Distribution network multi-business data correlation analysis and aggregation load model matching method

Abstract

The invention provides a distribution network multi-business data correlation analysis and aggregation load model matching method, and relates to the field of power system load modeling, and the method comprises the steps: obtaining distribution network multi-business data, carrying out the standardization preprocessing of the distribution network multi-business data, and obtaining a standardized data set; performing quality evaluation on the standardized data set to obtain a corresponding data credibility weight, and constructing an operation condition representation information set based on the standardized data set and the data credibility weight; obtaining a preset aggregation load model library, and calculating the correlation between the operation condition representation information set and candidate aggregation load models in the aggregation load model library; and performing weighted scoring on the correlation to obtain a correlation evaluation result, and matching a corresponding candidate aggregation load model. According to the method, operation condition representation information which can be used for aggregation load model structural element recognition and model library matching is constructed, and a correlation evaluation result is formed.

Description

Technical Field

[0001] This invention relates to the field of power system load modeling technology, specifically to a method for correlation analysis of multi-business data in distribution networks and matching of aggregated load models. Background Technology

[0002] As the power distribution network undergoes digital transformation, a large number of intelligent devices and information systems have been deployed, generating massive amounts of multi-source, heterogeneous data during the operation and management of the distribution network. This data includes marketing data, equipment ledger data, control data, fault recording data, and broadband dynamic characteristic monitoring data. These multi-source, heterogeneous data exhibit significant differences in sampling frequency, data structure, semantic expression, and quality level, and their complex relationships with the distribution network topology make it difficult to directly align and correlate the same distribution unit across multiple systems. Furthermore, data quality fluctuations such as missing data, anomalies, and inconsistent definitions objectively exist, making it difficult for multi-data types to form a stable input that can be directly used for load aggregation model structure matching, thus limiting its application in load modeling scenarios.

[0003] Existing methods still suffer from insufficient data processing in complex distribution network scenarios, and their output format is often difficult to directly serve power system load modeling. They also lack correlation analysis of multi-business data in distribution networks for matching aggregated load model structure. Summary of the Invention

[0004] This invention addresses the problems existing in the prior art by providing a method for correlation analysis of multi-business data in distribution networks and matching of aggregated load models. Based on the fusion of multi-business data in distribution networks, it introduces a data credibility characterization and load model library matching scoring mechanism, outputting the confidence level of structural elements and the probability of candidate model categories, providing support for subsequent parameter identification and online updates.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Obtain multi-business data of the distribution network, and perform standardized preprocessing on the multi-business data of the distribution network to obtain a standardized dataset; The standardized dataset is subjected to quality assessment to obtain the corresponding data credibility weights. Based on the standardized dataset and the data credibility weights, a set of operating condition characterization information is constructed. Obtain a pre-defined aggregated load model library and calculate the correlation between the set of operating condition characterization information and the candidate aggregated load models in the aggregated load model library; The correlation is weighted and scored to obtain the correlation evaluation result, and the corresponding candidate aggregate load model is matched.

[0006] In some embodiments, the acquisition of multi-business data in the distribution network involves standardizing and preprocessing the multi-business data in the distribution network to obtain a standardized dataset. Obtain the distribution network topology and collect multi-business data of the distribution network, wherein the multi-business data of the distribution network includes, but is not limited to: control cloud data, distribution network measurement data, marketing ledger data and online monitoring data of secondary equipment; Map the data of multiple business types in the power distribution network to a unified object identifier; Align the time granularity, perform linear interpolation on low-frequency distribution network multi-business data, and resample high-frequency distribution network multi-business data to obtain distribution network multi-business data with a unified time resolution; Establish a correspondence between the unified object identifier and the nodes of the distribution network topology to form a topology association; The distribution network multi-business data with the unified time resolution is deduplicated and filtered to obtain a standardized data set.

[0007] In some embodiments, the process of performing quality assessment on the standardized dataset to obtain corresponding data credibility weights includes: For each type of data from the same power distribution unit at the same time in the standardized dataset, calculate the quality assessment index separately; Calculate the uncertainty index for each data point based on the aforementioned quality assessment indicators; The uncertainty index is mapped to obtain the corresponding data credibility weight by a monotonically decreasing weight mapping function; The quality assessment indicators include anomaly probability, missing probability, and consistency score.

[0008] In some embodiments, the process of constructing a set of operational condition characterization information based on the standardized dataset and data credibility weights includes: Extract multi-business-type representational features related to the structural elements of the aggregated load model from the standardized dataset. Data credibility weights are mapped to representational features to obtain feature credibility weights corresponding to the representational features; A set of operational condition representation information is constructed based on representation features and feature credibility weights; Among them, the characterization features include distributed photovoltaic-related characterization features, electric vehicle-related characterization features, and general operating condition characterization features; In some embodiments, the construction process of the preset aggregated load model library includes: Several candidate aggregated load models are set up, and each candidate aggregated load model corresponds to a set of structural elements; A library of aggregated load models is established by compiling all candidate aggregated load models. The structural elements include one or more of the following: traditional static load elements, electric motor elements, distributed photovoltaic elements, and electric vehicle elements.

