An active power distribution network comprehensive fault diagnosis method based on a fusion algorithm and related devices

Abstract

The application provides an active power distribution network comprehensive fault diagnosis method based on a fusion algorithm and related devices, and belongs to the technical field of power system fault diagnosis. The application first collects multi-source monitoring data of the active power distribution network; then, abnormal data and dimensional differences are eliminated through data preprocessing, and a weighted data layer fusion strategy positively correlated with historical credibility of the data source is adopted to generate a complete fusion data set; subsequently, fault features are extracted from the fusion data; on this basis, candidate fault sections are screened and multi-class independent positioning probability evidence is generated in combination with the multi-source monitoring data and the fault features; finally, D-S evidence theory is adopted to fuse the multi-class independent evidence for decision-making, thereby realizing accurate fault positioning under the conditions of bidirectional power flow and incomplete information, comprehensively improving fault tolerance, accuracy and scene adaptability of the fault diagnosis method, and meeting the needs of safe and stable operation of the active power distribution network.

Description

Technical Field

[0001] This invention belongs to the field of power system fault diagnosis technology, specifically relating to an active distribution network integrated fault diagnosis method and related devices based on a fusion algorithm. Background Technology

[0002] In the power system sector, the distribution network is a crucial link in power transmission and distribution, directly impacting the quality of electricity supply and reliability for users. With adjustments to the energy structure and technological advancements, the large-scale integration of distributed generation (DG) has become a significant trend in distribution network development.

[0003] The integration of distributed generation (DG) transforms the distribution network from a traditional passive network into an active distribution network with multiple power sources and bidirectional power flow. This transformation brings a series of new changes to the operation and management of the distribution network, among which the changes in fault characteristics are particularly significant. For example, the amplitude and phase characteristics of fault currents become more complex, and power flow becomes bidirectional.

[0004] Traditional fault diagnosis methods for distribution networks often rely on a single data source or a single algorithm. When faced with problems such as weak fault characteristics, bidirectional power flow, and easy information distortion in active distribution networks, they suffer from insufficient fault tolerance, limited accuracy, and poor adaptability to different scenarios, making it difficult to meet the needs of safe and stable operation of active distribution networks. Summary of the Invention

[0005] Therefore, it is necessary to provide a comprehensive fault diagnosis method and related device for active distribution networks based on fusion algorithms to address the above-mentioned technical problems, so as to solve the diagnosis problems caused by weak fault characteristics, bidirectional power flow, and information distortion in active distribution networks.

[0006] In a first aspect, the present invention provides a comprehensive fault diagnosis method for active distribution networks based on a fusion algorithm, comprising the following steps:

[0007] Collect multi-source monitoring data from the active power distribution network;

[0008] Multi-source monitoring data are preprocessed to obtain a standardized monitoring dataset;

[0009] The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to their corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0010] Feature extraction is performed on the complete fused dataset to obtain the fault feature set;

[0011] Based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined.

[0012] The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion results of each segment.

[0013] Furthermore, the multi-source monitoring data is preprocessed to obtain a standardized monitoring dataset, including:

[0014] Outlier removal is performed on the multi-source monitoring data to obtain the first multi-source monitoring data.

[0015] Missing values ​​were filled into the first multi-source monitoring data to obtain the second multi-source monitoring data;

[0016] The second multi-source monitoring data is standardized to obtain a standardized monitoring dataset.

[0017] Furthermore, the standardized monitoring dataset undergoes data fusion processing to generate a complete fused dataset, including:

[0018] The reliability of each data source is calculated based on its historical operational data.

[0019] Based on credibility, the fusion weights of each data source are assigned; the fusion weights are positively correlated with credibility.

[0020] The standardized monitoring dataset is weighted and fused based on the fusion weights, and missing data is filled in to obtain a complete fused dataset.

[0021] Furthermore, feature extraction is performed on the complete fused dataset to obtain a fault feature set, including:

[0022] Extract initial fault features from the complete fused dataset;

[0023] Redundant features are removed from the initial fault features, and a fault feature set is obtained based on the removed fault features.

[0024] Furthermore, based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined, including:

[0025] Based on multi-source monitoring data, fault sections are screened to obtain a set of candidate fault sections;

[0026] For each fault segment in the candidate fault segment set, different evidence generation algorithms are used to generate multiple evidence in combination with the fault feature set, so as to obtain multi-class localization probability evidence for each segment; the evidence generation algorithm is an algorithm that uses the probability that each segment is a fault segment as evidence.

[0027] Furthermore, multi-evidence generation processing is performed to obtain multiple types of location probability evidence corresponding to each segment, including:

[0028] The first type of segment location probabilistic evidence is generated by an integer linear programming solution method based on an improved multiverse algorithm. The integer linear programming solution method based on the improved multiverse algorithm is an improved multiverse algorithm that introduces a chaotic initialization strategy and adopts a nonlinear adaptive universe expansion rate update formula. It is used to solve the fault location integer linear programming model constrained by power flow conservation and signal coverage.

[0029] The second type of segment location probability evidence is generated using the switch quantity statistical matching method;

[0030] The feature correlation matching method is used to generate the third type of segment location probability evidence.

[0031] Furthermore, the DS evidence theory is used to fuse the probabilistic evidence of each segment, and the final faulty segment is determined based on the fusion results of each segment, including:

[0032] For each fault segment in the candidate fault segment set, the basic probability allocation corresponding to each type of localization probability evidence is determined, resulting in multiple basic probability allocations.

[0033] The DS synthesis rule is used to fuse and calculate the basic probability assignments of multiple categories to obtain the comprehensive failure probability of each section;

[0034] The section whose overall failure probability meets the preset conditions is determined as the final failure section.

[0035] Secondly, the present invention provides an active power distribution network integrated fault diagnosis device based on a fusion algorithm, comprising:

[0036] The data acquisition module is used to collect multi-source monitoring data from active power distribution networks;

[0037] The data preprocessing module is used to preprocess multi-source monitoring data to obtain a standardized monitoring dataset;

[0038] The data layer fusion module is used to perform data layer fusion processing on standardized monitoring datasets to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0039] The feature extraction module is used to extract features from the complete fused dataset to obtain a fault feature set;

[0040] The candidate fault segment determination module is used to determine the set of candidate fault segments and the corresponding location probability evidence for each segment based on multi-source monitoring data and fault feature set;

[0041] The final fault segment determination module is used to fuse the location probability evidence of each segment using DS evidence theory, and determine the final fault segment based on the fusion processing results of each segment.

