Method and device for filling and quantifying multi-source data of power system based on bimodal hybrid algorithm in extreme weather

Abstract

The invention discloses a method and a device for filling and quantifying multi-source data of a power system based on a bimodal hybrid algorithm in extreme weather, which belong to the technical field of power distribution networks and are based on a reverse chaos whale transfer algorithm and combined with meteorological characteristics to dynamically adjust and construct a load prediction model to fill missing data and output distribution parameters and weight coefficients. Meanwhile, a meteorological condition matrix is introduced, a Gaussian mixture model is constructed, and the uncertainty of data is quantized and filled through a Monte Carlo subset deduction algorithm. The method effectively solves the problems that a traditional method is low in processing efficiency and large in error in extreme weather, experiments show that the method is high in convergence speed and high in estimation precision, the accuracy and reliability of power system data are remarkably improved, and guarantee is provided for stable operation of a power dispatching system.

Description

Technical Field

[0001] The present invention belongs to the technical field of power distribution networks, and in particular relates to a method and device for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions. Background Art

[0002] With the rapid development of smart grid technology, the sources of power big data are becoming increasingly rich, showing the characteristics of large data volume, various types, low value density and fast processing speed. However, these data originate from different systems and follow a unified modeling standard, resulting in a serious problem of missing data from multiple sources, especially in extreme weather scenarios (such as typhoons and blizzards). The missing data problem in distribution networks presents multi-dimensional complexity. For example, sudden regional missing data (such as data discontinuities of tens of kilometers caused by typhoons in coastal wind farms) causes the moving average interpolation error that relies on time continuity to surge by more than 300%; non-Gaussian distribution characteristics (such as multimodal load fluctuations caused by lightning strikes) cause the clustering accuracy of the EM algorithm based on the Gaussian assumption to drop sharply to 58%; strong noise interference (such as outliers with wind speeds > 80m / s in hurricanes) causes methods such as missForest (predicting and filling missing data based on a random forest iterative model) to generate out-of-limit filling values due to the lack of physical constraints; and the breakdown of spatiotemporal correlation (such as regional load differentiation caused by blizzards) causes the geographic proximity assumption of the kNN algorithm (data filling based on similarity) to fail, and the similarity measurement error increases by 2.4 times, highlighting the systematic limitations of traditional methods in complex environments. The above problems not only affect the integrity and stability of the data, but also pose obstacles to future analysis and application of critical data. Therefore, it is crucial to develop methods that can estimate parameters in incomplete datasets and fill in missing data in extreme environments.

[0003] Under extreme weather conditions, traditional distribution network data processing has problems of missing data and low efficiency. Due to the failure to fully consider the objectivity and complexity of the data gap filling process, the data interpolation process may be time-consuming and have large errors, which cannot fully meet the needs of efficient data gap filling in the power system. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0005] A dual-mode hybrid algorithm for filling and quantifying multi-source data in power systems under extreme weather conditions, including:

[0006] Step 1, parameter estimation phase, includes:

[0007] Step 1.1: Using extreme weather parameters, we select the missing data areas in the power dispatching system, introduce extreme weather influencing factors, perform preprocessing on the power system data, remove abnormal and duplicate data, and then standardize the missing data intervals.

[0008] Step 1.2, construct a weather-load data prediction model;

[0009] Step 1.3: Use the reverse chaotic whale transfer algorithm to obtain the optimal solution for missing data. The reverse chaotic whale transfer algorithm includes three steps: chaotic mapping with meteorological characteristics, reverse propagation, and whale group collaborative strategy to jump to the optimal solution;

[0010] Step 2, the uncertainty quantification stage, includes:

[0011] Step 2.1: Based on the parameter estimation results, construct a complex probability distribution model for missing data;

[0012] Step 2.2, outputting distribution parameters through the reverse chaotic whale transfer algorithm, including training using a model with an adaptive strategy;

[0013] Step 2.3, generate a large number of independent samples in the probability space to capture the mean and variance of the filling results;

[0014] Step 2.4: Based on the statistical characteristics of the imputed values, calculate the confidence interval and probability density function of the imputed values to quantify the random effect of missing data.

