is trained using the training set, and its performance is evaluated on the validation set. If the loss function value on the validation set does not decrease for several consecutive rounds, the second prediction model is considered to have converged, and training is stopped. The relevant key data for the future is imported into the second forecast model to obtain the second load forecast value.

[0010] Preferably, obtaining power prediction results includes the following steps: Based on the power generation principle of photovoltaic cells, combined with solar radiation intensity data and temperature data, the output power of photovoltaic cells is obtained, and the solar power generation is analyzed. By inputting solar radiation intensity data and temperature data into a neural network model, the predicted solar power generation is obtained, and the predicted solar power generation is analyzed. Wind energy resource distribution data includes wind speed, wind turbine angle of attack, and wind energy density. The wind energy resource distribution data and wind power generation data are recorded as wind power generation time series data and imported into an autoregressive integral moving average model to obtain the predicted wind power generation. The predicted wind power generation is then analyzed. The formula for calculating the output power of a photovoltaic cell is: ; In the formula, The output power of the photovoltaic cell. The reference output power is stored in the database. G represents the reference solar radiation intensity stored in the database, and G represents the actual solar radiation intensity. The reference operating temperature is stored in the database. The power temperature coefficient is expressed in units of 1 / ℃. This refers to the actual operating temperature. The formula for calculating solar power generation is: ; In the formula, for to Solar power generation during the time period for to The output power of the photovoltaic cell at time t within the time period. At the starting time, The end time; The expression for the autoregressive integral moving average model is: ; In the formula, Here, B is the autoregressive coefficient, B is the lag operator, and d is the difference order. Let be the wind power generation capacity at time t. The moving average coefficient is... It is a white noise sequence. for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right The influence coefficient, where jq is the order of the moving average; The formula for calculating wind power generation is: ; In the formula, for to Wind power generation during the time period for to The wind power generation capacity at time t within the time period.

[0011] Preferably, obtaining the load forecast correction value includes the following steps: Obtain the renewable energy power generation and the power generation that has been absorbed from the database, and calculate the renewable energy absorption rate. The formula for calculating the renewable energy absorption rate is: ; In the formula, Let be the renewable energy consumption rate at time t. Let be the power generation capacity of the h-th renewable energy source at time t. Let H be the renewable energy power generation of the h-th renewable energy source at time t, where h is the renewable energy source number and H is the total number of renewable energy sources. The ridge regression model is trained by obtaining the historical dataset stored in the database, and the load forecast correction value is calculated. The historical dataset includes historical renewable energy absorption rate data, historical renewable energy power generation data, and historical load forecast correction value data. The ridge regression model formula is: ; In the formula, This is the load forecast correction value at time t. Let be the renewable energy consumption rate at time t. Let be the total renewable energy generation capacity at time t, and k be the consumption-load coupling coefficient. is the regularization coefficient, M is the total number of sliding windows, and m is the time window number.

[0012] Preferably, constructing a power balance model includes the following steps: Based on load forecasting results and power supply forecasting results, the decision variables in the power balance model are determined. The decision variables are power generation capacity, energy storage device charging and discharging power, and energy storage device power. Define the model objectives, which include the objectives of balancing electricity supply and demand, minimizing grid operating costs, and maximizing the absorption of new energy sources. Establish constraints, including power balance constraints, power generation constraints, and energy storage device constraints. By integrating decision variables, model objectives, and constraints, a power balance model is established. The objective function for balancing electricity supply and demand is: ; In the formula, Let a be the power generation capacity of the a-th power source. The charging and discharging power of the b-th energy storage device. Let be the power consumption of the g-th load, where 'a' is the power source number, 'b' is the energy storage device number, and 'g' is the load number. This is the load forecast correction value for the g-th load. The objective function for minimizing power grid operating costs is: ; In the formula, C represents the total cost of power grid operation. Let be the power generation cost coefficient of the a-th power source. Let b be the operating cost coefficient for the b-th energy storage device. Let g be the power shortage cost coefficient for the g-th load. The formula for maximizing the utilization of new energy sources is: ; In the formula, To maximize the utilization rate of new energy sources, Let h be the power generation capacity of the h-th renewable energy source that is absorbed. The renewable energy power generation capacity of the h-th renewable energy source is... This is the correction term for the h-th new energy power source. Let h be the node allocation coefficient for the h-th renewable energy source. This is the load forecast correction value for the h-th renewable energy source. Power generation capacity constraints, the specific constraints are as follows: The power generation capacity of the power source is within its power generation capacity configuration range, that is: ; in, Let be the minimum generating power of the a-th power source. The maximum power output of the a-th power source is... Let g be the power consumption of the g-th load. This is the total load correction for the power supply zone where power source a is located; Constraints on energy storage devices, specifically: The energy storage device's capacity is within its configured capacity range, i.e.: ; in, Let b be the minimum power of the energy storage device. Let b be the maximum power of the energy storage device. This represents the actual power output of the b-th energy storage device.

[0013] Preferably, the established power balance model is solved using a genetic algorithm to determine the operating scheme, including the following steps: S1. Encode the decision variables; S2. Randomly generate the initial population; S3. Constructing the fitness function: Normalize and weight the objectives of power supply and demand balance, minimum grid operation cost, and maximum renewable energy consumption; S4. Select superior individuals from the current population to enter the next generation population based on the roulette wheel selection method; S5. Perform a crossover operation on the selected next generation population to obtain the crossover population; S6. Perform mutation operation on the crossover population to obtain the mutated population; S7. Repeat steps S4-S6 until the maximum number of iterations is reached. Stop the calculation and output the current optimal solution. The current optimal solution is the running plan.

[0014] Preferred method: Evaluate the determined operating plan, including the following steps: After obtaining the grid power after operation, analyze and obtain the power supply and demand balance assessment indicators, and conduct a power supply and demand balance assessment. After the grid is put into operation, the grid stability assessment index is obtained by analysis, and the grid stability assessment is carried out. Calculate the renewable energy consumption index after operation, and evaluate the renewable energy consumption effect based on the renewable energy consumption index; If the power supply and demand balance assessment index, the power grid stability assessment index, and the new energy consumption index are all less than the corresponding thresholds, then the power supply and demand balance assessment index, the power grid stability assessment index, and the new energy consumption index will be comprehensively analyzed to obtain the comprehensive operation assessment index. The overall operational evaluation index is compared with the operational evaluation thresholds stored in the database. If the overall operational evaluation index is less than the operational evaluation threshold, the corresponding operational plan meets the requirements. If the overall operational evaluation index is not less than the operational evaluation threshold, then the corresponding operational plan does not meet the requirements and needs to be readjusted. If any of the power supply and demand balance assessment indicators, power grid stability assessment indicators, and new energy consumption indicators is not less than the corresponding threshold, then the parameters of the scheme corresponding to that indicator will be adjusted.