[0009] In some embodiments, the process of weighting the correlation to obtain a correlation evaluation result includes: For each candidate aggregated load model in the model library, the relevance is weighted and scored according to the feature confidence weight to obtain the corresponding matching score; Normalize all matching scores of candidate aggregated load models to obtain the corresponding candidate aggregated load model category probabilities; Calculate the confidence level of structural elements based on the feature subsets corresponding to the structural elements; The category probability of candidate aggregated load models and the confidence level of structural elements are used as the results of correlation evaluation.

[0010] In some embodiments, the process of matching the corresponding candidate aggregate load model includes: Filter out candidate aggregate load models whose class probability is below a threshold; Delete candidate aggregated load models whose structural element confidence does not meet the preset business requirements; Calculate the weighted sum of the category probability of the candidate aggregated load model and the confidence of the structural elements, and select the candidate aggregated load model corresponding to the maximum weighted sum; Periodically recalculate the class probabilities and structural element confidence scores of candidate aggregated load models, and adaptively match new candidate aggregated load models.

[0011] This invention proposes a system for aggregated load model matching based on the correlation of multi-business data in a distribution network, comprising: The data acquisition unit is configured to acquire multi-business data of the distribution network, and to perform standardized preprocessing on the multi-business data of the distribution network to obtain a standardized dataset; The construction unit is configured to perform quality assessment on the standardized dataset, obtain corresponding data credibility weights, and construct a set of operating condition characterization information based on the standardized dataset and the data credibility weights. The correlation unit is configured to obtain a preset aggregated load model library and calculate the correlation between the set of operating condition characterization information and the candidate aggregated load models in the aggregated load model library. The matching unit is configured to perform weighted scoring on the correlation to obtain the correlation evaluation result and match the corresponding candidate aggregate load model.

[0012] This invention proposes a computer device, comprising: At least one processor; and a memory storing a computer program that can run on the processor, wherein the processor executes the program to perform the steps of the method for correlation analysis of multi-business data in a distribution network and matching of aggregated load models.

[0013] This invention proposes a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the method for correlation analysis of multi-business data in a distribution network and matching of aggregated load models.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention proposes a method for correlation analysis of multi-business data in distribution networks and matching of aggregated load models. The method includes: acquiring multi-business data of the distribution network; performing standardized preprocessing on the multi-business data to obtain a standardized dataset; conducting quality assessment on the standardized dataset to obtain corresponding data credibility weights; constructing an operating condition characterization information set based on the standardized dataset and the data credibility weights; acquiring a preset aggregated load model library; calculating the correlation between the operating condition characterization information set and candidate aggregated load models in the aggregated load model library; performing weighted scoring on the correlation to obtain a correlation evaluation result; and matching the corresponding candidate aggregated load models.

[0015] This invention addresses the inconsistencies in identification, time granularity, and topological relationships among data from various sectors such as distribution network marketing, equipment, and control. It constructs operational condition characterization information that can be used for identifying structural elements of aggregated load models and matching them with model libraries. Under the constraint of data credibility weights, it generates correlation evaluation results and outputs the confidence level of structural elements and the probability of candidate aggregated load model categories, providing a basis for subsequent parameter identification and online updates.

[0016] In its operation, this invention standardizes inputs by unifying object identification, aligning time granularity, and associating topological data from various sectors of the distribution network, including marketing, equipment, and control. Furthermore, it generates data credibility weights based on missing, anomaly, and consistency assessments to reduce the interference of incomplete or conflicting data on structural judgment. On this basis, it constructs a set of operational condition characterization information and matches and scores it with the aggregated load model's structural element set and model library. The output includes the confidence scores of structural elements such as distributed photovoltaic access and output status, and electric vehicle charging status, as well as the probability of candidate model categories. This allows the correlation analysis results to directly serve the aggregated load model's structural matching and provides reliable prior evidence and screening support for subsequent parameter identification and online updates of the load model. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these drawings without creative effort.

[0018] Figure 1 The flowchart illustrates a method for correlation analysis of multi-business data and matching of aggregated load models in a power distribution network, as provided by this invention.

[0019] Figure 2 This is a system module diagram for matching aggregated load models based on the correlation of multi-business data in the power distribution network, as provided by the present invention.

[0020] Figure 3 A schematic diagram of the structure of an embodiment of the computer device provided by the present invention.

[0021] Figure 4 This is a schematic diagram of an embodiment of the computer-readable storage medium provided by the present invention.

[0022] Figure 5 This is a flowchart of one embodiment of a method for correlation analysis of multi-business data and matching of aggregated load models in a distribution network provided by the present invention. Detailed Implementation

[0023] The present invention will now be further described with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention. It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application.

[0024] It should be noted that all uses of "first" and "second" in the embodiments of the present invention are for the purpose of distinguishing two entities or parameters with the same name but different names. It is clear that "first" and "second" are only for the convenience of expression and should not be construed as limiting the embodiments of the present invention. Subsequent embodiments will not explain this in detail.