[0015] Step 3, data completion and result output, combines the optimal solution for missing data obtained in step 1.3 with the probability density function obtained in step 2.4 to form a closed-loop complementary technology system, generates the final filling value, and inputs the filled data set into the power system.

[0016] A device for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions, comprising:

[0017] Parameter estimation module, including:

[0018] Through extreme weather parameters, we select the missing data areas in the power dispatching system, introduce extreme weather influencing factors, perform preprocessing operations on the power system data, remove abnormal and duplicate data, and then standardize the missing data intervals;

[0019] Build a weather-load data forecasting model;

[0020] The optimal solution for missing data is obtained using the reverse chaotic whale transfer algorithm, which includes three steps: chaotic mapping with meteorological characteristics, reverse propagation, and whale group collaborative strategy to jump to the optimal solution.

[0021] Uncertainty quantification module, including:

[0022] Based on the parameter estimation results, a complex probability distribution model of missing data is constructed;

[0023] Outputting distribution parameters through the reverse chaotic whale transfer algorithm, including training with a model with an adaptive strategy;

[0024] Generate a large number of independent samples in the probability space to capture the mean and variance of the filling results;

[0025] According to the statistical characteristics of the imputation results, the confidence interval and probability density function of the imputation value are calculated to quantify the random impact of missing data;

[0026] The data completion and result output module combines the optimal solution of missing data obtained by the parameter estimation module with the probability density function obtained by the uncertainty quantification module to form a closed-loop complementary technology system, generate the final filling value, and input the filled data set into the power system.

[0027] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the processor implements the steps of the method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions.

[0028] A non-transitory computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions.

[0029] The present invention has the following beneficial effects:

[0030] (1) The present invention accelerates convergence and performs chaotic interference operations by using the reverse chaotic whale transfer algorithm. It takes only 15 iterations for the objective function value to reach the optimal solution. In dealing with the problem of missing multi-source heterogeneous data under extreme weather conditions, the reverse chaotic whale transfer algorithm achieves the goal of quickly finding the optimal solution in the global scope and minimizing the error of the optimal solution. In the process of filling missing data under extreme weather conditions, compared with the traditional whale algorithm, the convergence speed of the reverse chaotic whale transfer algorithm based on the present invention is improved. In addition, the global optimization success rate of measured data in typhoon scenarios is greatly improved, achieving global optimization requirements.

[0031] (2) In the modeling of data distribution, the present invention improves the accuracy of non-Gaussian data clustering by adopting a Gaussian mixture model weighted by meteorological characteristics, making the effect of obtaining the deficiency compensation value more accurate, and supports multi-modal load fluctuation modeling, so that the model can simultaneously characterize opposing modes such as "sudden drop" and "surge", providing key technical support for improving the resilience of power systems under extreme weather conditions.

[0032] (3) The present invention uses the Monte Carlo subset deduction algorithm to make the result of obtaining the shortfall compensation value more accurate, and supports multi-modal load fluctuation modeling, so that the model can simultaneously characterize opposing modes such as "sudden drop" and "surge", providing key technical support for improving the resilience of the power system under extreme weather conditions.

[0033] (4) The adaptive penalty mechanism driven by geographic vulnerability proposed in this invention solves the problem that traditional static models cannot adapt to dynamic extreme weather disasters by dynamically adjusting the spatiotemporal correlation weights by integrating real-time meteorological data (such as wind speed and snow thickness). This mechanism introduces geographic vulnerability coefficients (such as weighted calculations of altitude, vegetation density, and equipment age), quantifies physical environmental differences as mathematical constraints, and significantly suppresses the risk of false correlation in high / low vulnerable areas. At the same time, combined with physical law constraints such as power conservation and equipment capacity limitations, it ensures that the fill value conforms to the spatiotemporal distribution law of disasters (such as the movement path of blizzards), thereby improving data rationality. Compared with traditional methods, this technology achieves significant optimization in spatiotemporal correlation accuracy, false correlation suppression, and real-time response efficiency.