[0042] Thirdly, the present invention provides a computer device, the device including a processor and a memory:

[0043] The memory is used to store computer programs and send the instructions of the computer programs to the processor;

[0044] The processor executes the following steps according to the instructions of the computer program:

[0045] Collect multi-source monitoring data from the active power distribution network;

[0046] Multi-source monitoring data are preprocessed to obtain a standardized monitoring dataset;

[0047] The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to their corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0048] Feature extraction is performed on the complete fused dataset to obtain the fault feature set;

[0049] Based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined.

[0050] The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion results of each segment.

[0051] Fourthly, the present invention provides a computer-readable storage medium on which a computer program is stored, and when executed by a processor, the computer program performs the following steps:

[0052] Collect multi-source monitoring data from the active power distribution network;

[0053] Multi-source monitoring data are preprocessed to obtain a standardized monitoring dataset;

[0054] The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to their corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0055] Feature extraction is performed on the complete fused dataset to obtain the fault feature set;

[0056] Based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined.

[0057] The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion results of each segment.

[0058] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0059] Collect multi-source monitoring data from the active power distribution network;

[0060] Multi-source monitoring data are preprocessed to obtain a standardized monitoring dataset;

[0061] The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to their corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0062] Feature extraction is performed on the complete fused dataset to obtain the fault feature set;

[0063] Based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined.

[0064] The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion results of each segment.

[0065] In summary, this invention provides a comprehensive fault diagnosis method and related device for active distribution networks based on a fusion algorithm. First, it collects multi-source monitoring data from the active distribution network. Then, it eliminates abnormal data and dimensional differences through data preprocessing and employs a weighted data layer fusion strategy positively correlated with the historical reliability of the data source to effectively suppress the impact of information distortion, generating a complete fusion dataset with significantly improved data integrity and reliability. Subsequently, it extracts fault features from the fusion data to strengthen the expression of weak fault features, solving the problem of unclear fault features in active distribution networks. Based on this, it combines multi-source monitoring data and fault features to screen candidate fault sections and generate multiple types of independent location probability evidence. This not only avoids the limitations of a single algorithm through multi-dimensional evidence construction but also significantly reduces computational complexity through initial screening of candidate sections. Finally, it uses DS evidence theory to perform fusion decision-making on multiple types of independent evidence, fully leveraging its advantages in handling uncertain information and conflicting evidence. This achieves accurate fault location under conditions of bidirectional power flow and incomplete information, comprehensively improving the fault diagnosis method's fault tolerance, accuracy, and scenario adaptability, meeting the requirements for the safe and stable operation of active distribution networks. Attached Figure Description

[0066] 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 drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart illustrating an active power distribution network integrated fault diagnosis method based on a fusion algorithm in one embodiment of the present invention.

[0068] Figure 2 This is a flowchart illustrating the implementation of an active power distribution network integrated fault diagnosis method based on a fusion algorithm according to the present invention.

[0069] Figure 3 This is a block diagram of an active power distribution network integrated fault diagnosis device based on a fusion algorithm, as shown in one embodiment of the present invention.

[0070] Figure 4 This is a block diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0071] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0072] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0073] The background technology of this invention will be further introduced below.

[0074] With the accelerated pace of the global energy transition, the scale of distributed generation (DG), represented by photovoltaic and wind power, in distribution networks continues to expand. Traditional unidirectional, radial, passive distribution networks are gradually evolving into complex active distribution networks with multiple power sources, multiple nodes, and bidirectional power flow. This fundamental shift has greatly enhanced the distribution network's ability to absorb renewable energy and promoted the low-carbon development of the power system. However, it has also completely altered the fault characteristics and operating patterns of the distribution network, posing challenges to traditional fault diagnosis technologies.

[0075] The fault characteristics of active distribution networks differ significantly from those of traditional passive distribution networks. First, the integration of distributed generation means that fault current is no longer solely supplied by the system-side power source; the fault point simultaneously receives current from both the system side and multiple distributed generation sources, resulting in high uncertainty and complexity in the amplitude, phase, and direction of the fault current. This complex fault current distribution makes it difficult for traditional fault diagnosis methods based on single electrical quantity thresholds and unidirectional power flow assumptions to accurately capture effective fault characteristics, with weakened fault characteristics being particularly prominent. Second, the existence of bidirectional power flow renders traditional fault power direction judgment principles ineffective, making it impossible to determine the fault section through simple current direction comparisons, easily leading to misjudgments or omissions in fault location. Furthermore, the output of distributed generation exhibits significant intermittency and fluctuation, with substantial differences in output across different time periods and weather conditions, further increasing the diversity of fault characteristics and the difficulty of diagnosis.

[0076] Meanwhile, multi-source monitoring systems for active distribution networks also face the problem of inconsistent data quality. To achieve comprehensive awareness of the distribution network's operational status, numerous monitoring devices, such as feeder terminal units (FTUs), transformer terminal units (TTUs), and micro phasor measurement units (μPMUs), are widely deployed in the distribution network. However, these devices are mostly installed in harsh outdoor environments, making them susceptible to electromagnetic interference, lightning strikes, and unstable power supply. During data acquisition and transmission, issues such as data loss, data distortion, communication delays, and even false alarms frequently occur. The reliability of a single data source is difficult to guarantee, and fault diagnosis results based on unreliable data naturally cannot meet accuracy requirements.

[0077] Most existing fault diagnosis methods for distribution networks still follow the technical approach of traditional passive distribution networks, and generally suffer from limitations due to reliance on a single data source or a single algorithm. For example, while switch-based diagnostic methods based solely on FTU alarm signals are computationally simple and have good real-time performance, they are extremely sensitive to signal distortion and missing signals, exhibiting very poor fault tolerance. Fault location methods using only traditional mathematical optimization algorithms are prone to getting trapped in local optima, and their convergence speed is slow and accuracy is limited in complex fault scenarios. Diagnostic methods based on a single artificial intelligence algorithm require a large amount of high-quality labeled data for training, resulting in insufficient generalization ability and scenario adaptability. These methods all demonstrate significant shortcomings when facing complex problems such as weak fault characteristics, bidirectional power flow, and easy information distortion in active distribution networks, making it difficult to ensure the safe and stable operation of active distribution networks.