[0025] This invention proposes a method for correlation analysis of multi-business data in distribution networks and for matching aggregated load models. Please refer to [link / reference]. Figure 1 and Figure 5 ,include: S1. Obtain multi-business data of the distribution network, and perform standardized preprocessing on the multi-business data of the distribution network to obtain a standardized dataset; S2. Perform quality assessment on the standardized dataset to obtain the corresponding data credibility weights, and construct a set of operating condition characterization information based on the standardized dataset and the data credibility weights. S3. Obtain the preset aggregated load model library and calculate the correlation between the set of operating condition characterization information and the candidate aggregated load models in the aggregated load model library; S4. The correlation is weighted and scored to obtain the correlation evaluation result, and the corresponding candidate aggregate load model is matched.

[0026] The specific operation of this invention is as follows: Step 1: Collect data from various sectors of the power distribution network, including marketing, equipment, and control. Perform data preprocessing on the collected multi-source heterogeneous data and output standardized data after preprocessing. Step 2: Based on the assessment of missing, anomaly, and consistency of the standardized data, generate data credibility weights; Step 3: Based on the standardized data, construct a set of multi-business characteristic and operating condition representation information, and establish a set of aggregated load model structural elements and model library. The correlation is represented by the degree of matching between the set of operating condition representation information and the set of structural elements and model library. Step 4: Based on the data credibility weight, perform weighted scoring on the set of operating condition characterization information and the model library to obtain the correlation evaluation results. The output includes the confidence of structural elements corresponding to distributed photovoltaic access and output status, electric vehicle charging status, and the probability of candidate aggregated load model categories.

[0027] This invention acquires multi-business data from the distribution network and forms a unified dataset through standardized preprocessing. After obtaining data credibility weights through quality assessment, it constructs a set of operating condition characterization information and performs correlation calculations with candidate models in a pre-set aggregated load model library. The correlation results are weighted and scored based on the data weights to generate an evaluation result. The aggregated load model with the highest adaptability to the actual operating conditions is matched, effectively improving the accuracy of distribution network operating status perception.

[0028] In some embodiments, please refer to Figure 1 and Figure 5 The step involves acquiring multi-business data of the distribution network, performing standardized preprocessing on the multi-business data of the distribution network, and obtaining a standardized dataset. Obtain the distribution network topology and collect multi-business data of the distribution network, wherein the multi-business data of the distribution network includes, but is not limited to: control cloud data, distribution network measurement data, marketing ledger data and online monitoring data of secondary equipment; Map the data of multiple business types in the power distribution network to a unified object identifier; Align the time granularity, perform linear interpolation on low-frequency distribution network multi-business data, and resample high-frequency distribution network multi-business data to obtain distribution network multi-business data with a unified time resolution; Establish a correspondence between the unified object identifier and the nodes of the distribution network topology to form a topology association; The distribution network multi-business data with the unified time resolution is deduplicated and filtered to obtain a standardized data set.

[0029] This invention collects multi-business data from distribution networks, including but not limited to: control cloud data, distribution network measurement data, marketing ledger data, and online monitoring data of secondary equipment; it preprocesses the collected multi-source heterogeneous data to achieve unified object identification, time granularity alignment, and topological association, and outputs a standardized data set.

[0030] In a preferred embodiment of the present invention, the process for forming a standardized data foundation for correlation analysis is as follows.

[0031] (1) Automatically collect multi-business power data from the existing business system of the distribution network, including but not limited to control cloud data, distribution network measurement data, low-voltage distributed photovoltaic measurement data, marketing ledger data, dispatch report data, load forecast data, secondary equipment online monitoring data, meteorological forecast data, communication equipment ledger and maintenance data, new energy power forecast data, etc. (2) Based on the original data, complete the unified alignment of spatial dimensions, and match users, meters, transformers, transformer areas, feeders and other objects with their nodes and branches in the distribution network topology one by one, and map the data to a unified object identifier to realize the data merging of the same physical unit in different systems; (3) Based on the original data, complete the unified alignment of time granularity. For the sampling period differences of different data sources, perform linear interpolation on low-frequency data and resample high-frequency data to unify multi-source data to d suitable for load modeling.

[0032] (4) Based on the distribution network topology, establish a correspondence between the unified object identifier and the distribution network topology node to form the topology affiliation and upstream and downstream association for the aggregated load model.

[0033] (5) Further, systematic data cleaning is performed on the multi-source data. For duplicate reports and records, deduplication is performed according to timestamps. Preprocessing of the original data is achieved through convolutional smoothing filtering and median filtering to remove noise. Outliers are removed by replacing the center value of the sliding window with the median value, which can effectively filter out pulse-type instantaneous outliers. The specific steps of median filtering for noise removal are as follows: (a) Detect whether there are missing values. If there are missing values, fill them using linear interpolation. (b) Determine the window size k, where k is a positive odd number, slide the window and calculate the median within each window; (c) Calculate the replacement threshold T. When the error within this window is greater than T, replace the original value with the median. ; Where α is a constant, 3 < α < 5. d i The absolute residual within each window; Depending on the computation time requirements, the window size in step (b) can be adjusted. After determining the window size, start by gradually increasing α=3 until the replacement rate meets the expectations, thus completing the determination of the window size and the value of α.