[0034] (5) The dual-modal hybrid algorithm proposed in this invention is embodied in the organic combination of deterministic data filling and uncertainty quantification to form a closed-loop complementary technology system. The beneficial effects are reflected in the following aspects: in terms of coupling mechanism innovation, the reverse chaotic whale transfer algorithm (deterministic mode) is used to achieve accurate filling of missing values. At the same time, a dynamic probability distribution is constructed based on the Gaussian mixture model of the meteorological condition matrix. The Monte Carlo subset deduction algorithm is used to continuously inject uncertainty parameters during the filling process. The two modes dynamically interact through weight coefficients. When the typhoon wind speed is greater than 30m / s, the weight of the uncertainty quantification module is automatically increased to 0.85, which is significantly different from the serial mode of separation of filling and evaluation in traditional methods. This dual-modal architecture breaks through the technical limitations of the traditional method of "focusing on filling and neglecting evaluation". Probability distribution parameters are generated simultaneously during the filling process, providing a dual decision basis of "point estimation + interval estimation" for power grid dispatching. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0036] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0037] like Figure 1As shown, the specific implementation process of the power system multi-source data filling and quantification method of the dual-mode hybrid algorithm under extreme weather conditions of the present invention includes three steps: parameter estimation stage, uncertainty quantification stage and data completion and result output stage.

[0038] Step 1: Parameter estimation phase. Specifically includes:

[0039] In step 1.1, we select the missing data areas in the power dispatching system based on extreme weather parameters such as typhoon wind speed and snow cover area, introduce extreme weather influencing factors, perform preprocessing operations on the power system data, remove abnormal and duplicate data, and then standardize the missing data interval to improve the convergence speed of the algorithm.

[0040] Step 1.2, construct a meteorological-load data prediction model. The meteorological-load data prediction model consists of two parts: objective function and constraint conditions. The objective function is to minimize the prediction error of the model to pursue the accuracy of data prediction. The constraint conditions are defined by the physical laws and probability distribution in the power system to define the rationality boundary of the prediction data.

[0041] Step 1.2.1, objective function (Data completion loss function) design:

[0042] For the distribution parameters to be estimated , is the objective function, which represents the minimization of the prediction error of the weather-load data prediction model, that is, the error between the predicted value and the actual observed value, which can be expressed as:

[0043] ;

[0044] in, is the total number of samples, that is, the total number of data points in the power system dataset, For the samples, For the The actual observed value of the sample, is the weather-load data prediction model, are input features, and are related variables used to predict specific output values. is the distribution parameter to be estimated, and the vector of integrated multidimensional meteorological characteristics ,in, express Maximum wind speed at the typhoon center at the moment (m / s), express Snow thickness at any moment (cm), express The diameter of hail at the moment (mm), etc., define different meteorological characteristics according to the different environments. It only illustrates the three meteorological characteristics of wind, snow and hail. It can also represent other meteorological characteristics such as temperature and rainfall, which can be selected based on the actual situation; is the adaptive penalty coefficient, and the calculation formula is , is the total number of weight parameters, that is, the number of all trainable parameters in the model, For the weight parameters, For the The meteorological characteristic values of the weight parameters, For the The meteorological warning threshold of the weight parameter, For the The range of variation of the weight parameters.