[0078] To address the shortcomings of the existing technologies, this invention proposes a comprehensive fault diagnosis method and related device for active power distribution networks based on a fusion algorithm. By combining multi-source data fusion with multi-algorithm decision fusion, it effectively solves the technical challenges of insufficient fault tolerance, limited accuracy, and poor scenario adaptability in active power distribution network fault diagnosis. The various embodiments of this invention are described in detail below.

[0079] Please see Figure 1 This invention provides a comprehensive fault diagnosis method for active distribution networks based on a fusion algorithm, comprising the following steps:

[0080] S101: Collect multi-source monitoring data of active power distribution networks.

[0081] The active distribution network refers to a distribution network connected to distributed power sources (such as photovoltaic and wind power). Multi-source monitoring data includes overcurrent alarm signals from feeder terminal units (FTUs), undervoltage data from distribution transformer terminal units (TTUs), transient electrical quantity data from micro synchronous phasor measurement devices (μPMUs), distributed generation (DG) output data, and distribution network topology parameters. Specifically, FTU overcurrent alarm signals are collected by the FTU devices at the feeder sectionalizing switches and include three status indicators: forward overcurrent, reverse overcurrent, and no overcurrent. TTU undervoltage data is collected through the distribution transformer terminal, reflecting abnormal node voltage conditions. μPMU transient electrical quantity data includes high-precision synchronous data such as transient zero-sequence voltage and current. DG output data covers the real-time output value and fluctuation information of photovoltaic distributed power sources. Distribution network topology parameters include line impedance, node connection relationships, and DG connection locations.

[0082] It should be noted that conventional data collection methods for power distribution network fault diagnosis usually rely on a single terminal or a limited data source. This step, however, involves data collection using multiple terminals and multiple dimensions.

[0083] Optionally, data can be collected through equipment such as distribution network automation systems, marketing metering systems, and synchronous phasor measurement devices, according to the real-time / sampling rate requirements of different terminals (FTU / TTU data at the 20ms level, μPMU data at ≥5kHz), and topology parameters can be updated periodically.

[0084] S102: Preprocess the multi-source monitoring data to obtain a standardized monitoring dataset.

[0085] Preprocessing refers to the cleaning and standardization of the collected raw data; standardized monitoring datasets refer to datasets after removing anomalies and unifying the units of measurement.

[0086] It should be noted that data preprocessing includes operations such as outlier removal, missing value imputation, and unit unification, with the aim of eliminating the impact of data noise and format differences on subsequent algorithms.

[0087] Optionally, outlier removal, missing value imputation, and standardization are performed sequentially on the multi-source data to obtain a standardized monitoring dataset.

[0088] S103: Perform data layer fusion processing on the standardized monitoring dataset to generate a complete fused dataset; data layer fusion processing is an operation of fusion of monitoring data from different data sources according to corresponding weights; the corresponding weights of each data source are positively correlated with the reliability of the data source.

[0089] Among them, data layer fusion processing refers to the operation of merging monitoring data from different data sources according to corresponding weights; data source credibility refers to the probability of no distortion in the data source based on historical operating data; fusion weight is positively correlated with data source credibility.

[0090] It should be noted that data layer fusion includes weighted fusion, missing data completion, etc., with the aim of integrating complementary information from multiple sources to generate fused data that is complete and highly reliable.

[0091] Optionally, the credibility is calculated based on the historical distortion rate of each data source, dynamic weights are assigned according to the credibility, and a complete fused dataset is generated through weighted fusion and missing data completion.

[0092] S104: Extract features from the complete fused dataset to obtain the fault feature set.

[0093] Among them, the fault feature set refers to the set of features extracted from the complete fusion dataset that can reflect the essence of faults in the active distribution network, and redundant features with low contribution to fault diagnosis are eliminated.

[0094] It should be noted that feature extraction can be implemented in various ways, including time-domain / frequency-domain feature extraction and correlation analysis, with the aim of reducing data dimensionality and highlighting the essential characteristics of the fault.

[0095] Optionally, initial fault features can be extracted from the complete fused dataset, and redundant features can be removed to obtain a fault feature set.

[0096] S105: Based on multi-source monitoring data and fault feature sets, determine the candidate fault segment set and the corresponding location probability evidence for each segment.

[0097] Among them, the candidate fault segment set refers to a set of several segments that may be faulty, which are initially screened based on multi-source data; the location probability evidence refers to the probability information reflecting that each segment is a faulty segment.

[0098] It should be noted that the methods for screening fault sections and generating evidence include signal correlation analysis and probability model calculation, with the aim of narrowing the scope of diagnosis and constructing multiple types of independent evidence.

[0099] Optionally, candidate fault segments can be screened based on multi-source monitoring data, and for each candidate segment, multiple independent segments location probability evidence can be generated by combining fault feature sets.

[0100] S106: The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion processing results of each segment.

[0101] Among them, DS evidence theory is a reasoning method for dealing with uncertain information. It achieves the fusion of multiple types of evidence by constructing a basic probability allocation function and a synthesis rule. The fusion processing result refers to the comprehensive failure probability of each segment after the fusion of multiple types of evidence.

[0102] It should be noted that the DS evidence theory is a commonly used method in fault diagnosis to handle conflicts of multiple sources of evidence and improve the reliability of decision-making. The implementation methods include defining an identification framework, constructing a basic probability allocation, evidence synthesis, and decision-making process.

[0103] Optionally, a basic probability allocation is constructed for each type of location probability evidence, and the comprehensive failure probability of each segment is calculated by fusing the DS synthesis rules. The segment that meets the preset conditions is determined as the final failure segment.