[0034] (d) For the initial and final boundaries, fill the data with mirrors and determine if there are outliers.

[0035] (5) Output the standardized dataset D: ; Where x i For observation data, s i Used as a data source identifier.

[0036] This invention acquires the distribution network topology and collects data from various sectors, including control cloud, measurement, marketing ledgers, and secondary equipment monitoring. It maps these data to a unified object identifier to achieve data entity alignment. Combined with time granularity alignment technology, it interpolates and resamples data of different frequencies to ensure unified time resolution and establishes associations with topology nodes. Finally, after deduplication filtering, it forms a standardized data set, effectively solving the inconsistency problems of multi-source heterogeneous data in object identifiers, time dimensions, and topology space, and obtaining high-quality, structured, and spatiotemporally consistent data.

[0037] In some embodiments, please refer to Figure 1 and Figure 5 The process of evaluating the quality of the standardized dataset and obtaining the corresponding data credibility weights includes: For each type of data from the same power distribution unit at the same time in the standardized dataset, calculate the quality assessment index separately; Calculate the uncertainty index for each data point based on the aforementioned quality assessment indicators; The uncertainty index is mapped to obtain the corresponding data credibility weight by a monotonically decreasing weight mapping function; The quality assessment indicators include anomaly probability, missing probability, and consistency score.

[0038] This invention assesses data quality based on the missing, anomaly, and consistency of the standardized data, generating data credibility weights to characterize the credibility level of different data. For various types of power data from the same distribution unit at the same time, the anomaly probability is calculated separately. P anom missing probability P miss and consistency score c i And based on the aforementioned indicators, an uncertainty index is constructed for each data point. u iAccording to the aforementioned uncertainty index u i The corresponding data credibility weights are determined using a monotonically decreasing weight mapping function. ω ,in, Furthermore, the greater the uncertainty, the smaller the credibility weight.

[0039] In a preferred embodiment of the present invention, based on the obtained standardized data, a system is used to detect and quantify abnormal, missing, and conflicting data, and then calculate the credibility weight ω of each data point. The weight ω ranges from 0 to 1, with a larger weight indicating more credible data. The specific process is as follows.

[0040] (1) For each data sample x i Perform anomaly detection, anomaly probability P anom The Gaussian component parameters are obtained by calculating from the Gaussian mixture model and using the expectation-maximization algorithm. π k , μ k , Σ k For each sample point, calculate its likelihood value under the mixture model and normalize it to a probabilistic form. The calculation formula is as follows: ; in, γ This is a parameter for adjusting the steepness; The log-likelihood of each sample under a Gaussian mixture distribution. ; for The mean.

[0041] The steps of the Expectation-Maximization (EM) algorithm are as follows: (a) Set the Gaussian function based on the data π k , μ k , Σ k We first determine the initial values, and then calculate the corresponding log-likelihood function: ; (b) Step E: Introducing latent variables z Calculate the first k Posterior probability of clustering of loads γ ( z nk ): ; (c) M-step: Substitute the latent variables obtained in the E-step into the maximum likelihood estimation formula for the parameters to be determined, and iterate to obtain new results. π k , μ k and Σ k .

[0042] ; ; ; (d) Recalculate the log-likelihood function from step (a) and check whether the parameters or the log-likelihood function converge. If they do not converge, return to step (b). If they converge, the Gaussian function parameters are obtained. π k , μ k and Σ k .

[0043] (2) For each data sample x i Perform missing detection, missing probability P miss The determination is based on the data timestamp, and the calculation formula is as follows: ; Where, Δ t i For the current time interval, T 0 represents the desired sampling period.

[0044] (3) For each data sample x i Perform consistency checks and consistency scoring. c i The calculation formula is based on the correlation of multi-source observations or the time-series prediction error: ; in, σ j Indicates the first j Measurement variance of each measurement source, consistency score c i ∈[0,1], the closer to 1, the more consistent they are.

[0045] (4) Calculate the data sample x i Uncertainty index u i And credibility weight ω iThe uncertainty index is determined by the following comprehensive function: ; in, u i As an uncertainty index, ; P anom , P miss and c i These are the outlier probability, missing probability, and data consistency score, respectively. β 1. β 2. β 3 represents a non-negative weighting coefficient. Preferably, it can be empirically set based on distribution network operation experience and the degree of impact of different quality problems on the fusion results, satisfying... β 1+ β 2+ β 3=1, ω i As a credibility weight, in another implementation, the weight coefficient can also be calibrated offline based on historical operating data by minimizing the error between the fusion result and the reference value. This invention does not limit this. α This is the uncertainty adjustment coefficient.