[0045] Among them, the weather-load data prediction model The formula is as follows:

[0046] ;

[0047] in, is the intercept term or bias term of the model, which means that when all input feature values The predicted value when both are 0, For the parameters, is the total number of input features, For the The coefficient of an input feature indicates the degree of influence of the feature on the predicted output. For the The first sample Input feature values, is the total number of weight parameters, For the weight parameters, For the The weather influence weight matrix with learnable weight parameters, For the The meteorological characteristic conversion function with a weight parameter (such as the exponential decay function for typhoon wind speed) ,in Indicates wind speed in meters per second). For the In the iteration The weight parameter value, is the geographic vulnerability coefficient, calculated as ,in is the altitude coefficient, is the vegetation density, is the age of the device; For geographical features, Expressed as 、 and Their respective values, such as .

[0048] Step 1.2.2, constraint design:

[0049] Physical constraints: power conservation and equipment capacity limitations;

[0050] Probability distribution constraints: missing values must obey the preset distribution ,in is the mean (expressed as the expected value of missing data), is the covariance (expressed as the uncertainty range for missing data).

[0051] Step 1.3, use the reverse chaotic whale transfer algorithm to obtain the optimal solution for missing data. The reverse chaotic whale transfer algorithm includes three steps: chaotic mapping with meteorological characteristics, reverse propagation, and whale group collaborative strategy to jump to the optimal solution.

[0052] In step 1.3.1, the population of the whale optimization algorithm is initialized using the chaotic map embedded with meteorological features.

[0053] Each parameter to be evaluated is taken as an example of a Logistic map (a chaotic map that can map a value to another value between 0 and 1):

[0054] ;

[0055] in, is the control parameter, is the current iteration value, It is the next iteration value, For the Number of iterations.

[0056] In typhoon situations, ;

[0057] in, is the wind speed in meters per second.

[0058] Mapping is extended to multi-dimensional parameter space: for each parameter to be estimated (such as the distribution parameter and weight coefficient of missing values), an independent chaotic sequence is generated through chaotic mapping, and the linear transformation is used to transform the chaotic sequence into a series of chaotic sequences. (in, is the scaling factor of the linear transformation, is the actual value after linear transformation, The chaotic sequence is generated by the chaotic map, and these values are in the interval [0,1] and exhibit chaotic characteristics) The chaotic sequence is mapped to the solution space to generate a chaotic initialized population ,in, is the d-th dimension parameter of the i-th individual, and are the upper and lower bounds of the parameter, A random number used to adjust value, so that it and Mapping chaotic sequences to solution space is a technique for enhancing global search capabilities in optimization algorithms. It increases the diversity of the population by utilizing the randomness and unpredictability of chaotic sequences, prevents the algorithm from falling into local optimal solutions, and improves the efficiency of the algorithm.

[0059] Step 1.3.2: In order to further enhance the diversity of the population, a reverse propagation strategy is used: for the chaotic initialized population, calculate its reverse population , merge the chaotically initialized population and its reverse population, and select individuals with better fitness to form a new initial population.

[0060] Step 1.3.3: After obtaining the new initial population, we further use the whale group collaboration strategy to jump out of the local optimal solution and output the distribution parameters and weight coefficients of the missing values. The specific implementation steps are as follows:

[0061] Step 1.3.3.1, Encirclement Phase: The pilot whale (optimal solution) guides the population to gather in the global optimal area, and other whales adjust their positions based on the pilot whale's position:

[0062] ;

[0063] in, is the coefficient that decays with the number of iterations, For the The position of each individual in the search space, is the location of the optimal solution, that is, the location of the "pilot whale".

[0064] Step 1.3.3.2, pursuit phase: When pilot whales approach their prey (local optimum), they need to switch to random walk mode to explore new areas:

[0065] ;

[0066] in, is a uniformly distributed random number.

[0067] Step 1.3.3.3, diversity maintenance: retain a certain proportion of suboptimal solutions in each position iteration, dynamically adjust the intensity of chaotic perturbation, and balance global search and local mining.

[0068] In step 1.3.3.4, the whale group collaborative strategy effectively escapes the local optimal solution and ultimately outputs the optimal distribution parameters and weight coefficients of the missing values. This approach not only improves the accuracy of data recovery but also enhances the model's global optimization capabilities.