[0104] This embodiment provides a comprehensive fault diagnosis method for active distribution networks based on a fusion algorithm. The method first collects multi-source monitoring data from the active distribution network. Then, it eliminates abnormal data and dimensional differences through data preprocessing, and employs a weighted data layer fusion strategy positively correlated with the historical reliability of the data source to effectively suppress the impact of information distortion, generating a complete fusion dataset with significantly improved data integrity and reliability. Subsequently, fault features are extracted from the fusion data to strengthen the expression of weak fault features, solving the problem of unclear fault features in active distribution networks. Based on this, candidate fault sections are screened by combining multi-source monitoring data and fault features, and multiple types of independent location probability evidence are generated. This not only avoids the limitations of a single algorithm through multi-dimensional evidence construction but also significantly reduces computational complexity through initial screening of candidate sections. Finally, the DS evidence theory is used to fuse multiple types of independent evidence for decision-making, fully leveraging its advantages in handling uncertain information and conflicting evidence. This achieves accurate fault location under conditions of bidirectional power flow and incomplete information, comprehensively improving the fault diagnosis method's fault tolerance, accuracy, and scenario adaptability, meeting the requirements for the safe and stable operation of active distribution networks.

[0105] In one exemplary embodiment, multi-source monitoring data is preprocessed to obtain a standardized monitoring dataset, including:

[0106] S201: Perform outlier removal processing on the multi-source monitoring data to obtain the first multi-source monitoring data.

[0107] Among them, outlier removal refers to the operation of identifying and removing distorted data that exceeds the normal data distribution range; the first multi-source monitoring data refers to the multi-source monitoring data after outlier removal.

[0108] For example, the 3σ criterion is used to calculate the mean and standard deviation of data from each data source. Data that exceed the mean ± 3 times the standard deviation are identified as outliers and removed. Median filtering is applied to the μPMU transient data to further suppress impulse noise interference, thus obtaining the first multi-source monitoring data.

[0109] S202: Fill in missing values ​​in the first multi-source monitoring data to obtain the second multi-source monitoring data.

[0110] Among them, missing value filling refers to the operation of filling in the missing parts of the data; the second multi-source monitoring data refers to the multi-source monitoring data after filling in the missing values.

[0111] For example, when no more than 5 consecutive data points are missing, linear interpolation is used to fill the missing data; when more than 5 consecutive data points are missing, the filling is optimized by combining the trend characteristics of similar data in adjacent time periods; when multiple data sources are missing at the same time, grey relational analysis is used to calculate the fill value to obtain the second multi-source monitoring data.

[0112] S203: Standardize the second multi-source monitoring data to obtain a standardized monitoring dataset.

[0113] Standardization refers to the operation of eliminating the differences in the units of different data sources and mapping the data to a unified range; standardized monitoring datasets refer to multi-source monitoring data after standardization.

[0114] For example, the Z-score standardization method is used for continuous data, and the units are unified by the formula z=(x-σ) / μ (where μ is the data mean and σ is the data standard deviation); the status identification data (0 / 1 / -1) of FTU and TTU are normalized separately and mapped to the interval [0,1] to obtain the standardized monitoring dataset.

[0115] This embodiment effectively eliminates noise, missing values, and dimensional differences in multi-source monitoring data through phased outlier removal, missing value filling, and standardization.

[0116] In one exemplary embodiment, a data layer fusion process is performed on the standardized monitoring dataset to generate a complete fused dataset, including:

[0117] S301: Calculate the credibility of each data source based on its historical operational data.

[0118] Among them, data source credibility refers to the probability of no distortion in the data source based on the statistics of historical operating data, reflecting the reliability level of the data source.

[0119] For example, the number of historical distorted data points (false positives, false negatives, and data anomalies) for each data source is counted against the total number of data collections. The reliability value of each data source is then calculated using the following formula:

[0120] Credibility = 1 - (Number of distortions / Total number of data collections);

[0121] The reliability value ranges from [0,1], with a higher value indicating a stronger reliability of the data source.

[0122] S302: Assign fusion weights to each data source based on credibility; fusion weights are positively correlated with credibility.

[0123] Among them, the fusion weight refers to the weighting coefficient of each data source when the data layer is fused, and it is positively correlated with the credibility of the data source.

[0124] For example, based on the credibility values ​​of each data source, the fusion weight (i.e., dynamic weight) of each data source is calculated using the following formula:

[0125] Dynamic weight = credibility / Σ credibility of all data sources;

[0126] This formula ensures that high-confidence data dominates the fusion process, reducing the impact of low-confidence data on the fusion results.

[0127] S303: Based on the fusion weight, the standardized monitoring dataset is weighted and fused, and missing data is filled in to obtain a complete fused dataset.

[0128] Among them, weighted fusion refers to the operation of weighting the data from each data source according to the fusion weight; complete fusion dataset refers to the fusion result of multi-source data without missing data and with high reliability.

[0129] For example, the standardized monitoring dataset is weighted and fused according to the calculated fusion weights; when a single data source is missing data, the K-nearest neighbor algorithm (K=5) is used to retrieve similar data under similar working conditions for weighted completion; when multiple data sources are missing at the same time, a completion model is constructed based on grey relational analysis, the correlation between data is calculated and completion values ​​are generated, and finally a complete fused dataset with no missing data and high reliability is obtained.

[0130] This embodiment achieves effective fusion of multi-source data through credibility assessment, dynamic weight allocation, and missing data completion, thus solving the problem of unreliable data caused by distortion or missing data from a single data source.

[0131] In one exemplary embodiment, feature extraction is performed on the complete fused dataset to obtain a fault feature set, including:

[0132] S401: Extract initial fault features from the complete fused dataset.

[0133] The initial fault features refer to the set of time-domain / frequency-domain features that may reflect fault information, which are directly extracted from the complete fusion dataset.

[0134] For example, three core initial fault features are extracted from the complete fused dataset:

[0135] (1) The phase differential current is calculated by the current difference between the two ends of the section, reflecting the current imbalance in the section. The calculation formula is as follows:

[0136] ;

[0137] In the formula, For the current at the beginning of the section, This is the current at the end of the section.

[0138] (2) Node voltage deviation characteristics are calculated by the ratio of the voltage difference before and after the fault to the rated voltage, reflecting the degree of node voltage abnormality. The calculation formula is as follows:

[0139] ;

[0140] In the formula, This is the actual voltage. This is the rated voltage.