[0046] This invention calculates quality assessment indicators such as anomaly probability, missing probability, and consistency score for various types of data from the same distribution unit at the same time in a standardized dataset. Based on these indicators, it quantifies the uncertainty of each data point and then uses a monotonically decreasing weight mapping function to transform the uncertainty into data credibility weights. This effectively achieves dynamic quantitative assessment of multi-source data quality, obtains data weights, and improves the accuracy and robustness of the matching aggregated load model results.

[0047] In some embodiments, please refer to Figure 1 and Figure 5 The process of constructing a set of operational condition characterization information based on the standardized dataset and data credibility weights includes: Extract multi-business-type representational features related to the structural elements of the aggregated load model from the standardized dataset. Data credibility weights are mapped to representational features to obtain feature credibility weights corresponding to the representational features; A set of operational condition representation information is constructed based on representation features and feature credibility weights; Among them, the characterization features include distributed photovoltaic-related characterization features, electric vehicle-related characterization features, and general operating condition characterization features.

[0048] In a preferred embodiment of the present invention, the obtained standardized data and the obtained data credibility weights are further transformed into inputs that can be used for identifying structural elements of the aggregated load model and matching with the model library, and the meaning of the correlation is clarified. The specific process is as follows.

[0049] (1) Based on standardized data, extract multi-business characteristic features related to the structural elements of the aggregated load model to form feature vectors. ; in, n Indicates the number of features. x j For the first j Each characterization feature may include at least one or more of the following categories: (a) Relevant characteristics of distributed photovoltaic systems: ledger capacity indication, weather, reverse power transmission indication, etc.; (b) Relevant characteristics of electric vehicles: charging pile ledger indications, time-of-use electricity price period indications, nighttime load surge magnitude and duration indicators, etc.; (c) General operating condition characteristics: season, temperature, workday, holiday, load fluctuation index, etc.

[0050] The aforementioned features can originate from a single system or from the joint calculation results of multiple systems after alignment; this invention does not limit the feature construction method. Optionally, the characterization features can be calculated and normalized by statistical correlation indicators such as Pearson correlation coefficient, mutual information, or Granger causality test, and then proceed to the weighted scoring in step four.

[0051] (2) Map the data credibility weights to the feature level to obtain the data credibility weights corresponding to the features. ω j The larger the value, the more reliable the representation feature.

[0052] (3) Based on the representation features and their credibility weights, construct a set of operating condition representation information: ; in,( x j , ω j ) indicates the first j Each characterization feature and its credibility weight.

[0053] This invention accurately extracts multi-business characteristic features closely related to the structural elements of the aggregated load model, such as distributed photovoltaic, electric vehicles, and general operating conditions, from a standardized dataset. It then scientifically maps data credibility weights to each characteristic feature to form feature credibility weights, constructing a set of operating condition characterization information containing feature values ​​and their credibility weights. This achieves feature fusion of multi-source heterogeneous data, providing comprehensive and reliable operating condition inputs for the aggregated load model and improving the model's ability to match complex distribution network operating scenarios.

[0054] In some embodiments, please refer to Figure 1 and Figure 5 The construction process of the preset aggregated load model library includes: Several candidate aggregated load models are set up, and each candidate aggregated load model corresponds to a set of structural elements; A library of aggregated load models is established by compiling all candidate aggregated load models. in ， The structural elements include one or more of the following: traditional static load elements, electric motor elements, distributed photovoltaic elements, and electric vehicle elements.

[0055] Establish a set of structural elements and a model library for the aggregated load model, and define the correlation as the degree of matching between the set of operating condition characterization information and the set of structural elements and the model library.

[0056] Establish aggregated load model library ,in K This represents the number of candidate models. Each candidate model... M k Corresponding to a set of structural elements Ω k This describes the types of structural elements included in the candidate model; the structural elements include one or more of the following: traditional static load elements, electric motor elements, distributed photovoltaic elements, and electric vehicle elements. Preferably, the model library can be configured to contain a candidate set of 16 transformer substation structural types.

[0057] Correlation is defined as a set of information characterizing operating conditions. X With candidate models M k and its structural elements set Ω k The degree of matching between them, that is, the degree to which the characterization information supports the composition of the structural elements of the candidate model under a given distribution unit and time, the higher the degree of matching, the stronger the correlation.

[0058] This invention achieves comprehensive coverage of the load characteristics of different business types in the distribution network, realizes the matching of operating condition characterization information with the model, enhances the adaptability of the model to complex load composition scenarios, and improves the parameter optimization efficiency of the model through structural element decoupling.

[0059] In some embodiments, please refer to Figure 1 and Figure 5 The process of weighting and scoring the correlation to obtain the correlation evaluation result includes: For each candidate aggregated load model in the model library, the relevance is weighted and scored according to the feature confidence weight to obtain the corresponding matching score; Normalize all matching scores of candidate aggregated load models to obtain the corresponding candidate aggregated load model category probabilities; Calculate the confidence level of structural elements based on the feature subsets corresponding to the structural elements; The category probability of candidate aggregated load models and the confidence level of structural elements are used as the results of correlation evaluation.