[0069] Step 2, uncertainty quantification stage. Specifically includes:

[0070] Step 2.1: Based on the parameter estimation results (optimal distribution parameters and weight coefficients of missing values), a complex probability distribution model for missing data is constructed. The complex probability distribution model for missing data uses a Gaussian mixture model and introduces a meteorological condition matrix, which is mathematically defined as:

[0071] ;

[0072] in, For the meteorological conditions matrix Under the condition of missing data The probability distribution of For missing data, The meteorological condition matrix includes dynamic meteorological parameters (such as wind speed, snow depth, lightning density, etc.) and geographical vulnerability factors (such as altitude factor, equipment age and vegetation density); For the Mixture components represent sub-distributions under different meteorological-geographical coupling scenarios, for example: Corresponding to the typhoon dominant mode, It is a lightning-heavy rainfall coupled model, etc. is the total number of mixed ingredients; is the mixing ratio, indicating the The distribution of seeds under current meteorological conditions The weight under is a normal distribution, For the The mean and variance of the class. For the The mean of the sub-distribution represents the optimal filling value under a specific meteorological-geographical coupling scenario; For the The variance of the subset distribution is used to quantify the uncertainty of the fill value.

[0073] Step 2.2: Output the distribution parameters through the reverse chaotic whale transfer algorithm, including: mixing ratio Initialization, mean and variance Initialization, as the statistical value of the data sample; use a model with an adaptive strategy (EM algorithm) for training to reduce the computational dimension.

[0074] In the E step (expected step), the posterior probability of each data point belonging to each latent variable is calculated. The formula is as follows:

[0075] ;

[0076] in, The data point belongs to The probability of a mixture component; It is the loop index variable used to traverse all the mixed components; For meteorological conditions Next Initial probability weight under disaster-like mode, For the The expected value of data filling under disaster-like mode, For the Uncertainty in data filling under disaster-like models.

[0077] M step (maximization step), update parameters To maximize the likelihood function:

[0078] ;

[0079] in, is the log-likelihood function, For the data set data points, For the A mixed ingredient, the upper bound chosen for the mixture components (the total number of mixture components), is the mixing ratio, is the probability density of the normal distribution, expressed as data points In the The probability density of the mixture components.

[0080] Until the log-likelihood change is less than the threshold Or when the maximum number of iterations is reached, the iteration is stopped to obtain the trained complex probability distribution model, and the silhouette coefficient is used to verify the rationality of the model:

[0081] ;

[0082] in, is the silhouette coefficient (the closer the silhouette coefficient is to 1, the better the clustering effect of the model), is the total number of data points in the dataset, For the data set data points, For data points To the cluster other than your own The minimum square distance between cluster centers (means), For data points The maximum square distance to the cluster center (mean).

[0083] Step 2.3, generate a large number of independent samples in the probability space and capture the mean of the filling results and variance , as follows:

[0084] ;

[0085] ;

[0086] in, is the total number of independent samples generated, For the independent samples, For the The imputation results of independent samples.

[0087] In step 2.4, based on the statistical characteristics of the imputation results, the confidence interval and probability density function of the imputed value are calculated to quantify the random impact of missing data.

[0088] The formula for calculating the confidence interval of the imputed value is as follows:

[0089] ;

[0090] in, To fill the results The confidence interval of , It is usually a small positive number, such as 0.05, indicating that there is a 95% confidence that the true parameter value falls within this interval. To fill the result, To fill the mean of the results, For the standard normal distribution (1- ) quantiles (e.g. =0.05 corresponds to =1.96), The sample standard deviation of the filling result is used to measure the dispersion of the filling value.