[0141] (3) The ratio of high-frequency to low-frequency energy of the line is calculated by decomposing the transient electrical quantity into high-frequency (1-5kHz) and low-frequency (0-1kHz) energy through wavelet packet decomposition, and the ratio of the two is used to reflect the frequency distribution characteristics of the fault transient signal. The calculation formula is as follows:

[0142] ;

[0143] In the formula, For high-frequency energy, It is low-frequency energy.

[0144] S402: Redundant feature removal is performed on the initial fault features, and a fault feature set is obtained based on the fault features after the removal process.

[0145] Among them, redundant feature removal refers to the operation of removing features that are irrelevant to or highly relevant to fault diagnosis; the fault feature set refers to the set of core features that contribute highly to fault diagnosis after removing redundant features.

[0146] For example, firstly, the Pearson correlation coefficient between each initial fault feature is calculated, and highly correlated redundant features with an absolute value of correlation coefficient > 0.8 are removed; then, the mutual information entropy between each feature and the fault type is calculated, and effective features with mutual information entropy > 0.6 are retained, finally forming a core fault feature set that includes phase differential current, node voltage offset features, and the high- and low-frequency energy ratio of the line.

[0147] This embodiment achieves effective fusion and dimensionality reduction of fault features through initial fault feature extraction and redundant feature removal, retaining the core features that reflect the essence of the fault while eliminating redundant information.

[0148] In an exemplary embodiment, based on multi-source monitoring data and a fault feature set, a set of candidate fault segments and corresponding location probability evidence for each segment are determined, including:

[0149] S501: Based on multi-source monitoring data, the fault section is screened to obtain a set of candidate fault sections.

[0150] Among them, fault segment screening refers to the operation of initially narrowing down the fault range based on the correlation of multi-source monitoring data; candidate fault segment set refers to a set of several segments that may fail after screening.

[0151] For example, fault segment selection is performed based on the correlation between FTU overcurrent alarm signals and TTU undervoltage signals. If a segment meets the criteria of "upstream FTU detecting an overcurrent signal" and "downstream TTU detecting an undervoltage signal," then that segment is included in the candidate fault segment set, ultimately outputting multiple candidate fault segments to narrow down the scope of subsequent diagnosis. Preferably, the number of segments in the candidate fault segment set is 1-3.

[0152] S502: For each fault segment in the candidate fault segment set, and in conjunction with the fault feature set, different evidence generation algorithms are used to perform multi-evidence generation processing to obtain multi-class location probability evidence for each segment; the evidence generation algorithm is an algorithm that uses the probability that each segment is a fault segment as evidence.

[0153] Among them, multi-evidence generation processing refers to the operation of generating independent evidence reflecting the failure probability of each section using different algorithms; location probability evidence refers to the probability information reflecting that each section is a failure section.

[0154] For example, for each candidate fault segment, three independent evidence generation algorithms are used to process it in combination with the fault feature set, so as to obtain three independent localization probability evidences for each segment.

[0155] This embodiment narrows the diagnostic scope by screening faulty sections and generates multiple independent sections location probability evidence to provide reliable input for DS evidence theory fusion diagnosis.

[0156] In an exemplary embodiment, multi-evidence generation processing is performed to obtain multiple types of location probability evidence corresponding to each segment, including:

[0157] S601: The first type of segment location probability evidence is generated by using an integer linear programming solution method based on an improved multiverse algorithm. The integer linear programming solution method based on the improved multiverse algorithm is an improved multiverse algorithm that introduces a chaotic initialization strategy and adopts a nonlinear adaptive universe expansion rate update formula. It is used to solve the fault location integer linear programming model constrained by power flow conservation and signal coverage.

[0158] Among them, the IMVO (Improved Multi-Verse Optimizer) integer linear programming solution method refers to the method of solving the fault location integer linear programming model based on the improved multi-verse algorithm. The fault location integer linear programming model is an optimization model with the decision variables of whether the section is faulty and whether the monitoring point signal is effective, the objective function of minimizing the signal distortion rate and the power flow deviation, and the constraints of power flow conservation, DG output, and signal coverage.

[0159] For example, first construct an integer linear programming model for fault location, with x as the decision variable. j(Whether candidate segment j is faulty, 1 = faulty, 0 = normal), y k (Whether the signal at monitoring point k is valid, 1 = valid, 0 = distorted), the objective function is:

[0160] ;

[0161] In the formula, and The weights are denoted by , and the constraints include power flow conservation constraints, DG output constraints, and signal coverage constraints.

[0162] The improved multiverse algorithm (IMVO) was then used to solve the model. The improvement strategies included:

[0163] (1) Introducing a Logistic chaotic mapping initialization strategy to generate an initial population improves population diversity and avoids initial clustering of solutions. The formula is:

[0164] ;

[0165] In the formula, For a given initialization speed, .

[0166] (2) The cosmological expansion rate is updated using a nonlinear adaptive cosmological expansion rate update formula to improve the algorithm's convergence performance. The formula is:

[0167] ;

[0168] In the formula, 'a' controls the expansion rate, 't' is the current iteration number, and 'T' is the maximum iteration number.

[0169] The termination condition is set as follows: when the change in the fitness function value over 10 consecutive iterations is less than 1 / 3. The algorithm terminates when the maximum number of iterations is reached. After solving the model using IMVO, the failure probability of each candidate segment is output as the first type of segment location probability evidence m1(•).

[0170] S602: Use the switch quantity statistical matching method to generate the second type of segment location probability evidence.

[0171] Among them, the switch quantity statistical matching method refers to the method of calculating the section fault probability based on the statistical matching relationship between FTU overcurrent data and TTU undervoltage data.

[0172] For example, based on the historical statistical patterns of FTU overcurrent alarm signals (forward overcurrent, reverse overcurrent, no overcurrent) and TTU undervoltage data, a matching model between the section fault probability and the switching signal is constructed; by using the currently monitored FTU overcurrent data and TTU undervoltage data, the fault probability of each candidate section is statistically matched as the second type of section location probability evidence m2(•).