[0060] For the k-th candidate aggregated load model in the model library, the matching score of the candidate model is calculated based on the representation feature xj in the set of operating condition representation information, the data confidence weight ωj corresponding to the representation feature, and the representation feature importance coefficient αj. ; in, For characterization features x j For the k The matching contribution function of each candidate model; the matching score S k Normalization is performed to obtain the class probabilities of candidate aggregated load models. p k , ; The confidence levels of structural elements such as distributed photovoltaic access and output status, and electric vehicle charging status are obtained by weighted summarization of the representation features corresponding to the structural elements and the representation features subsets corresponding to the structural elements.

[0061] In a preferred embodiment of the present invention, the method is used to calculate the correlation evaluation results based on the set of operating condition characterization information and the model library, and output the confidence level of structural elements and the probability of candidate model categories. The specific process is as follows.

[0062] (1) For each candidate model in the model library M k Calculate matching score S k : ; in, α j For the first j The importance coefficients of each representation feature satisfy the following conditions: α j ≥0 and , ω j For the first j The data credibility weight of each characteristic feature To characterize features x j For the k The matching contribution function of each candidate model.

[0063] Alternatively, the matching contribution function can take the form that is easy to implement in engineering: (a) If the candidate model M k Inclusion and features x j The corresponding structural elements are then taken (b) If the candidate model M k Not containing and features x j The corresponding structural elements are then taken (2) Normalize the matching scores of all candidate models to obtain the probability of each candidate model category: ; in, p k Candidate models M k The class probability, whose value ranges from 0 to 1 and .

[0064] (3) Calculate and output the confidence scores of structural elements, using the feature subsets corresponding to structural element e. Calculate the confidence score of structural elements using the input. ; in, This represents the confidence level of structural element e. Structural element e includes distributed photovoltaic grid connection and output status, electric vehicle charging status, etc.

[0065] (4) Output the correlation evaluation results, including the candidate aggregated load model category probability {pk} and the confidence level of structural elements. The output results can be used for aggregated load model structure selection and provide input for subsequent parameter identification and online updates.

[0066] This invention obtains the model category probability by weighting and normalizing the relevance of each candidate aggregated load model based on feature confidence weights, and calculates the structural element confidence by combining structural element feature subsets, forming a comprehensive relevance evaluation result that includes model matching probability and structural reliability. This quantifies the degree of fit between different models and actual working conditions, and ensures the model's ability to explain key load characteristics through structural element confidence assessment, providing a basis for the selection of aggregated load models.

[0067] In some embodiments, please refer to Figure 1 and Figure 5 The process of matching the corresponding candidate aggregated load model includes: Filter out candidate aggregate load models whose class probability is below a threshold; Delete candidate aggregated load models whose structural element confidence does not meet the preset business requirements; Calculate the weighted sum of the category probability of the candidate aggregated load model and the confidence of the structural elements, and select the candidate aggregated load model corresponding to the maximum weighted sum; Periodically recalculate the class probabilities and structural element confidence scores of candidate aggregated load models, and adaptively match new candidate aggregated load models.

[0068] This invention filters low-probability models, eliminates models with substandard confidence levels for structural elements, and selects the best-fitting model based on a weighted sum of class probability and structural element confidence levels. Simultaneously, a dynamic update mechanism is established to periodically re-evaluate model suitability, effectively eliminating candidate models with poor matching to actual operating conditions. Multi-dimensional quantitative indicators ensure the selected model's ability to interpret key load characteristics, enabling adaptive optimization of the aggregated load model as the distribution network's operating status changes. This significantly improves the model's applicability and prediction accuracy in complex dynamic scenarios.

[0069] This invention proposes a system for aggregated load model matching based on the correlation of multi-business data in the distribution network. Please refer to [link / reference]. Figure 2 ,include: The acquisition unit 100 is configured to acquire multi-business data of the distribution network, and to perform standardized preprocessing on the multi-business data of the distribution network to obtain a standardized dataset; The construction unit 200 is configured to perform quality assessment on the standardized dataset, obtain corresponding data credibility weights, and construct a set of operating condition characterization information based on the standardized dataset and the data credibility weights. The correlation unit 300 is configured to obtain a preset aggregated load model library and calculate the correlation between the set of operating condition characterization information and the candidate aggregated load models in the aggregated load model library. The matching unit 400 is configured to perform weighted scoring on the correlation to obtain the correlation evaluation result and match the corresponding candidate aggregate load model.

[0070] In some embodiments, please refer to Figure 5 The specific process of this invention is as follows: Data acquisition and standardization preprocessing involve collecting multi-business data, including marketing data, equipment data, and control data from the power distribution network. The data undergoes time-aligned processing and convolutional smoothing filtering. Missing values ​​are checked; if any are found, linear interpolation is performed, and the center value is replaced using a sliding window. Finally, the dataset is standardized.

[0071] Data quality assessment and data credibility weight generation evaluate data quality by calculating anomaly probability, missing probability, and consistency score. Data credibility weights are then generated based on the assessment results.

