New energy power distribution fault diagnosis system and method based on multi-source data fusion

Abstract

The invention provides a new energy power distribution fault diagnosis system and method based on multi-source data fusion, and relates to the technical field of power system fault diagnosis. The system comprises a data processing module, a fusion reasoning module, a graph structure diagnosis module and a control response module. The data processing module performs unified structured processing on the multi-source operation data to generate a feature tensor; the fusion reasoning module constructs a dynamic fusion weight based on the information entropy of each data source, and outputs a fusion confidence judgment result; the graph structure diagnosis module is combined with the distribution network topology to construct a graph structure model, and node-level fault positioning is realized through a graph neural network; and the control response module generates a control instruction based on the diagnosis result, and acquires and executes feedback to update graph structure attributes, thereby realizing a diagnosis closed loop. The method has the advantages of high diagnosis precision, strong adaptability, timely response and the like, and is suitable for intelligent fault identification and dynamic processing in a new energy power distribution scene.

Description

Technical Field

[0001] The present invention relates to the technical field of power system fault diagnosis, and in particular to a new energy distribution fault diagnosis system and method based on multi-source data fusion. Background Art

[0002] With the continuous increase in the penetration rate of new energy, distributed photovoltaic, wind power, energy storage and other new power sources are connected to the distribution network in large quantities. The structure and operating status of the distribution system are becoming increasingly complex. The traditional fault diagnosis method that relies on single-source measurement data and fixed model identification rules can no longer meet the "high real-time, high accuracy, and high robustness" fault location requirements in new energy scenarios.

[0003] In existing technologies, fault status identification is often performed using classification models based on single-channel data such as voltage and current. Some solutions introduce algorithms such as neural networks and decision trees to improve classification accuracy. However, due to the lack of cross-channel information fusion mechanisms, the following technical limitations still exist: Insufficient fusion of multi-source information: The data collected by various edge devices have large differences in dimensions and inconsistent sampling frequencies. Traditional feature-level splicing methods have difficulty achieving information alignment, which can easily lead to fusion failure or redundant interference; The fusion mechanism lacks adaptive capabilities: Most fusion algorithms have fixed weights or are set based solely on prior experience, and cannot dynamically reflect changes in the credibility of each data source. This leads to unstable fusion results, especially in scenarios with data anomalies or localized failures; Graph structures are not incorporated into the diagnostic closed loop: Most current distribution diagnosis methods based on graph neural networks are based on static graph structures and only use one-time graph modeling in the inference phase. Control feedback information is not incorporated into the graph state update process, lacking the ability for system evolution and strategy self-adjustment; Fault response is disconnected from diagnostic logic: In most existing solutions, diagnostic results are difficult to feedback to subsequent diagnostic mechanisms. System behavior is a linear "perception-judgment-execution" chain, and a learnable and evolvable closed-loop control system has not been formed.

[0004] Therefore, the current new energy distribution system urgently needs an intelligent fault diagnosis method and system structure that can integrate multi-source heterogeneous data, has information confidence adjustment capabilities, and can collaboratively evolve graph structure and control feedback, so as to improve the diagnostic intelligence level, fault tolerance and response efficiency in new energy scenarios. Summary of the Invention

[0005] In response to the problems in existing new energy distribution systems, such as the difficulty in fusing multi-source heterogeneous data, the lack of adaptive adjustment capability of the fusion mechanism, the disconnection between graph structure diagnosis and control feedback, and the inability of the system to evolve in a closed loop, the present invention provides a new energy distribution fault diagnosis system and method based on multi-source data fusion.

[0006] In order to solve the above problems, the present invention is implemented through the following technical solutions:

[0007] A new energy distribution fault diagnosis system based on multi-source data fusion, including:

[0008] A data processing module is used to receive multi-source operating data from the new energy distribution system and perform unified structured processing on the multi-source operating data, including normalization, standardization and multi-dimensional feature extraction, to generate corresponding feature tensors;

[0009] A fusion reasoning module is configured to receive the feature tensor, calculate an information entropy value based on the fault classification probability distribution of each data source within a preset time window, and construct a dynamic fusion weight vector based on the information entropy value. The dynamic fusion weight vector is used to guide the weighted fusion processing of multi-source confidence information and output a fusion confidence judgment result for each type of fault;

[0010] A graph structure diagnosis module is configured to receive the fusion confidence judgment result and construct a graph structure model based on the topology of the current power distribution network. The graph structure model includes nodes representing power distribution equipment and edges or structural attribute fields representing electrical connection relationships. The fusion result is mapped into embedded features of nodes in the graph, and node-level fault identification and location are performed through a graph neural network model including an attention mechanism, and the fault diagnosis result is output;

[0011] A control response module is used to receive the fault diagnosis results, generate corresponding isolation control instructions, and receive control execution feedback, and update the feedback information to the structural attribute fields of the nodes or edges in the graph structure model for dynamic correction of the graph structure in subsequent diagnosis cycles.