[0015] A multi-directional collaborative system for energy sources, grids, loads, and energy storage in a high-penetration renewable energy grid, used to implement the above method, includes: The data acquisition module is used to acquire historical datasets of the power system, socio-economic data, solar radiation, wind energy resource distribution data, and historical meteorological data, and to transmit and integrate the acquired data. The data communication and transmission module is used to receive data acquired by the data acquisition module. The data communication and transmission module is connected to the load forecasting module, the power supply forecasting module, the load forecasting correction module, the power balance model construction module, the operation scheme determination module, the scheme evaluation module, and the data storage module. The load forecasting module is used to acquire historical power system data and socio-economic data within the region to obtain load forecasting results. The historical power system data includes historical load data, power generation, voltage, current, historical meteorological data, and historical electricity consumption data. The socio-economic data includes total population, population density, and GDP. The power prediction module is used to acquire data on solar radiation, wind energy resource distribution, and historical meteorological data within the region to obtain power prediction results, including predicted solar power generation, predicted solar power output, predicted wind power generation, and predicted wind power output. The load forecasting correction module is used to input the power forecasting results from the power forecasting module into the load forecasting module to obtain the load forecasting correction value. The power balance model building module is used to build a power balance model based on load forecast correction values ​​and power source forecast results. The operation scheme determination module is used to solve the established power balance model based on the genetic algorithm and determine the operation scheme. The scheme evaluation module is used to evaluate the determined operation scheme; The data storage module is used to store the collected data.

[0016] The present invention has the following beneficial effects: This invention overcomes the limitations of traditional forecasting, which relies solely on a single type of data, by comprehensively considering socio-economic data and various historical power system data. It can comprehensively capture load change patterns and effectively solve the problem of inaccurate load forecasting caused by limited data in existing technologies. By constructing first and second forecasting models and comprehensively analyzing their results, it can balance linear and nonlinear relationships, adapt to complex and ever-changing power system environments, improve overall forecast accuracy, and correct load forecast deviations in real time, ensuring forecast accuracy. Using power generation capacity, energy storage device charging and discharging power, and energy consumption as decision variables, it constructs a power balance model that comprehensively covers key elements of power generation, grid, load, and storage, achieving multiple synergies, optimizing power generation and energy storage strategies, reducing operating costs while increasing the absorption rate of new energy sources, improving the reliability and economy of power system operation, and ensuring the safe and stable operation of the power grid. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system module diagram of the present invention. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0019] Example 1: As Figure 1 As shown, a multi-directional collaborative method for energy source, grid, load, and storage in a high-penetration renewable energy power grid includes the following steps: The system acquires historical power system data and socioeconomic data within the region, constructs a prediction model to obtain load forecast results, and the historical power system data includes historical load data, power generation, voltage, current, historical meteorological data, and historical electricity consumption data. The socioeconomic data includes total population, population density, and GDP. Data on solar radiation, wind energy resource distribution, and historical meteorological data within the region are acquired to obtain power generation forecasts, including predicted solar power output, predicted solar power generation, predicted wind power output, and predicted wind power generation. Based on the power supply forecast results, the load forecast correction value is obtained; A power balance model is constructed based on the load forecast correction value and the power source forecast result; The established power balance model is solved using a genetic algorithm to determine the operating scheme. The determined operation plan is evaluated. If the operation plan meets the requirements, the operation plan is output for multi-directional coordination of source, grid, load and storage.

[0020] The process of constructing a forecasting model to obtain load forecasting results includes the following steps: Obtain load-related parameters, combine them with historical load data, construct the first prediction model, and obtain the first load prediction value; Based on historical load data and load-related parameters, a second prediction model is constructed to obtain the second load prediction value; The combined load forecast value is obtained by combining the first load forecast value and the second load forecast value. The formula for calculating the comprehensive load forecast is as follows: ; In the formula, This is the comprehensive load forecast value. This is the first load forecast value. This is the second load forecast value. for Weighting factors for Weighting factors; , It is obtained from the database and calculated based on historical data. , as well as ,Establish , The mapping set of its corresponding weight factors is used to obtain the current... , The following text , , , , , All of these are obtained through a mapping set of historical data and weight factors established in the database, that is, the corresponding weight factors are obtained based on the current data.

[0021] Obtaining load-related parameters includes the following steps: The load-related parameters include multiple related parameters. Pearson correlation analysis is used to analyze the correlation coefficients between each related parameter and the load power. The correlation coefficients obtained from the analysis are compared with the correlation thresholds stored in the database. If the correlation coefficient is greater than the correlation threshold, the relevant parameter corresponding to the correlation coefficient is selected and recorded as the key parameter. If the correlation coefficient is not greater than the correlation threshold, the relevant parameter corresponding to that correlation coefficient will not be selected. The obtained key parameters are recorded as relevant key data, including temperature, temperature change rate, humidity, and solar radiation intensity.

[0022] The formula for calculating the correlation coefficient is: ; In the formula, Here, E represents the comprehensive load forecast, Z represents the relevant parameters, f represents the load power, and s represents the sample number. Let f be the f-th sample value of E. Let E be the mean. Let f be the f-th sample value of Z. Let Z be the mean of Z.

[0023] The first prediction model is constructed by the following steps: The relevant key data is imported into the neural network model to obtain the first prediction model; The least squares method is used to estimate the model parameters, calculate the coefficient of determination, and perform model testing. If the test statistic is greater than the critical value of the test statistic in the database, then the first prediction model is considered to be significant overall. Obtain relevant key data for the future, including predicted temperature, predicted rate of temperature change, predicted humidity, and predicted solar radiation intensity, import them into the first prediction model, and obtain the first load prediction. The formula for calculating the first load forecast value is: ; In the formula, T is the temperature. For reference temperature, For the rate of temperature change, Let S be the reference temperature change rate, and S be the solar radiation intensity. For reference to solar radiation intensity, H represents humidity. For reference humidity, The intercept is... for The slope, for The slope, for The slope, for The slope.