[0072] The process involves constructing and defining the correlations of operational condition characterization information sets, extracting characterization features from multiple business types, including general operational condition characterization features, distributed photovoltaic-related characterization features, and electric vehicle-related characterization features. An operational condition characterization information set is constructed, and an aggregated load model library is built. The correlations between features are defined.

[0073] The system calculates the weighted score and outputs the results of model library matching, including the matching score and class probability of candidate models. The final result is output based on the confidence level of structural elements.

[0074] It should be noted that this invention utilizes multi-source data fusion technology to map control cloud, measurement, marketing ledger, and secondary equipment monitoring data to a unified object identifier. Combined with time granularity alignment and topological association, a standardized dataset is constructed. To address data quality differences, an uncertainty index is calculated using anomaly probability, missing probability, and consistency score, and feature credibility weights are generated through a nonlinear weight mapping function.

[0075] The model library construction phase adopts a modular design, covering combinations of structural elements such as traditional static loads, electric motors, distributed photovoltaics, and electric vehicles, forming a model library containing 16 categories of candidate models. In the relevance evaluation phase, the relevance between operating conditions and model structural elements is weighted and scored based on feature credibility weights. For example, at 14:00, the standardized dataset for a certain regional distribution network shows that the confidence weight of distributed photovoltaic power output is 0.82, and that of electric vehicle charging is 0.78. In the model library, the ZIP model has a class probability of 0.79 and structural element confidence scores: static load 0.92, photovoltaic 0.85; the exponential recovery model has a class probability of 0.68 and structural element confidence scores: static load 0.88, motor 0.72. After weighted calculation, the ZIP model's overall score is 0.79×0.85+0.82×0.92=0.83, which is higher than the exponential recovery model's score of 0.68×0.72+0.72×0.88=0.71. Therefore, the ZIP model is selected. This effectively solves the problems of uncontrollable data quality and poor model adaptability in traditional modeling methods, reducing the prediction error of the aggregated load model under complex operating conditions and significantly improving the accuracy of distribution network operation control.

[0076] Based on the same inventive concept, according to another aspect of the present invention, such as Figure 3 As shown, an embodiment of the present invention also provides a computer device 30, which includes a processor 310 and a memory 320. The memory 320 stores a computer program 321 that can be run on the processor. When the processor 310 executes the program, it performs the steps of the method described above.

[0077] Based on the same inventive concept, according to another aspect of the present invention, such as Figure 4 As shown, embodiments of the present invention also provide a computer-readable storage medium 40, which stores a computer program 410 that, when executed by a processor, performs the methods described above.

[0078] Embodiments of the present invention may also include a corresponding computer device. The computer device includes a memory, at least one processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes any of the methods described above when executing the program.

[0079] The memory, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as program instructions / modules in the embodiments of this application. The processor executes various functional applications and data processing of the device by running the non-volatile software programs, instructions, and modules stored in the memory, thereby implementing the above-described method.

[0080] The memory may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the device. Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In embodiments, the memory may optionally include memory remotely located relative to the processor, which can be connected to the local module via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0081] Finally, it should be noted that those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The storage medium for the program can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. The above computer program embodiments can achieve the same or similar effects as any of the corresponding foregoing method embodiments.

[0082] Those skilled in the art will also understand that the various exemplary logic blocks, modules, circuits, and algorithm steps described in conjunction with the disclosure herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the functionality of various illustrative components, blocks, modules, circuits, and steps has been generally described. Whether this functionality is implemented as software or as hardware depends on the specific application and the design constraints imposed on the system as a whole. Those skilled in the art can implement the functionality in various ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the embodiments disclosed herein.

[0083] The above are exemplary embodiments disclosed in this invention. However, it should be noted that various changes and modifications can be made without departing from the scope of the embodiments of this invention as defined by the claims. The functions, steps, and / or actions of the methods according to the disclosed embodiments described herein do not need to be performed in any particular order. The sequence numbers of the disclosed embodiments of this invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. Furthermore, although the elements disclosed in the embodiments of this invention may be described or claimed individually, they may be understood as multiple unless explicitly limited to a singular number.

[0084] It should be understood that, as used herein, the singular form “a” is intended to include the plural form as well, unless the context clearly supports an exception. It should also be understood that, as used herein, “and / or” refers to any and all possible combinations of one or more of the associated listed items.

[0085] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention (including the claims) is limited to these examples. Within the framework of the invention, technical features of the above embodiments or different embodiments can be combined, and many other variations of different aspects of the invention exist, which are not provided in the details for the sake of brevity. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the protection scope of the invention.

Claims

1. A method for correlation analysis of multi-business data in a power distribution network and matching of aggregated load models, characterized in that, include: Obtain multi-business data of the distribution network, and perform standardized preprocessing on the multi-business data of the distribution network to obtain a standardized dataset; The standardized dataset is subjected to quality assessment to obtain the corresponding data credibility weights. Based on the standardized dataset and the data credibility weights, a set of operating condition characterization information is constructed. Obtain a pre-defined aggregated load model library and calculate the correlation between the set of operating condition characterization information and the candidate aggregated load models in the aggregated load model library; The correlation is weighted and scored to obtain the correlation evaluation result, and the corresponding candidate aggregate load model is matched.