[0173] S603: Use the feature correlation matching method to generate the third type of segment location probability evidence.

[0174] Among them, the feature correlation matching method refers to the method of calculating the probability of section faults based on the correlation between the fault feature set and the fault status of each section.

[0175] For example, based on three core fault characteristics—phase differential current, node voltage offset, and line high- and low-frequency energy ratio—a correlation matching model is constructed between the characteristics and the fault status of each section. By calculating the correlation coefficient between the current fault characteristics and the fault status of each section, the association probability of each candidate section is obtained, which serves as the evidence m3(•) for the third type of section location probability.

[0176] This embodiment uses three independent evidence generation algorithms to generate segment location probability evidence from different dimensions, realizing a multi-perspective expression of fault information and solving the problems of incomplete and easily interfered single evidence information.

[0177] In an exemplary embodiment, the DS evidence theory is used to fuse the probabilistic evidence of each segment, and the final faulty segment is determined based on the fusion results of each segment, including:

[0178] S701: For each fault segment in the candidate fault segment set, determine the basic probability allocation corresponding to each type of localization probability evidence to obtain multiple basic probability allocations.

[0179] In this context, Basic Probability Assignment (BPA) refers to the degree of confidence in the probability assigned to each element within the identification frame in DS evidence theory; the identification frame refers to the mutually exclusive set consisting of all candidate fault segments. (N is the total number of distribution network segments).

[0180] For example, first define the recognition framework. The elements within the framework are all candidate fault segments, and the elements are mutually exclusive. Then, for each candidate fault segment, the location probability evidence of the first, second, and third types of segments is mapped to the corresponding basic probability allocation functions m1(•), m2(•), and m3(•), respectively, to obtain the three types of basic probability allocations.

[0181] S702: The DS synthesis rule is used to perform fusion calculation on the allocation of multiple basic probabilities to obtain the comprehensive failure probability of each section.

[0182] Among them, the DS synthesis rule refers to the formula in DS evidence theory used to fuse multiple basic probability assignments. Its core is to obtain a comprehensive probability assignment through intersection operation and conflict coefficient normalization.

[0183] For example, the DS synthesis rule is used to fuse the three basic probability assignments, and the formula is as follows:

[0184] ;

[0185] In the formula, A represents the faulty segment within the identification framework, B, C, and D are the focal elements corresponding to the three types of evidence, the numerator is the sum of the probabilities of the intersection of the three types of evidence being A, and the denominator is the normalized term of the conflict coefficient (i.e., the sum of the probabilities of the conflicting parts of the three types of evidence). The comprehensive fault probability of each segment is calculated using this formula. .

[0186] S703: The section whose overall failure probability meets the preset conditions is determined as the final failure section.

[0187] Among them, the preset conditions refer to the fault judgment threshold set according to the magnitude of the comprehensive fault probability; the final fault section refers to the distribution network section that is determined to have a fault.

[0188] For example, in the case of a single fault, the segment with the largest m(A) after fusion can be selected as the fault segment. For complex faults, a threshold needs to be set; values ​​exceeding a certain threshold are directly output, while values ​​within a certain range are subject to secondary verification using core features from the feature layer to ensure diagnostic reliability. For instance, the comprehensive fault probability of each segment can be used as a basis for further analysis. The final faulty section is determined as follows:

[0189] (1) If This section was directly identified as the final fault section.

[0190] (2) If The determination is made after secondary verification based on core characteristics such as excessive phase differential current and voltage deviation.

[0191] (3) If Return to the candidate segment screening step and re-execute to finally determine the faulty segment and diagnostic confidence, and output the complete diagnostic results.

[0192] This embodiment effectively solves the problems of information conflict and strong uncertainty in active power distribution network fault diagnosis by integrating multiple types of location probability evidence through DS evidence theory, and achieves accurate determination of fault sections.

[0193] This invention provides a comprehensive fault diagnosis method for active distribution networks based on a fusion algorithm. First, it collects multi-source monitoring data, including FTU overcurrent alarms, TTU undervoltage, μPMU transient electrical quantities, distributed generation output, and distribution network topology parameters. Data preprocessing is completed through anomaly removal, missing data filling, and dimensional standardization. Second, a reliability assessment model is constructed based on the historical distortion probability of the data sources. Dynamic fusion weights are assigned according to reliability, and missing data is filled using the K-nearest neighbor algorithm and grey relational analysis. The complementary information from the multi-source data is integrated to generate a highly reliable and complete fusion dataset, achieving data-level fusion processing. Finally, the phase differential current is extracted from the fusion data. This method identifies three core fault characteristics: node voltage deviation, line high-frequency / low-frequency energy ratio, etc. Redundant features are eliminated using Pearson correlation coefficient and mutual information entropy, resulting in a simplified set of effective fault features for feature-level fusion processing. Next, a small number of candidate fault sections are screened based on multi-source signal correlation. Then, three independent types of fault location probability evidence are generated through integer linear programming (using an improved multiverse algorithm), statistical matching of switch data, and correlation matching of core features. Finally, DS evidence theory synthesis rules are applied to complete multi-evidence fusion. Based on the fused comprehensive fault probability classification, the fault section and diagnostic confidence are output, thus achieving decision-level fusion processing. This three-level fusion architecture adapts to the operating conditions of active distribution networks with bidirectional power flow, weak fault characteristics, and easily distorted monitoring information, effectively improving fault diagnosis's fault tolerance, location accuracy, and adaptability to real-world scenarios.

[0194] Please see Figure 2 , Figure 2 The implementation flow of an active distribution network integrated fault diagnosis method based on a fusion algorithm designed according to the above embodiments is shown, including:

[0195] Step 1: Collect multi-source monitoring data of active power distribution network.

[0196] Data such as FTU, TTU, μPMU, DG, and topology parameters are obtained through multi-terminal and multi-dimensional data collection.

[0197] Step 2: Preprocess the collected data to obtain a standardized dataset.

[0198] By removing outliers, filling in missing values, and standardizing the data, noise, missing values, and dimensional differences are eliminated, and standardized data is generated.

[0199] Step 3: Data layer fusion and credibility assessment.