[0024] Constructing a second prediction model includes the following steps: The relevant key data is dimensionless (e.g., Z-score normalization) and imported into a neural network model (a feedforward neural network is used in this embodiment) to obtain the second prediction model. The input layer nodes include historical load data, temperature data, temperature change rate data, solar radiation intensity data, and humidity data. The first hidden layer has 10-15 neurons, and the second hidden layer has 5-8 neurons, using the ReLU activation function. The output layer nodes are the second load prediction values. Based on the historical load data and load-related parameters, a training set and a validation set are created. The second prediction model is trained using the training set, and its performance is evaluated on the validation set. If the loss function value on the validation set does not decrease for several consecutive rounds, the second prediction model is considered converged, and training is stopped. Future relevant key data is then imported into the second prediction model to obtain the second load prediction values.

[0025] Load forecasting is performed by acquiring historical power system datasets and socioeconomic data, comprehensively considering historical load data, power generation, voltage, current, meteorological data, electricity consumption data, as well as factors such as total population, population density, and GDP. This reflects various factors influencing load from different perspectives, comprehensively capturing patterns in load changes and improving forecast accuracy.

[0026] A first and second prediction model are constructed, and their prediction results are combined and analyzed to obtain a comprehensive load forecast. The first prediction model uses multiple linear regression to establish a linear relationship based on key parameters, which can intuitively reflect the linear correlation between parameters and load. The second prediction model utilizes neural networks, whose powerful nonlinear fitting ability can uncover complex nonlinear relationships in the data. The two models complement each other, improving the adaptability to different load change patterns and thus enhancing the overall prediction accuracy.

[0027] Pearson correlation analysis was used to screen key parameters among load-related parameters, identifying temperature, temperature change rate, humidity, and solar radiation intensity as key influencing factors. From numerous potential influencing factors, those with strong correlation to load power were identified, avoiding interference from irrelevant or weakly correlated factors in the prediction model. This allowed the model to focus more on key factors, improving its efficiency and prediction accuracy.

[0028] The parameters of the first forecasting model were estimated using the least squares method, the coefficient of determination was calculated, and the model was tested to evaluate its goodness of fit to the data and its significance. The coefficient of determination measures the model's ability to interpret the data, and the comparison of the test statistic with the critical value determines whether the model is significant overall. If the model passes the test, it indicates that it is reliable and effective, can reasonably describe the relationship between key factors and load, and provides a reliable basis for load forecasting.

[0029] The second prediction model determines convergence based on changes in the loss function value on the validation set. This prevents overfitting and ensures good generalization ability even on unknown data. Training stops only when the model converges, guaranteeing continuous optimization during training and improving predictive performance.

[0030] Load forecasting, as a crucial input to power balance models, directly impacts decisions regarding power generation, energy storage charging and discharging power, and electricity consumption. Load forecasting helps in the rational planning of power generation, avoiding over- or under-generation; optimizing energy storage charging and discharging strategies to improve energy efficiency; achieving coordination among power generation, grid, load, and storage; ensuring stable power system operation; promoting the efficient absorption of new energy sources; and reducing grid operating costs.

[0031] Obtaining power prediction results includes the following steps: Based on the power generation principle of photovoltaic cells, combined with solar radiation intensity data and temperature data, the output power of photovoltaic cells is obtained, and the solar power generation is analyzed. By inputting solar radiation intensity data and temperature data into a neural network model, the predicted solar power generation is obtained, and the predicted solar power generation is analyzed. Wind energy resource distribution data includes wind speed, wind turbine angle of attack, and wind energy density. The wind energy resource distribution data and wind power generation data are recorded as wind power generation time series data and imported into an autoregressive integral moving average model to obtain the predicted wind power generation. The predicted wind power generation is then analyzed. The formula for calculating the output power of a photovoltaic cell is: ; In the formula, The output power of the photovoltaic cell. The reference output power is stored in the database. G represents the reference solar radiation intensity stored in the database, and G represents the actual solar radiation intensity. The reference operating temperature is stored in the database. The power temperature coefficient is expressed in units of 1 / ℃. This refers to the actual operating temperature. The formula for calculating solar power generation is: ; In the formula, for to Solar power generation during the time period for to The output power of the photovoltaic cell at time t within the time period. At the starting time, The end time; The expression for the autoregressive integral moving average model is: ; In the formula, Here, B is the autoregressive coefficient, B is the lag operator, and d is the difference order. Let be the wind power generation capacity at time t. The moving average coefficient is... It is a white noise sequence. for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right The influence coefficient, where jq is the order of the moving average; The formula for calculating wind power generation is: ; In the formula, for to Wind power generation during the time period for to The wind power generation capacity at time t within the time period.

[0032] Reference output power and reference solar radiation intensity data provide a standardized benchmark for calculations, while the power temperature coefficient considers the impact of temperature on power generation, making the calculation results more realistic. Simultaneously, a neural network model is used to process solar radiation intensity and temperature data to predict solar power generation. The powerful nonlinear fitting ability of neural networks can uncover complex potential relationships between data points, improving prediction accuracy. Accurate solar power generation forecasts provide reliable information for grid dispatch, helping to rationally plan the generation of other power sources and ensure a balance between power supply and demand.

[0033] This method effectively captures the time-series characteristics of wind power generation, considering the correlation between power at different times in historical data (reflected by autoregressive coefficients) and random fluctuations (reflected by white noise sequences and moving average coefficients). By comprehensively considering these factors, it achieves accurate prediction of wind power generation, providing strong support for the rational integration and dispatch of wind power.

[0034] In power balance models, charging and discharging strategies for energy storage devices are rationally arranged based on the power generation capacity and load demand to achieve power supply and demand balance, reduce grid operating costs, and improve the absorption rate of new energy sources. Accurate power source forecasting results enable models to more accurately simulate the operating state of the power system, providing a data foundation for formulating reasonable operating plans.

[0035] When renewable energy generation is sufficient, priority should be given to consuming renewable energy, reducing traditional energy generation, and lowering carbon emissions. When renewable energy generation is insufficient, energy storage devices should be discharged or traditional energy generation should be increased to ensure a stable power supply. Optimization of collaborative strategies based on accurate forecasting can improve the overall operating efficiency and economy of the power system.