[0012] As a preferred solution of the present invention, the data processing module includes:

[0013] A data receiving unit is used to receive operating data from multiple edge acquisition terminals, including electrical quantities, state quantities, and environmental parameters, and align and encapsulate the operating data according to timestamps to construct a structured data input format;

[0014] a processing unit, configured to perform normalization processing based on Box-Cox transformation and Z-score standardization on the structured data;

[0015] The feature extraction unit is used to extract multidimensional features including RMS value, kurtosis, wavelet energy density, spectral center of gravity and frequency offset from each data channel within the sliding time window, and generate a feature tensor for subsequent processing.

[0016] As a preferred solution of the present invention, the fusion reasoning module includes:

[0017] A feature archiving unit, configured to receive the feature tensors, classify and archive the feature tensors according to data source identifiers, and construct a fusion input set;

[0018] Uncertainty quantification unit, used to determine the probability distribution P of faults based on each data source within the target time window i ={p i1 ,p i2 ,…,p iK}, calculate the information entropy value, the calculation formula is:

[0019]

[0020] Among them: H i is the information entropy value, i represents the data source number, k represents the fault type number, K is the total number of fault categories, and p ik represents the probability of the i-th data source distinguishing the k-th type of fault;

[0021] Entropy weight generation unit, used to construct dynamic fusion weight w according to the information entropy value i , the calculation formula is:

[0022]

[0023] Where: N is the total number of data sources involved in the fusion, j is the data source number in the fusion, H j The information entropy value calculated for the j-th data source;

[0024] Confidence fusion unit, used to combine the confidence evidence of each data source with the dynamic fusion weight w i Perform weighted synthesis and output the fusion confidence result corresponding to the fault type;

[0025] The fusion feedback adjustment unit is used to adaptively update the weight generation strategy or confidence threshold parameters in the subsequent time window based on the fusion confidence result.

[0026] As a preferred solution of the present invention, the fusion reasoning module further includes:

[0027] A feature compression unit is used to perform dimensionality compression processing on the received feature tensor before performing uncertainty quantization. The compression processing is based on the principal component analysis (PCA) algorithm. After constructing the sample covariance matrix, the first several principal components are extracted and low-order feature components whose cumulative contribution rate is lower than a preset threshold are eliminated.

[0028] The compressed tensor output by the feature compression unit serves as the input basis for subsequent information entropy value calculation and weight generation, and is used to improve the feature alignment and fusion consistency between multi-source data. The compression dimension can be dynamically adjusted according to the fusion feedback results.

[0029] As a preferred solution of the present invention, the graph structure diagnosis module includes:

[0030] Graph model building unit, used to build a graph structure model based on the topology of the current new energy distribution system and equipment connection information

[0031] G = (V, E), where the node set V represents the electrical equipment unit, and the edge set E represents the physical connection relationship. Each node and edge in the graph structure model is configured with a structural attribute field containing a state flag, device type, and control attribute;

[0032] A feature mapping unit is used to map the fusion confidence judgment result to the corresponding node of the graph structure model according to the node device number, and splice it with the historical state feature in the graph structure model to form an input feature vector;

[0033] A graph neural reasoning unit, configured to execute a graph neural network model with attention based on the graph structure model and node feature input to identify the fault and fault type label of each node;

[0034] The graph embedding output unit is used to jointly output the intermediate state embedding results of the graph neural network and the fault discrimination results to form a structured diagnosis tensor.

[0035] As a preferred solution of the present invention, the graph neural reasoning unit adopts a structural attribute-aware attention mechanism, and the attention weight coefficient between nodes is calculated according to the following formula:

[0036]

[0037] Where: α ij represents the attention weight coefficient assigned by the i-th node to its neighboring node j; h i , h j are the input feature vectors of nodes i and j respectively; W1 and W2 are trainable feature mapping matrices used to embed node features into a unified representation space; s i , s j are the structural state attribute vectors of nodes i and j, including device type, current fault status, and historical control records; φ(s i ,s j ) is the joint embedding function between node attributes, which is used to reflect the degree of association between structural states; is a trainable attention parameter vector; σ(·) is the activation mapping function; ‖ represents the vector concatenation operation; h k Represents adjacent nodes The input feature vector of is the set of adjacent nodes of node i.