[0200] By integrating complementary information from multiple data sources through credibility assessment, dynamic weight allocation, and missing data completion, a complete and reliable fused dataset without missing data is generated, thus solving the problems of distortion and missing data from a single data source.

[0201] Step 4: Feature layer fusion extracts core fault features and derives different evidence weights.

[0202] By extracting initial fault features and removing redundant features, a core fault feature set reflecting the essence of the fault is obtained, and the weights of different pieces of evidence are determined based on the feature contribution.

[0203] Step 5: The decision-making level adopts a hierarchical diagnostic logic and combines DS evidence theory to output diagnostic results.

[0204] By screening candidate fault segments, generating multiple types of independent evidence, and fusing DS evidence, comprehensive reasoning based on multi-source evidence is achieved, ultimately outputting the fault segment and diagnostic confidence level, thus solving the problems of poor fault tolerance and insufficient scenario adaptability of traditional methods.

[0205] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0206] Based on the same inventive concept, this application also provides a device for implementing the above-mentioned active distribution network integrated fault diagnosis method based on fusion algorithm. The solution provided by this device is similar to the implementation scheme described in the above method. Therefore, the specific limitations in the embodiments of the active distribution network integrated fault diagnosis device based on fusion algorithm provided below can be found in the limitations of the active distribution network integrated fault diagnosis method based on fusion algorithm described above, and will not be repeated here.

[0207] Please see Figure 3 This invention also provides an active power distribution network integrated fault diagnosis device based on a fusion algorithm, comprising:

[0208] The data acquisition module is used to collect multi-source monitoring data from active power distribution networks;

[0209] The data preprocessing module is used to preprocess multi-source monitoring data to obtain a standardized monitoring dataset;

[0210] The data layer fusion module is used to perform data layer fusion processing on standardized monitoring datasets to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0211] The feature extraction module is used to extract features from the complete fused dataset to obtain a fault feature set;

[0212] The candidate fault segment determination module is used to determine the set of candidate fault segments and the corresponding location probability evidence for each segment based on multi-source monitoring data and fault feature set;

[0213] The final fault segment determination module is used to fuse the location probability evidence of each segment using DS evidence theory, and determine the final fault segment based on the fusion processing results of each segment.

[0214] In one exemplary embodiment, multi-source monitoring data is preprocessed to obtain a standardized monitoring dataset, including:

[0215] Outlier removal is performed on the multi-source monitoring data to obtain the first multi-source monitoring data.

[0216] Missing values ​​were filled into the first multi-source monitoring data to obtain the second multi-source monitoring data;

[0217] The second multi-source monitoring data is standardized to obtain a standardized monitoring dataset.

[0218] In one exemplary embodiment, a data layer fusion process is performed on the standardized monitoring dataset to generate a complete fused dataset, including:

[0219] The reliability of each data source is calculated based on its historical operational data.

[0220] Based on credibility, the fusion weights of each data source are assigned; the fusion weights are positively correlated with credibility.

[0221] The standardized monitoring dataset is weighted and fused based on the fusion weights, and missing data is filled in to obtain a complete fused dataset.

[0222] In one exemplary embodiment, feature extraction is performed on the complete fused dataset to obtain a fault feature set, including:

[0223] Extract initial fault features from the complete fused dataset;

[0224] Redundant features are removed from the initial fault features, and a fault feature set is obtained based on the removed fault features.

[0225] In an exemplary embodiment, based on multi-source monitoring data and a fault feature set, a set of candidate fault segments and corresponding location probability evidence for each segment are determined, including:

[0226] Based on multi-source monitoring data, fault sections are screened to obtain a set of candidate fault sections;

[0227] For each fault segment in the candidate fault segment set, different evidence generation algorithms are used to generate multiple evidence in combination with the fault feature set, so as to obtain multi-class localization probability evidence for each segment; the evidence generation algorithm is an algorithm that uses the probability that each segment is a fault segment as evidence.

[0228] In an exemplary embodiment, multi-evidence generation processing is performed to obtain multiple types of location probability evidence corresponding to each segment, including:

[0229] The IMVO integer linear programming method is used to generate the first type of segment location probability evidence. The IMVO integer linear programming method is an improved multiverse algorithm that introduces a chaotic initialization strategy and adopts a nonlinear adaptive universe expansion rate update formula. It is used to solve the fault location integer linear programming model constrained by power flow conservation and signal coverage.

[0230] The second type of segment location probability evidence is generated using the switch quantity statistical matching method;

[0231] The feature correlation matching method is used to generate the third type of segment location probability evidence.

[0232] In an exemplary embodiment, the DS evidence theory is used to fuse the probabilistic evidence of each segment, and the final faulty segment is determined based on the fusion results of each segment, including:

[0233] For each fault segment in the candidate fault segment set, the basic probability allocation corresponding to each type of localization probability evidence is determined, resulting in multiple basic probability allocations.

[0234] The DS synthesis rule is used to fuse and calculate the basic probability assignments of multiple categories to obtain the comprehensive failure probability of each section;

[0235] The section whose overall failure probability meets the preset conditions is determined as the final failure section.

[0236] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0237] Reference Figure 4 This invention also provides a computer device, including: a memory and a processor, and a computer program stored in the memory. When the computer program is executed on the processor, it performs the following steps:

[0238] Collect multi-source monitoring data from the active power distribution network;

[0239] Multi-source monitoring data are preprocessed to obtain a standardized monitoring dataset;

[0240] The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to their corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0241] Feature extraction is performed on the complete fused dataset to obtain the fault feature set;

[0242] Based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined.

[0243] The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion results of each segment.

[0244] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 4The examples of computer devices are merely examples and do not constitute a limitation on computer devices. They may include more or fewer components than shown in the illustration, or combinations of certain components, or different components. For example, they may also include input / output devices, network access devices, etc.

[0245] The processor referred to can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0246] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard drive or RAM. In other embodiments, the memory may be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory may include both internal and external storage units of the computer device. The memory is used to store the operating system, applications, boot loader, data, and other programs, such as the program code of the computer program. The memory can also be used to temporarily store data that has been output or will be output.