[0036] By understanding the power and electricity output of new energy generation in advance, the power grid can prioritize the allocation of new energy power to loads with demand, make full use of new energy resources, increase the proportion of new energy in the power grid, promote the transformation of the energy structure towards cleaner and lower carbon emissions, realize the coordinated operation of energy storage devices and new energy power generation, and improve the efficiency of new energy consumption.

[0037] Obtaining the load forecast correction includes the following steps: Obtain the renewable energy power generation and the power generation that has been absorbed from the database, and calculate the renewable energy absorption rate. The formula for calculating the renewable energy absorption rate is: ; In the formula, Let be the renewable energy consumption rate at time t. Let be the power generation capacity of the h-th renewable energy source at time t. Let H be the renewable energy power generation of the h-th renewable energy source at time t, where h is the renewable energy source number and H is the total number of renewable energy sources. The ridge regression model is trained by obtaining the historical dataset stored in the database, and the load forecast correction value is calculated. The historical dataset includes historical renewable energy absorption rate data, historical renewable energy power generation data, and historical load forecast correction value data. The ridge regression model formula is: ; In the formula, This is the load forecast correction value at time t. Let be the renewable energy consumption rate at time t. Let be the total renewable energy generation capacity at time t, and k be the consumption-load coupling coefficient. is the regularization coefficient, M is the total number of sliding windows, and m is the time window number.

[0038] The prediction feedback loop dynamically couples renewable energy absorption rate with load demand, corrects prediction deviations in real time, and injects them into the power balance model. It utilizes a hybrid architecture of linear regression and neural networks to take into account both physical laws and nonlinear characteristics, and continuously optimizes the dynamic coupling coefficients through an online learning mechanism. It simultaneously integrates absorption rate, economic efficiency, and stability objectives into the power balance model, and expands the operating boundaries of power sources and energy storage through elastic constraints to achieve minute-level response and regional autonomous optimization, reducing cross-regional power impact. Relying on the closed-loop iterative mechanism of "prediction-scheduling-evaluation", it drives the model to adaptively evolve, improving the absorption capacity and operational economy in scenarios with a high proportion of renewable energy.

[0039] Constructing a power balance model includes the following steps: Based on load forecasting results and power supply forecasting results, the decision variables in the power balance model are determined. The decision variables are power generation capacity, energy storage device charging and discharging power, and energy storage device power. Define the model objectives, which include the objectives of balancing electricity supply and demand, minimizing grid operating costs, and maximizing the absorption of new energy sources. Establish constraints, including power balance constraints, power generation constraints, and energy storage device constraints. By integrating decision variables, model objectives, and constraints, a power balance model is established. The objective function for balancing electricity supply and demand is: ; In the formula, Let a be the power generation capacity of the a-th power source. The charging and discharging power of the b-th energy storage device. Let be the power consumption of the g-th load, where 'a' is the power source number, 'b' is the energy storage device number, and 'g' is the load number. This is the load forecast correction value for the g-th load. The objective function for minimizing power grid operating costs is: ; In the formula, C represents the total cost of power grid operation. Let be the power generation cost coefficient of the a-th power source. Let b be the operating cost coefficient for the b-th energy storage device. Let g be the power shortage cost coefficient for the g-th load. The formula for maximizing the utilization of new energy sources is: ; In the formula, To maximize the utilization rate of new energy sources, Let h be the power generation capacity of the h-th renewable energy source that is absorbed. The renewable energy power generation capacity of the h-th renewable energy source is... This is the correction term for the h-th new energy power source. Let h be the node allocation coefficient for the h-th renewable energy source. This is the load forecast correction value for the h-th renewable energy source. Power generation capacity constraints, the specific constraints are as follows: The power generation capacity of the power source is within its power generation capacity configuration range, that is: ; in, Let be the minimum generating power of the a-th power source. The maximum power output of the a-th power source is... Let g be the power consumption of the g-th load. This is the total load correction for the power supply zone where power source a is located; Constraints on energy storage devices, specifically: The energy storage device's capacity is within its configured capacity range, i.e.: ; in, Let b be the minimum power of the energy storage device. Let b be the maximum power of the energy storage device. This represents the actual power output of the b-th energy storage device.

[0040] The model explicitly uses power generation capacity, energy storage device charging and discharging power, and electricity consumption as decision variables, and sets power supply and demand balance, minimum grid operating costs, and maximum renewable energy consumption as its objectives. This approach comprehensively covers key elements across all stages of the power system, breaking away from the traditional model of independent operation of different parts and enabling coordinated planning and optimization of power generation, energy storage, and electricity consumption. By comprehensively considering these factors, the power system can ensure power supply reliability while reducing operating costs and maximizing the utilization of renewable energy, thus promoting energy structure transformation.

[0041] After determining the decision variables and objectives, the model, by constructing corresponding objective functions and constraints, can find the optimal operating scheme while meeting various practical limitations. The model can rationally allocate the power generation capacity of each power source and the charging and discharging power of energy storage devices, ensuring precise matching between power supply and load demand, and avoiding energy waste and power shortages. Under the objective function of minimizing grid operating costs, the model comprehensively considers the power generation costs of different power sources and the operating costs of energy storage devices, optimizing power generation and energy storage strategies to reduce overall operating costs.

[0042] Power balance constraints are fundamental to the normal operation of a power system. These constraints coordinate power production and consumption in real time, preventing grid frequency fluctuations and voltage instability caused by power imbalances, ensuring the safe and stable operation of the grid, and providing users with a reliable power supply. Power generation capacity constraints and energy storage device constraints prevent damage to power generation equipment due to overload or underload operation, while protecting energy storage devices from overcharging and discharging, extending their lifespan.

[0043] Incorporating renewable energy integration rate into the model's objectives encourages the power system to prioritize the integration and consumption of renewable energy during operation. In actual operation, the model will rationally adjust the power generation capacity and energy storage charging and discharging strategies based on renewable energy generation and load demand, thereby improving the utilization rate of renewable energy, reducing wind and solar curtailment, and promoting the widespread application of renewable energy in the power system.

[0044] By optimizing decision variables, the model can select the lowest-cost power generation and energy storage schemes while meeting electricity demand, thereby reducing the overall operating cost of the power system. The introduction of a load shortage cost coefficient ensures that the model fully considers the degree to which load demand is met during the optimization process. By rationally allocating power generation capacity and energy storage charging and discharging power, the occurrence of load shortages can be minimized, reducing economic losses to users and indirectly improving the economic efficiency of the power system.