[0091] The formula for calculating the probability density function is as follows:

[0092] ;

[0093] in, For the point The estimated density function at is a data point in the dataset, ranging from 1 to , is the total number of data points in the dataset, K is the kernel function (such as Gaussian kernel), which is used to calculate the contribution of each data point to the estimated probability density, is the bandwidth. The choice of bandwidth affects the smoothness of the kernel density estimate (which can be optimized by maximum likelihood estimation). is the point whose probability density needs to be estimated, For the data set The value of the data point (i.e. is the input feature).

[0094] The statistical characteristics calculated above are passed to the subsequent power dispatch parameter processing module to quantify the random impact of missing data on the final result.

[0095] Step 3: Data completion and result output stage. Specifically includes:

[0096] The optimal solution (data deterministic imputation) obtained from the reverse chaotic whale transfer in step 1.3.3.4, namely the optimal distribution parameters and weight coefficients of missing values (where the distribution parameters include the mean, which is the determined missing data), are combined with the probability distribution (data uncertainty range) obtained from the Monte Carlo subset deduction in step 2.4, namely the probability density function obtained in step 2.4, to form a closed-loop complementary technical system to generate the final imputed value. The imputed data set is then input into the power system to support its normal operation and optimized scheduling. This dual-modal architecture overcomes the technical limitations of traditional methods that "emphasize imputation over evaluation" by simultaneously generating probability distribution parameters during the imputation process, providing a dual decision-making basis for power grid scheduling: "point estimation + interval estimation."

[0097] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk drives, CD-ROMs, optical storage devices, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0098] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0099] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0101] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0102] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for filling and quantifying multi-source data of power systems using a dual-mode hybrid algorithm under extreme weather conditions, characterized in that: include: Step 1, parameter estimation phase, includes: Step 1.1: Using extreme weather parameters, we select the missing data areas in the power dispatching system, introduce extreme weather influencing factors, perform preprocessing on the power system data, remove abnormal and duplicate data, and then standardize the missing data intervals. Step 1.2, construct a weather-load data prediction model; Step 1.3: Use the reverse chaotic whale transfer algorithm to obtain the optimal solution for missing data. The reverse chaotic whale transfer algorithm includes three steps: chaotic mapping with meteorological characteristics, reverse propagation, and whale group collaborative strategy to jump to the optimal solution; Step 2, the uncertainty quantification stage, includes: Step 2.1: Based on the parameter estimation results, construct a complex probability distribution model for missing data; Step 2.2, outputting distribution parameters through the reverse chaotic whale transfer algorithm, including training using a model with an adaptive strategy; Step 2.3, generate a large number of independent samples in the probability space to capture the mean and variance of the filling results; Step 2.4: Based on the statistical characteristics of the imputed values, calculate the confidence interval and probability density function of the imputed values to quantify the random effect of missing data. Step 3, data completion and result output, combines the optimal solution for missing data obtained in step 1.3 with the probability density function obtained in step 2.4 to form a closed-loop complementary technology system, generates the final filling value, and inputs the filled data set into the power system.

2. The method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions according to claim 1 is characterized in that: Step 1.2 includes the objective function (Data completion loss function) design: For the distribution parameters to be estimated , The objective function is expressed as: ； in, is the total number of samples, that is, the total number of data points in the power system dataset, For the samples, For the The actual observed value of the sample, is the weather-load data prediction model, are input features, and are related variables used to predict specific output values. is the distribution parameter to be estimated, and the vector of integrated multidimensional meteorological characteristics ,in, express Maximum wind speed at the typhoon center at the moment (m / s), express Snow thickness at any moment (cm), express The diameter of hail at the moment (mm), etc., define different meteorological characteristics according to the different environments. It only illustrates the three meteorological characteristics of wind, snow and hail. It can also represent other meteorological characteristics such as temperature and rainfall, which can be selected based on the actual situation; is the adaptive penalty coefficient, and the calculation formula is , is the total number of weight parameters, that is, the number of all trainable parameters in the model, For the weight parameters, For the The meteorological characteristic values of the weight parameters, For the The meteorological warning threshold of the weight parameter, For the The range of variation of the weight parameters; Among them, the weather-load data prediction model The formula is as follows: ； in, is the intercept term or bias term of the model, which means that when all input feature values The predicted value when both are 0, For the parameters, is the total number of input features, For the The coefficient of an input feature indicates the degree of influence of the feature on the predicted output. For the The first sample Input feature values, is the total number of weight parameters, For the weight parameters, For the The weather influence weight matrix with learnable weight parameters, For the The meteorological characteristic conversion function with weight parameters; For the In the iteration The weight parameter value, is the geographic vulnerability coefficient, calculated as ,in is the altitude coefficient, is the vegetation density, is the age of the device; For geographical features, Expressed as 、 and Their respective values, such as .