[0247] This invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, performs the following steps:

[0248] Collect multi-source monitoring data from the active power distribution network;

[0249] Multi-source monitoring data are preprocessed to obtain a standardized monitoring dataset;

[0250] The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to their corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0251] Feature extraction is performed on the complete fused dataset to obtain the fault feature set;

[0252] Based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined.

[0253] The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion results of each segment.

[0254] In this embodiment, if the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0255] This invention provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0256] Collect multi-source monitoring data from the active power distribution network;

[0257] Multi-source monitoring data are preprocessed to obtain a standardized monitoring dataset;

[0258] The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset. The data layer fusion processing is an operation that merges monitoring data from different data sources according to their corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source.

[0259] Feature extraction is performed on the complete fused dataset to obtain the fault feature set;

[0260] Based on multi-source monitoring data and fault feature sets, a set of candidate fault segments and corresponding location probability evidence for each segment are determined.

[0261] The DS evidence theory is used to fuse the location probability evidence of each segment, and the final fault segment is determined based on the fusion results of each segment.

[0262] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0263] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0264] In the embodiments disclosed in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0265] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A comprehensive fault diagnosis method for active distribution networks based on a fusion algorithm, characterized in that, Includes the following steps: Collect multi-source monitoring data from the active power distribution network; The multi-source monitoring data is preprocessed to obtain a standardized monitoring dataset; The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset; the data layer fusion processing is an operation of fusing monitoring data from different data sources according to corresponding weights; The corresponding weights of each data source are positively correlated with the credibility of the data source; Feature extraction is performed on the complete fused dataset to obtain a fault feature set; Based on the multi-source monitoring data and the fault feature set, a set of candidate fault segments and the corresponding location probability evidence for each segment are determined. The location probability evidence of each segment is fused using the DS evidence theory, and the final fault segment is determined based on the fusion processing results of each segment.

2. The active distribution network integrated fault diagnosis method based on fusion algorithm according to claim 1, characterized in that, The multi-source monitoring data is preprocessed to obtain a standardized monitoring dataset, including: The outlier removal process is performed on the multi-source monitoring data to obtain the first multi-source monitoring data; Missing values ​​are filled into the first multi-source monitoring data to obtain the second multi-source monitoring data; The second multi-source monitoring data is standardized to obtain the standardized monitoring dataset.

3. The active distribution network integrated fault diagnosis method based on fusion algorithm according to claim 1, characterized in that, The standardized monitoring dataset is subjected to data layer fusion processing to generate a complete fused dataset, including: The reliability of each data source is calculated based on its historical operational data. Based on the credibility, a fusion weight is assigned to each data source; the fusion weight is positively correlated with the credibility. The standardized monitoring dataset is weighted and fused based on the fusion weights, and missing data is filled in to obtain the complete fused dataset.

4. The active distribution network integrated fault diagnosis method based on fusion algorithm according to claim 1, characterized in that, Feature extraction is performed on the complete fused dataset to obtain a fault feature set, including: Extract initial fault features from the complete fused dataset; The initial fault features are subjected to redundant feature removal processing, and the fault feature set is obtained based on the fault features after the removal processing.

5. The active distribution network integrated fault diagnosis method based on fusion algorithm according to claim 1, characterized in that, Based on the multi-source monitoring data and the fault feature set, a candidate fault segment set and corresponding location probability evidence for each segment are determined, including: Based on the multi-source monitoring data, fault sections are screened to obtain the candidate fault section set; For each fault segment in the candidate fault segment set, and in conjunction with the fault feature set, different evidence generation algorithms are used to perform multi-evidence generation processing to obtain multi-class location probability evidence for each segment; the evidence generation algorithm is an algorithm that uses the probability that each segment is a fault segment as evidence.

6. The active distribution network integrated fault diagnosis method based on fusion algorithm according to claim 5, characterized in that, Multi-evidence generation processing is performed to obtain multiple types of probabilistic location evidence for each segment, including: The first type of segment location probability evidence is generated by an integer linear programming solution method based on an improved multiverse algorithm. The integer linear programming solution method based on the improved multiverse algorithm is an improved multiverse algorithm that introduces a chaotic initialization strategy and adopts a nonlinear adaptive universe expansion rate update formula. It is used to solve the fault location integer linear programming model constrained by power flow conservation and signal coverage. The second type of segment location probability evidence is generated using the switch quantity statistical matching method; The feature correlation matching method is used to generate the third type of segment location probability evidence.

7. The active distribution network integrated fault diagnosis method based on fusion algorithm according to claim 1, characterized in that, The location probability evidence for each segment is fused using the DS evidence theory, and the final fault segment is determined based on the fusion results of each segment, including: For each fault segment in the candidate fault segment set, the basic probability allocation corresponding to each type of localization probability evidence is determined to obtain multiple basic probability allocations. The DS synthesis rule is used to fuse and calculate the multi-class basic probability allocation to obtain the comprehensive fault probability of each section; The segment whose overall failure probability meets the preset conditions is determined as the final failure segment.

8. An active power distribution network integrated fault diagnosis device based on a fusion algorithm, characterized in that, include: The data acquisition module is used to collect multi-source monitoring data from active power distribution networks; The data preprocessing module is used to preprocess the multi-source monitoring data to obtain a standardized monitoring dataset; The data layer fusion module is used to perform data layer fusion processing on the standardized monitoring dataset to generate a complete fused dataset. The data layer fusion processing is an operation of fusing monitoring data from different data sources according to corresponding weights. The corresponding weights of each data source are positively correlated with the reliability of the data source. The feature extraction module is used to extract features from the complete fused dataset to obtain a fault feature set; The candidate fault segment determination module is used to determine a set of candidate fault segments and the location probability evidence for each segment based on the multi-source monitoring data and the fault feature set. The final fault segment determination module is used to fuse the location probability evidence of each segment using DS evidence theory, and determine the final fault segment based on the fusion processing results of each segment.

9. A computer device, characterized in that, The device includes a processor and a memory: The memory is used to store computer programs and send the instructions of the computer programs to the processor; The processor executes, according to the instructions of the computer program, a method for comprehensive fault diagnosis of active power distribution networks based on a fusion algorithm as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for comprehensive fault diagnosis of active power distribution networks based on a fusion algorithm as described in any one of claims 1-7.

Images