[0045] The established power balance model is solved using a genetic algorithm to determine the operating scheme, including the following steps: S1. Encode the decision variables; S2. Randomly generate the initial population; S3. Constructing the fitness function: Normalize and weight the objectives of power supply and demand balance, minimum grid operation cost, and maximum renewable energy consumption; S4. Select superior individuals from the current population to enter the next generation population based on the roulette wheel selection method; S5. Perform a crossover operation on the selected next generation population to obtain the crossover population; S6. Perform mutation operation on the crossover population to obtain the mutated population; S7. Repeat steps S4-S6 until the maximum number of iterations is reached. Stop the calculation and output the current optimal solution. The current optimal solution is the running plan.

[0046] The fitness function is: ; In the formula, Let x be the fitness value of the x-th individual. Let the objective function value be the balance between electricity supply and demand. The objective function value is the value that minimizes the operating cost of the power grid. To maximize the objective function value for renewable energy consumption, for Weighting factors for Weighting factors for Weighting factors.

[0047] In power grids with high penetration of renewable energy, the operating state of the power system is complex and variable, influenced by numerous factors. Genetic algorithms can find the globally optimal or near-globally optimal solution from a large number of possible operating schemes. For example, considering factors such as the generation characteristics of different power sources, the charging and discharging efficiency of energy storage devices, load variation patterns, and the intermittency and volatility of renewable energy sources, they can accurately find the optimal combination of power generation capacity, energy storage device charging and discharging power, and energy quantity that meets multiple objective requirements, ensuring the stable and economical operation of the power system under various complex operating conditions.

[0048] In power grids with high penetration of renewable energy, the operational status of each link—source, grid, load, and storage—changes continuously over time. For example, renewable energy generation is intermittent and fluctuates due to weather conditions, and load demand also fluctuates with seasonal and temporal factors. Genetic algorithms, through continuous iterative optimization during the solution process, can quickly adapt to these changes. In each iteration, a new population is generated through crossover and mutation operations, introducing new gene combinations (corresponding to different operating schemes), enabling the algorithm to adjust its operating scheme according to the real-time state of the system.

[0049] The evaluation of the determined operational plan includes the following steps: After obtaining the grid power after operation, analyze and obtain the power supply and demand balance assessment indicators, and conduct a power supply and demand balance assessment. After the grid is put into operation, the grid stability assessment index is obtained by analysis, and the grid stability assessment is carried out. Calculate the renewable energy consumption index after operation, and evaluate the renewable energy consumption effect based on the renewable energy consumption index; If the power supply and demand balance assessment index, the power grid stability assessment index, and the new energy consumption index are all less than the corresponding thresholds, then the power supply and demand balance assessment index, the power grid stability assessment index, and the new energy consumption index will be comprehensively analyzed to obtain the comprehensive operation assessment index. The overall operational evaluation index is compared with the operational evaluation thresholds stored in the database. If the overall operational evaluation index is less than the operational evaluation threshold, the corresponding operational plan meets the requirements. If the overall operational evaluation index is not less than the operational evaluation threshold, then the corresponding operational plan does not meet the requirements and needs to be readjusted. If any of the power supply and demand balance assessment indicators, power grid stability assessment indicators, and new energy consumption indicators is not less than the corresponding threshold, then the parameters of the scheme corresponding to that indicator will be adjusted.

[0050] The formula for calculating the power supply and demand balance assessment index is as follows: ; In the formula, It serves as an indicator for assessing the balance between power supply and demand.

[0051] The formula for calculating the power grid stability assessment index is as follows: ; In the formula, W is the power grid stability assessment index. Voltage deviation rate, Where U is the frequency deviation rate and U is the actual voltage. Where is the rated voltage and f is the actual frequency. This is the rated frequency.

[0052] The formula for calculating the renewable energy consumption index is as follows: ; In the formula, X represents the renewable energy consumption index.

[0053] The formula for calculating the comprehensive evaluation indicators is as follows: ; In the formula, YX is the fitness value of the x-th individual. for Weighting factors The weighting factor for W, Let X be the weighting factor.

[0054] In power grids with high penetration of renewable energy, the increased volatility of renewable energy generation and the increased uncertainty of load demand make it difficult to maintain a balance between power supply and demand. To achieve this balance, it is necessary to calculate power supply and demand balance assessment indicators. These indicators can promptly reflect the deviation between the two; if the deviation exceeds a threshold, it indicates a power supply and demand imbalance, requiring adjustments to the system. This allows for the timely detection and resolution of power shortages or surpluses, ensuring a stable power supply and meeting users' electricity needs.

[0055] These indicators are monitored and calculated in real time. Once an indicator is found to be close to or exceed the threshold, the operation plan is adjusted to effectively avoid problems such as voltage fluctuations and frequency anomalies, ensure the stable operation of the power grid, and reduce equipment damage and power outages caused by power grid instability.

[0056] In a grid with high penetration of new energy sources, the new energy absorption rate can quantify the actual utilization of new energy. If the absorption index is not ideal, parameters can be adjusted for relevant schemes, such as optimizing the charging and discharging strategies of energy storage equipment or adjusting the power generation plan, in order to increase the absorption of new energy, reduce wind and solar curtailment, and promote the transformation of the energy structure towards a cleaner energy source.

[0057] The comprehensive evaluation indicators consider multiple key factors, reflecting the overall performance of the operational plan. Comparison with operational evaluation thresholds determines whether the plan meets requirements; if not, adjustments are made. This avoids focusing on a single indicator while neglecting other aspects, helping to find better operational plans and achieve multi-directional coordinated optimization of source, grid, load, and storage.