3. The method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions according to claim 1 is characterized in that: Step 1.2 involves designing constraints: Physical constraints: power conservation and equipment capacity limitations; Probability distribution constraints: missing values must obey the preset distribution ,in is the mean, is the covariance.

4. The method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions according to claim 1 is characterized in that: Step 1.3 includes: Step 1.3.1, initialize the population of the whale optimization algorithm using the chaotic map embedded with meteorological features; Each parameter to be evaluated is mapped to Logistic: ； in, is the control parameter, is the current iteration value, It is the next iteration value, For the Number of iterations; In typhoon situations, ; in, is the wind speed in meters per second; For each parameter to be estimated, an independent chaotic sequence is generated through chaotic mapping, and the linear transformation , is the scaling factor of the linear transformation, is the actual value after linear transformation, The chaotic sequence is generated by the chaotic map, and these values are in the interval [0,1] and exhibit chaotic characteristics; the chaotic sequence is mapped to the solution space to generate a chaotic initialized population ,in, is the d-th dimension parameter of the i-th individual, and are the upper and lower bounds of the parameter, A random number used to adjust value, so that it and Evenly distributed between Step 1.3.2, adopt the reverse propagation strategy: for the chaotic initialized population, calculate its reverse population , merge the chaotic initialized population and its reverse population, and select individuals with better fitness to form a new initial population; In step 1.3.3, after obtaining the new initial population, the whale group collaborative strategy is further used to jump out of the local optimal solution and output the distribution parameters and weight coefficients of the missing values.

5. The method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions according to claim 4 is characterized in that: Step 1.3.3 includes: Step 1.3.3.1, Encirclement Phase: The pilot whale guides the population to gather in the global optimal area, and other whales adjust their positions according to the position of the pilot whale: ； in, is the coefficient that decays with the number of iterations, For the The position of each individual in the search space, is the location of the optimal solution; Step 1.3.3.2, pursuit phase: When pilot whales approach their prey, they need to switch to random walk mode to explore the new area: ； in, is a uniformly distributed random number; Step 1.3.3.3, diversity maintenance: retain a certain proportion of suboptimal solutions in each position iteration, dynamically adjust the intensity of chaotic perturbations, and balance global search and local mining; In step 1.3.3.4, through the whale group collaborative strategy, we can effectively jump out of the local optimal solution and finally output the optimal distribution parameters and weight coefficients of the missing values.

6. The method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions according to claim 1 is characterized in that: In step 2.1, the complex probability distribution model of missing data adopts the Gaussian mixture model, which introduces the meteorological condition matrix and is mathematically defined as: ； in, For the meteorological conditions matrix Under the condition of missing data The probability distribution of For missing data, is the meteorological condition matrix, which includes dynamic meteorological parameters and geographical vulnerability coefficients; For the Mixture components represent sub-distributions under different meteorological-geographical coupling scenarios, for example: Corresponding to the typhoon dominant mode, It is a lightning-heavy rainfall coupled model, etc. is the total number of mixed ingredients; is the mixing ratio, indicating the The distribution of seeds under current meteorological conditions The weight under is a normal distribution, For the The mean and variance of the class; For the The mean of the sub-distribution represents the optimal filling value under a specific meteorological-geographical coupling scenario; For the The variance of the subset distribution is used to quantify the uncertainty of the fill value.