2. The method for correlation analysis of multi-business data and matching of aggregated load models in a distribution network according to claim 1, characterized in that, The process involves acquiring multi-business data from the distribution network, performing standardized preprocessing on the multi-business data, and obtaining a standardized dataset. Obtain the distribution network topology and collect multi-business data of the distribution network, wherein the multi-business data of the distribution network includes, but is not limited to: control cloud data, distribution network measurement data, marketing ledger data and online monitoring data of secondary equipment; Map the data of multiple business types in the power distribution network to a unified object identifier; Align the time granularity, perform linear interpolation on low-frequency distribution network multi-business data, and resample high-frequency distribution network multi-business data to obtain distribution network multi-business data with a unified time resolution; Establish a correspondence between the unified object identifier and the nodes of the distribution network topology to form a topology association; The distribution network multi-business data with the unified time resolution is deduplicated and filtered to obtain a standardized data set.

3. The method for correlation analysis of multi-business data and matching of aggregated load models in a distribution network according to claim 1, characterized in that, The process of evaluating the quality of the standardized dataset and obtaining the corresponding data credibility weights includes: For each type of data from the same power distribution unit at the same time in the standardized dataset, calculate the quality assessment index separately; Calculate the uncertainty index for each data point based on the aforementioned quality assessment indicators; The uncertainty index is mapped to obtain the corresponding data credibility weight by a monotonically decreasing weight mapping function; The quality assessment indicators include anomaly probability, missing probability, and consistency score.

4. The method for correlation analysis of multi-business data and matching of aggregated load models in a distribution network according to claim 1, characterized in that, The process of constructing a set of operational condition characterization information based on the standardized dataset and data credibility weights includes: Extract multi-business-type representational features related to the structural elements of the aggregated load model from the standardized dataset. Data credibility weights are mapped to representational features to obtain feature credibility weights corresponding to the representational features; A set of operational condition representation information is constructed based on representation features and feature credibility weights; Among them, the characterization features include distributed photovoltaic-related characterization features, electric vehicle-related characterization features, and general operating condition characterization features.

5. The method for correlation analysis of multi-business data and matching of aggregated load models in a distribution network according to claim 4, characterized in that, The construction process of the preset aggregated load model library includes: Several candidate aggregated load models are set up, and each candidate aggregated load model corresponds to a set of structural elements; A library of aggregated load models is established by compiling all candidate aggregated load models. The structural elements include one or more of the following: traditional static load elements, electric motor elements, distributed photovoltaic elements, and electric vehicle elements.

6. The method for correlation analysis of multi-business data and matching of aggregated load models in a distribution network according to claim 5, characterized in that, The process of weighting and scoring the correlation to obtain the correlation evaluation result includes: For each candidate aggregated load model in the model library, the relevance is weighted and scored according to the feature confidence weight to obtain the corresponding matching score; Normalize all matching scores of candidate aggregated load models to obtain the corresponding candidate aggregated load model category probabilities; Calculate the confidence level of structural elements based on the feature subsets corresponding to the structural elements; The category probability of candidate aggregated load models and the confidence level of structural elements are used as the results of correlation evaluation.

7. The method for correlation analysis of multi-business data and matching of aggregated load models in a distribution network according to claim 6, characterized in that, The process of matching the corresponding candidate aggregate load model includes: Filter out candidate aggregate load models whose class probability is below a threshold; Delete candidate aggregated load models whose structural element confidence does not meet the preset business requirements; Calculate the weighted sum of the category probability of the candidate aggregated load model and the confidence of the structural elements, and select the candidate aggregated load model corresponding to the maximum weighted sum; Periodically recalculate the class probabilities and structural element confidence scores of candidate aggregated load models, and adaptively match new candidate aggregated load models.

8. A system for matching aggregated load models based on the correlation of multi-business data in a distribution network, characterized in that, include: The data acquisition unit is configured to acquire multi-business data of the distribution network, and to perform standardized preprocessing on the multi-business data of the distribution network to obtain a standardized dataset; The construction unit is configured to perform quality assessment on the standardized dataset, obtain corresponding data credibility weights, and construct a set of operating condition characterization information based on the standardized dataset and the data credibility weights. The correlation unit is configured to obtain a preset aggregated load model library and calculate the correlation between the set of operating condition characterization information and the candidate aggregated load models in the aggregated load model library. The matching unit is configured to perform weighted scoring on the correlation to obtain the correlation evaluation result and match the corresponding candidate aggregate load model.

9. A computer device, comprising: At least one processor; The processor includes a memory storing a computer program that can run on the processor, characterized in that the processor executes the program by performing the steps of the method for correlation analysis of multi-business data in a distribution network and matching of aggregated load models as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it performs the steps of the method for correlation analysis of multi-business data in a distribution network and matching of aggregated load models as described in any one of claims 1 to 7.

Images