[0058] Example 2: Based on Example 1, when calculating the load forecast correction using the ridge regression model, the fixed regularization coefficient is no longer used. Instead of directly adjusting the regularization coefficient, the model dynamically adjusts the regularization coefficient based on the real-time operating status of the power grid and load characteristics to improve its adaptability to complex operating conditions and the accuracy of its corrections. The dynamic impact index set of the regularization coefficient is defined, including the real-time fluctuation range of renewable energy output, the proportion of load type (the respective proportions of industrial load, residential load, and commercial load), and the historical correction error feedback value. Among them, the real-time fluctuation range of renewable energy output is measured by the ratio of the maximum difference of the total renewable energy power generation in the past hour to the average power, and the historical correction error feedback value is the average of the load forecast correction value and the actual load deviation in the past 5 time windows. Based on historical operating data, a dynamic mapping model of operating conditions and regularization coefficients is constructed. This dynamic mapping model takes the scenarios corresponding to different fluctuation amplitudes, different load types, and different historical error feedback values ​​as inputs, and takes the regularization coefficient that minimizes the load prediction correction error under the scenario as output. It is generated by training through the random forest algorithm and can accurately capture the nonlinear relationship between multiple factors and the optimal regularization coefficient. Real-time data on dynamic impact indicators of the power grid are collected, including the fluctuation range of current renewable energy output, the current proportion of various loads, and the real-time feedback value of historical correction errors. This data is input into the dynamic mapping model of operating conditions and regularization coefficients, which outputs dynamic regularization coefficients adapted to the current scenario. Finally, the dynamic regularization coefficients are substituted into the ridge regression model, combined with the real-time renewable energy absorption rate, total renewable energy generation, and absorption-load coupling coefficient, to calculate the dynamically optimized load forecast correction value.

[0059] By adopting dynamic adjustment, the ridge regression model can more flexibly cope with complex changes when facing scenarios such as drastic fluctuations in new energy output (when the fluctuation range is large, the regularization coefficient is dynamically increased to reduce the model's sensitivity to abnormal data), sudden changes in load type structure (when the proportion of industrial load rises sharply, the regularization coefficient is adjusted to enhance the model's adaptation to changes in high-power load), and large historical correction errors (when the feedback value exceeds the threshold, the regularization coefficient is finely adjusted to quickly correct the model deviation). The dynamic regularization coefficient solves the problems of decreased correction accuracy and insufficient robustness when the fixed regularization coefficient changes operating conditions.

[0060] When solving the established power balance model using a genetic algorithm, the weighted summation process of the fitness function is optimized by introducing a dynamic weight adjustment mechanism to replace the original fixed weight mode. Specifically: When constructing the fitness function, instead of using fixed weighting factors based on historical data mapping, the weighting of the power supply and demand balance target, the minimum power grid operating cost target, and the maximum renewable energy consumption target is dynamically adjusted according to the real-time operating conditions of the power grid. The core assessment dimensions of the real-time operating conditions of the power grid are clearly defined, including the current load factor of the power grid, the fluctuation of the output of new energy power generation, the proportion of the remaining power of energy storage devices, and the deviation rate of the power grid voltage. Among them, the fluctuation of the output of new energy is measured by the change in the total power generation of new energy within adjacent fixed time intervals (5-15 minutes). Based on historical operating data, the mapping relationship between different operating conditions (such as high load rate scenarios, scenarios with drastic fluctuations in renewable energy output, and scenarios with insufficient energy storage capacity) and the corresponding optimal weight combinations is sorted out, and a dynamic mapping model of operating conditions and weights is constructed. This dynamic mapping model can record the target weight configuration that optimizes the overall benefits of the power grid (taking into account stability, economy, and renewable energy consumption) under different scenarios. Real-time data on the actual operating conditions of the power grid is collected and input into the aforementioned dynamic mapping model. The gradient boosting tree algorithm is used to accurately calculate the dynamic weight factors suitable for the current scenario, ensuring that the sum of the dynamic weight factors for the three objectives remains constant at 1. Finally, the three objective function values ​​are normalized and substituted into the dynamic weight factors to complete the fitness function calculation.

[0061] With dynamic adjustment, in complex scenarios such as peak-valley load differences (prioritizing supply and demand balance during peak hours and prioritizing the absorption of new energy during off-peak hours), sudden changes in new energy output (prioritizing grid stabilization during periods of severe fluctuation), and changes in the status of energy storage devices (controlling operating costs when remaining power is insufficient), dynamic weights can automatically adjust the priority of each objective, solving the problem of poor coordination effect of fixed weights under multiple operating conditions. At the same time, the application of gradient boosting tree algorithm can more accurately capture the nonlinear relationship between operating conditions and optimal weights, improving the accuracy of the fitness function in guiding the solution direction of the genetic algorithm.

[0062] Example 3: Based on Example 1, after calculating the load forecast correction using the ridge regression model, an adaptive calibration of the absorption-load coupling coefficient k is added: The actual renewable energy absorption rate and load forecast error of the past 24 hours are paired hourly to form an error-absorption sequence. If the extreme combination of "absorption rate higher than 95% and load forecast error greater than 3%" or "absorption rate lower than 70% and load forecast error less than 1%" occurs for 3 consecutive hours, the online update of k is triggered: the ridge regression model is refitted with the data of the most recent 72 hours, and the original k value is replaced with the new k value before entering the subsequent power balance model.

[0063] This mechanism can keep the error of load forecast correction value less than 2% in both high and low absorption scenarios, avoiding overcompensation or undercompensation caused by fixed coefficients.

[0064] Example 4: Figure 2 As shown, a multi-directional collaborative system for energy sources, grids, loads, and storage in a high-penetration renewable energy grid is used to implement the methods in Embodiment 1, Embodiment 2, or Embodiment 3, comprising: The data acquisition module is used to acquire historical datasets of the power system, socio-economic data, solar radiation, wind energy resource distribution data, and historical meteorological data, and to transmit and integrate the acquired data. The data communication transmission module is used to receive data acquired by the data acquisition module. The data communication transmission module is connected to the load forecasting module, the power supply forecasting module, the load forecasting correction module, the power balance model construction module, the operation scheme determination module, the scheme evaluation module, and the data storage module. The load forecasting module is used to acquire historical power system data and socio-economic data within the region to obtain load forecasting results. The historical power system data includes historical load data, power generation, voltage, current, historical meteorological data, and historical electricity consumption data. The socio-economic data includes total population, population density, and GDP. The power prediction module is used to acquire data on solar radiation, wind energy resource distribution, and historical meteorological data within the region to obtain power prediction results, including predicted solar power generation, predicted solar power output, predicted wind power generation, and predicted wind power output. The load forecasting correction module is used to input the power forecasting results from the power forecasting module into the load forecasting module to obtain the load forecasting correction value. The power balance model building module is used to build a power balance model based on load forecast correction values ​​and power source forecast results. The operation scheme determination module is used to solve the established power balance model based on the genetic algorithm and determine the operation scheme. The scheme evaluation module is used to evaluate the determined operation scheme; The data storage module is used to store the collected data.