7. The method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions according to claim 1 is characterized in that: Step 2.2: Output the distribution parameters through the reverse chaotic whale transfer algorithm, including: mixing ratio Initialization, mean and variance Initialization, as the statistical value of the data sample; use the EM algorithm for training to reduce the calculation dimension; In step E, the posterior probability of each data point belonging to each latent variable is calculated using the following formula: ； in, The data point belongs to The probability of a mixture component; It is the loop index variable used to traverse all the mixed components; For meteorological conditions Next Initial probability weight under disaster-like mode, For the The expected value of data filling under disaster-like mode, For the Uncertainty in data filling under disaster-like models; M steps, update parameters To maximize the likelihood function: ； in, is the log-likelihood function, For the data set data points, For the A mixed ingredient, the upper bound chosen for the mixture components (the total number of mixture components), is the mixing ratio, is the probability density of the normal distribution, expressed as data points In the The probability density of the mixture components; Until the log-likelihood change is less than the threshold Or when the maximum number of iterations is reached, the iteration is stopped to obtain the trained complex probability distribution model, and the silhouette coefficient is used to verify the rationality of the model: ； in, is the silhouette coefficient, is the total number of data points in the dataset, For the data set data points, For data points To the cluster other than your own The minimum square distance between cluster centers, For data points The maximum square distance to the cluster center.

8. The method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions according to claim 1 is characterized in that: In step 2.4, the formula for calculating the confidence interval of the imputed value is as follows: ； in, To fill the results The confidence interval of , is a small positive number, To fill the result, To fill the mean of the results, For the standard normal distribution (1- ) quantile, is the sample standard deviation of the imputed result; The formula for calculating the probability density function is as follows: ； in, For the point The estimated density function at is a data point in the dataset, ranging from 1 to , is the total number of data points in the data set, K is the kernel function used to calculate the contribution of each data point to the estimated probability density, is the bandwidth. The choice of bandwidth will affect the smoothness of the kernel density estimation. is the point whose probability density needs to be estimated, For the data set The value of the data point; The statistical characteristics calculated above are passed to the subsequent power dispatch parameter processing module to quantify the random impact of missing data on the final result.

9. A device for filling and quantifying multi-source data of power system with dual-mode hybrid algorithm under extreme weather conditions, characterized in that: include: Parameter estimation module, including: Through extreme weather parameters, we select the missing data areas in the power dispatching system, introduce extreme weather influencing factors, perform preprocessing operations on the power system data, remove abnormal and duplicate data, and then standardize the missing data intervals; Build a weather-load data forecasting model; The optimal solution for missing data is obtained using the reverse chaotic whale transfer algorithm, which includes three steps: chaotic mapping with meteorological characteristics, reverse propagation, and whale group collaborative strategy to jump to the optimal solution. Uncertainty quantification module, including: Based on the parameter estimation results, a complex probability distribution model of missing data is constructed; Outputting distribution parameters through the reverse chaotic whale transfer algorithm, including training with a model with an adaptive strategy; Generate a large number of independent samples in the probability space to capture the mean and variance of the filling results; According to the statistical characteristics of the imputation results, the confidence interval and probability density function of the imputation value are calculated to quantify the random impact of missing data; The data completion and result output module combines the optimal solution of missing data obtained by the parameter estimation module with the probability density function obtained by the uncertainty quantification module to form a closed-loop complementary technology system, generate the final filling value, and input the filled data set into the power system.

10. An electronic 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 program, the steps of the power system multi-source data filling and quantification method of the dual-mode hybrid algorithm under extreme weather conditions as described in any one of claims 1 to 8 are implemented.

11. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for filling and quantifying multi-source data of a power system using a dual-mode hybrid algorithm under extreme weather conditions are implemented as described in any one of claims 1 to 8.

Images