Claims

1. A multi-directional collaborative method for energy source, grid, load, and storage in a high-penetration renewable energy power grid, characterized in that, Includes the following steps: The system acquires historical power system data and socioeconomic data within the region, constructs a prediction model to obtain load forecast results, and the historical power system data includes historical load data, power generation, voltage, current, historical meteorological data, and historical electricity consumption data. The socioeconomic data includes total population, population density, and GDP. Data on solar radiation, wind energy resource distribution, and historical meteorological data within the region are acquired to obtain power generation forecasts, including predicted solar power output, predicted solar power generation, predicted wind power output, and predicted wind power generation. Based on the power supply forecast results, the load forecast correction value is obtained; A power balance model is constructed based on the load forecast correction value and the power source forecast result; The established power balance model is solved using a genetic algorithm to determine the operating scheme. The determined operation plan is evaluated. If the operation plan meets the requirements, the operation plan is output for multi-directional coordination of source, grid, load and storage.

2. The method for multi-directional coordination of energy source, grid, load, and storage in a high-penetration power grid according to claim 1, characterized in that, The process of constructing a forecasting model to obtain load forecasting results includes the following steps: Obtain load-related parameters, combine them with historical load data, construct the first prediction model, and obtain the first load prediction value; Based on historical load data and load-related parameters, a second prediction model is constructed to obtain the second load prediction value; The combined load forecast value is obtained by combining the first load forecast value and the second load forecast value. The formula for calculating the comprehensive load forecast is as follows: ； In the formula, This is the comprehensive load forecast value. This is the first load forecast value. This is the second load forecast value. for Weighting factors for Weighting factors; Obtaining load-related parameters includes the following steps: The load-related parameters include multiple related parameters. Pearson correlation analysis is used to analyze the correlation coefficients between each related parameter and the load power. The correlation coefficients obtained from the analysis are compared with the correlation thresholds stored in the database. If the correlation coefficient is greater than the correlation threshold, the relevant parameter corresponding to the correlation coefficient is selected and recorded as the key parameter. If the correlation coefficient is not greater than the correlation threshold, the relevant parameter corresponding to that correlation coefficient will not be selected. The obtained key parameters are recorded as relevant key data, including temperature, temperature change rate, humidity, and solar radiation intensity.

3. The method for multi-directional coordination of energy source, grid, load, and storage in a high-penetration new energy power grid according to claim 2, characterized in that, The first prediction model is constructed by the following steps: The relevant key data is imported into the neural network model to obtain the first prediction model; The least squares method is used to estimate the model parameters, calculate the coefficient of determination, and perform model testing. If the test statistic is greater than the critical value of the test statistic in the database, then the first prediction model is considered to be significant overall. Obtain relevant key data for the future, including predicted temperature, predicted rate of temperature change, predicted humidity, and predicted solar radiation intensity, import them into the first prediction model, and obtain the first load prediction. The formula for calculating the first load forecast value is: ； In the formula, T is the temperature. For reference temperature, For the rate of temperature change, Let S be the reference temperature change rate, and S be the solar radiation intensity. For reference to solar radiation intensity, H represents humidity. For reference humidity, The intercept is... for The slope, for The slope, for The slope, for The slope.

4. The method for multi-directional coordination of energy source, grid, load, and storage in a high-penetration new energy power grid according to claim 2, characterized in that, Constructing a second prediction model includes the following steps: The relevant key data is imported into the neural network model to obtain the second prediction model; In the second prediction model: the input layer nodes include historical load data, temperature data, temperature change rate data, solar radiation intensity data, and humidity data. The first hidden layer has 10-15 neurons, the second hidden layer has 5-8 neurons, and the ReLU activation function is used. The output layer nodes are the second load prediction values. Based on historical load data and load-related parameters, a training set and a validation set are divided. The second prediction model is trained using the training set, and its performance is evaluated on the validation set. If the loss function value on the validation set does not decrease for several consecutive rounds, the second prediction model is considered to have converged, and training is stopped. The relevant key data for the future is imported into the second forecast model to obtain the second load forecast value.

5. The method for multi-directional coordination of energy source, grid, load, and storage in a high-penetration power grid according to claim 1, characterized in that, Obtaining power prediction results includes the following steps: Based on the power generation principle of photovoltaic cells, combined with solar radiation intensity data and temperature data, the output power of photovoltaic cells is obtained, and the solar power generation is analyzed. By inputting solar radiation intensity data and temperature data into a neural network model, the predicted solar power generation is obtained, and the predicted solar power generation is analyzed. Wind energy resource distribution data includes wind speed, wind turbine angle of attack, and wind energy density. The wind energy resource distribution data and wind power generation data are recorded as wind power generation time series data and imported into an autoregressive integral moving average model to obtain the predicted wind power generation. The predicted wind power generation is then analyzed. The formula for calculating the output power of a photovoltaic cell is: ； In the formula, The output power of the photovoltaic cell. The reference output power is stored in the database. G represents the reference solar radiation intensity stored in the database, and G represents the actual solar radiation intensity. The reference operating temperature is stored in the database. The power temperature coefficient is expressed in units of 1 / ℃. This refers to the actual operating temperature. The formula for calculating solar power generation is: ； In the formula, for to Solar power generation during the time period for to The output power of the photovoltaic cell at time t within the time period. At the starting time, The end time; The expression for the autoregressive integral moving average model is: ； In the formula, Here, B is the autoregressive coefficient, B is the lag operator, and d is the difference order. Let be the wind power generation capacity at time t. The moving average coefficient is... It is a white noise sequence. for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right Influence coefficient, for right The influence coefficient, where jq is the order of the moving average; The formula for calculating wind power generation is: ； In the formula, for to Wind power generation during the time period for to The wind power generation capacity at time t within the time period.

6. The method for multi-directional coordination of energy source, grid, load, and storage in a high-penetration new energy power grid according to claim 1, characterized in that, Obtaining the load forecast correction includes the following steps: Obtain the renewable energy power generation and the power generation that has been absorbed from the database, and calculate the renewable energy absorption rate. The formula for calculating the renewable energy absorption rate is: ； In the formula, Let be the renewable energy consumption rate at time t. Let be the power generation capacity of the h-th renewable energy source at time t. Let H be the renewable energy power generation of the h-th renewable energy source at time t, where h is the renewable energy source number and H is the total number of renewable energy sources. The ridge regression model is trained by obtaining the historical dataset stored in the database, and the load forecast correction value is calculated. The historical dataset includes historical renewable energy absorption rate data, historical renewable energy power generation data, and historical load forecast correction value data. The ridge regression model formula is: ； In the formula, This is the load forecast correction value at time t. Let be the renewable energy consumption rate at time t. Let be the total renewable energy generation capacity at time t, and k be the consumption-load coupling coefficient. is the regularization coefficient, M is the total number of sliding windows, and m is the time window number.

7. The method for multi-directional coordination of energy source, grid, load, and storage in a high-penetration power grid according to claim 1, characterized in that, Constructing a power balance model includes the following steps: Based on load forecasting results and power supply forecasting results, the decision variables in the power balance model are determined. The decision variables are power generation capacity, energy storage device charging and discharging power, and energy storage device power. Define the model objectives, which include the objectives of balancing electricity supply and demand, minimizing grid operating costs, and maximizing the absorption of new energy sources. Establish constraints, including power balance constraints, power generation constraints, and energy storage device constraints. By integrating decision variables, model objectives, and constraints, a power balance model is established. The objective function for balancing electricity supply and demand is: ； In the formula, Let a be the power generation capacity of the a-th power source. The charging and discharging power of the b-th energy storage device. Let be the power consumption of the g-th load, where 'a' is the power source number, 'b' is the energy storage device number, and 'g' is the load number. This is the load forecast correction value for the g-th load. The objective function for minimizing power grid operating costs is: ； In the formula, C represents the total cost of power grid operation. Let be the power generation cost coefficient of the a-th power source. Let b be the operating cost coefficient for the b-th energy storage device. Let g be the power shortage cost coefficient for the g-th load. The formula for maximizing the utilization of new energy sources is: ； In the formula, To maximize the utilization rate of new energy sources, Let h be the power generation capacity of the h-th renewable energy source that is absorbed. The renewable energy power generation capacity of the h-th renewable energy source is... This is the correction term for the h-th new energy power source. Let h be the node allocation coefficient for the h-th renewable energy source. This is the load forecast correction value for the h-th renewable energy source. Power generation capacity constraints, the specific constraints are as follows: The power generation capacity of the power source is within its power generation capacity configuration range, that is: ； in, Let be the minimum generating power of the a-th power source. The maximum power output of the a-th power source is... Let g be the power consumption of the g-th load. This is the total load correction for the power supply zone where power source a is located; Constraints on energy storage devices, specifically: The energy storage device's capacity is within its configured capacity range, i.e.: ； in, Let b be the minimum power of the energy storage device. Let b be the maximum power of the energy storage device. This represents the actual power output of the b-th energy storage device.

8. A multi-directional collaborative method for energy source, grid, load, and storage in a high-penetration power grid according to claim 7, characterized in that, The established power balance model is solved using a genetic algorithm to determine the operating scheme, including the following steps: S1. Encode the decision variables; S2. Randomly generate the initial population; S3. Constructing the fitness function: Normalize and weight the objectives of power supply and demand balance, minimum grid operation cost, and maximum renewable energy consumption; S4. Select superior individuals from the current population to enter the next generation population based on the roulette wheel selection method; S5. Perform a crossover operation on the selected next generation population to obtain the crossover population; S6. Perform mutation operation on the crossover population to obtain the mutated population; S7. Repeat steps S4-S6 until the maximum number of iterations is reached. Stop the calculation and output the current optimal solution. The current optimal solution is the running plan.

9. A multi-directional collaborative method for energy source, grid, load, and storage in a high-penetration power grid according to claim 1, characterized in that: The evaluation of the determined operational plan includes the following steps: After obtaining the grid power after operation, analyze and obtain the power supply and demand balance assessment indicators, and conduct a power supply and demand balance assessment. After the grid is put into operation, the grid stability assessment index is obtained by analysis, and the grid stability assessment is carried out. Calculate the renewable energy consumption index after operation, and evaluate the renewable energy consumption effect based on the renewable energy consumption index; If the power supply and demand balance assessment index, the power grid stability assessment index, and the new energy consumption index are all less than the corresponding thresholds, then the power supply and demand balance assessment index, the power grid stability assessment index, and the new energy consumption index will be comprehensively analyzed to obtain the comprehensive operation assessment index. The overall operational evaluation index is compared with the operational evaluation thresholds stored in the database. If the overall operational evaluation index is less than the operational evaluation threshold, the corresponding operational plan meets the requirements. If the overall operational evaluation index is not less than the operational evaluation threshold, then the corresponding operational plan does not meet the requirements and needs to be readjusted. If any of the power supply and demand balance assessment indicators, power grid stability assessment indicators, and new energy consumption indicators is not less than the corresponding threshold, then the parameters of the scheme corresponding to that indicator will be adjusted.

10. A multi-directional collaborative system for energy source, grid, load, and storage in a high-penetration power grid, used to implement the method described in any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire historical datasets of the power system, socio-economic data, solar radiation, wind energy resource distribution data, and historical meteorological data, and to transmit and integrate the acquired data. The data communication and transmission module is used to receive data acquired by the data acquisition module. The data communication and transmission module is connected to the load forecasting module, the power supply forecasting module, the load forecasting correction module, the power balance model construction module, the operation scheme determination module, the scheme evaluation module, and the data storage module. The load forecasting module is used to acquire historical power system data and socio-economic data within the region to obtain load forecasting results. The historical power system data includes historical load data, power generation, voltage, current, historical meteorological data, and historical electricity consumption data. The socio-economic data includes total population, population density, and GDP. The power prediction module is used to acquire data on solar radiation, wind energy resource distribution, and historical meteorological data within the region to obtain power prediction results, including predicted solar power generation, predicted solar power output, predicted wind power generation, and predicted wind power output. The load forecasting correction module is used to input the power forecasting results from the power forecasting module into the load forecasting module to obtain the load forecasting correction value. The power balance model building module is used to build a power balance model based on load forecast correction values ​​and power source forecast results. The operation scheme determination module is used to solve the established power balance model based on the genetic algorithm and determine the operation scheme. The scheme evaluation module is used to evaluate the determined operation scheme; The data storage module is used to store the collected data.

Images