[0038] As a preferred solution of the present invention, the attention graph neural network model further includes the following technical improvement modules:

[0039] The attribute-driven fine-tuning module is used to dynamically fine-tune the attention weight according to the differences in the node state attribute vectors in the graph structure during the attention coefficient calculation process;

[0040] The policy control module is used to adjust the scale factor in the normalization process according to the degree distribution characteristics of the nodes in the current graph structure;

[0041] The connection suppression module is used to prune the edge connections corresponding to the attention weights below the set threshold.

[0042] As a preferred solution of the present invention, the control response module includes:

[0043] A response instruction generating unit is used to extract the control attributes of the target node and the associated edges according to the fault diagnosis result, and generate an isolation control instruction including a control target identifier, an operation type, a priority, and execution strategy parameters;

[0044] An instruction issuing and executing unit, configured to issue the isolation control instruction to an execution terminal of a corresponding power distribution device and trigger corresponding switchgear actions, including opening and closing, load switching or feeder isolation control operations;

[0045] An execution status acquisition unit is used to collect operation feedback information of the execution terminal in real time, the feedback information including the actual status of the switch, the action completion flag, the response time and the abnormal execution flag;

[0046] The graph attribute update unit is used to map and update the operation feedback information to the state structure attribute field of the corresponding node or edge in the graph structure model, including the row status mark, execution success flag and response delay value, for dynamic adjustment and structural correction of the fusion strategy in the subsequent graph neural network diagnosis process.

[0047] As a preferred solution of the present invention, the control response module further includes:

[0048] A strategy adjustment unit is used to dynamically adjust the information entropy value threshold parameter, fusion weight generation strategy or node state confidence threshold in the fusion reasoning module according to the state parameters, response delay or execution abnormality flag contained in the control execution feedback;

[0049] The strategy adjustment unit constructs a fusion mechanism adaptive path driven by feedback information, so that the system can strategically optimize the fusion accuracy according to historical control behavior in the next diagnosis cycle.

[0050] A new energy distribution fault diagnosis method based on multi-source data fusion, using the new energy distribution fault diagnosis system based on multi-source data fusion as described above, the method comprises the following steps:

[0051] Step S1: receiving multi-source operating data from a new energy distribution system, including voltage, current, load power, and environmental state parameters; performing structured sampling on the multi-source operating data in a time series, completing unified format packaging, and generating a multi-source feature tensor through normalization, standardization, and feature extraction;

[0052] Step S2: performing principal component analysis on the feature tensor to remove redundant feature components whose cumulative contribution rate is lower than a preset threshold, thereby forming a compressed feature tensor;

[0053] Step S3: Construct the fault classification probability distribution within the time window corresponding to each data source, calculate the information entropy value, and construct the dynamic fusion weight based on the information entropy value. The fusion reasoning module performs weighted synthesis processing of the confidence information according to the weight and outputs the fusion confidence judgment result;

[0054] Step S4: Build a graph structure model by combining the topological structure relationship of the new energy distribution network and the device connection information, map the fusion confidence result into the feature vector of the node in the graph structure, perform node-level aggregation and reasoning through the structural attribute-aware attention graph neural network model, and generate the node embedding vector and the corresponding fault type label;

[0055] Step S5: generating an isolation control instruction including a control target identifier, an operation type, a priority, and policy parameters based on the diagnostic tag and the node control attribute, and sending the isolation control instruction to the corresponding power distribution equipment terminal for execution;

[0056] Step S6: collecting the terminal execution feedback information, including the action completion status, response time and abnormal flag, and writing the feedback result into the structural attribute field of the corresponding node or edge in the graph structure model;

[0057] Step S7: According to the updated graph structure state, dynamically adjust the information entropy value threshold parameter, fusion weight generation strategy or confidence judgment threshold to achieve adaptive evolution of the diagnosis strategy and form a closed-loop intelligent fault diagnosis and control mechanism.

[0058] The beneficial effects of the present invention are: the present invention has many advantages such as strong data fusion capability, high fault identification accuracy, and good response control closed-loop. The system uses the data processing module to perform unified structured processing on the multi-source operating data from the new energy distribution system, completes normalization, standardization and multi-dimensional feature extraction, and generates a feature tensor with good consistency, which effectively solves the problems of heterogeneity and time series asynchrony of multi-source data. The fusion reasoning module uses information entropy to measure the discrimination uncertainty of each data source, dynamically constructs fusion weights, guides the weighted fusion of multi-source confidence information, and improves the accuracy and robustness of the fusion results. The graph structure diagnosis module constructs a graph structure model based on the current distribution network topology, and maps the fusion results into node embedding features. It combines the attention mechanism graph neural network to realize node-level fault identification and positioning, and enhances the system's perception and diagnosis capabilities of complex electrical connection relationships. The control response module generates isolation control instructions based on the fault diagnosis results and issues them for execution. At the same time, it collects execution feedback information to dynamically update the structural attribute fields of the nodes or edges of the graph structure model, so that the graph structure evolves in real time with the control behavior, and forms a closed-loop mechanism of diagnosis-control-feedback-re-diagnosis, thereby significantly improving the intelligence level and adaptability of the system, and having good engineering feasibility and application promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0060] in:

[0061] Figure 1 Schematic diagram of the overall structure of the system of the present invention;

[0062] Figure 2 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0064] like Figure 1 FIG. 1 is an embodiment of the present invention, which provides a new energy distribution fault diagnosis system based on multi-source data fusion, including:

[0065] (1) Data processing module

[0066] Used to receive multi-source operating data from the new energy distribution system and perform unified structural processing on the multi-source operating data, including normalization, standardization and multi-dimensional feature extraction, to generate corresponding feature tensors;

[0067] The data processing module includes:

[0068] A data receiving unit is used to receive operating data from multiple edge acquisition terminals, including electrical quantities, state quantities, and environmental parameters, and align and encapsulate the operating data according to timestamps to construct a structured data input format;

[0069] a processing unit, configured to perform normalization processing based on Box-Cox transformation and Z-score standardization on the structured data;

[0070] The feature extraction unit is used to extract multidimensional features including RMS value, kurtosis, wavelet energy density, spectral center of gravity and frequency offset from each data channel within the sliding time window, and generate a feature tensor for subsequent processing.

[0071] In specific implementations, the feature extraction unit can further integrate an edge recognition strategy module to eliminate or weaken the weight of channels with abnormal data quality, thereby preventing low-confidence inputs from interfering with subsequent fusion models. At the same time, the sliding time window length and step size parameters can be dynamically set based on the actual power distribution operation characteristics. For example, the time window width can be adjusted based on the fault response sensitive interval to improve diagnostic accuracy while ensuring real-time performance. The generation of feature tensors supports a tensor parallel encapsulation mechanism, facilitating direct connection with the data interface of downstream fusion modules, reducing system memory load and latency pressure.

[0072] (2) Fusion Reasoning Module

[0073] The system is used to receive the feature tensor, calculate the information entropy value based on the fault classification probability distribution of each data source within a preset time window, and construct a dynamic fusion weight vector based on the information entropy value. The dynamic fusion weight vector is used to guide the weighted fusion processing of multi-source confidence information and output the fusion confidence judgment result of each type of fault;

[0074] In a preferred embodiment of the present invention, a fusion inference module is used to complete the entire process, from information credibility measurement to fusion discriminant output, based on data feature tensors from different sources. This module's implementation primarily includes four key subunits: uncertainty quantification, weight generation, confidence fusion, and feedback adjustment. Each unit has clear technical responsibilities and interactive data structures, ensuring high adaptability and robustness of fusion decisions.

[0075] The fusion reasoning module includes:

[0076] A feature archiving unit, configured to receive the feature tensors, classify and archive the feature tensors according to data source identifiers, and construct a fusion input set;

[0077] Uncertainty quantification unit, used to determine the probability distribution P of faults based on each data source within the target time window i ={p i1 ,p i2 ,…,p iK}, calculate the information entropy value, the calculation formula is:

[0078]

[0079] Among them: H i is the information entropy value, i represents the data source number, k represents the fault type number, K is the total number of fault categories, and p ik represents the probability of the i-th data source distinguishing the k-th type of fault;

[0080] It should be noted that, unlike the use of fixed fusion weights or data confidence priors in existing technologies, this step introduces information entropy as the basis for weight generation for the first time, and has real-time, in-source self-feedback capabilities.

[0081] Entropy weight generation unit, used to construct dynamic fusion weight w according to the information entropy value i , the calculation formula is:

[0082]

[0083] Where: N is the total number of data sources involved in the fusion, j is the data source number in the fusion, H j The information entropy value calculated for the j-th data source;

[0084] It should be noted that compared with traditional linear averaging and hard rule screening strategies, entropy-driven normalized weights have the characteristics of dynamic, interpretable, and adjustable, and are adaptable to various data quality states.

[0085] Confidence fusion unit, used to combine the confidence evidence of each data source with the dynamic fusion weight w i Perform weighted synthesis and output the fusion confidence result corresponding to the fault type;

[0086] The fusion feedback adjustment unit is used to adaptively update the weight generation strategy or confidence threshold parameters in the subsequent time window based on the fusion confidence result.

[0087] In this embodiment, the fusion feedback adjustment unit is responsible for monitoring the change trend of the fusion output for multiple consecutive cycles. If it is found that the entropy value of a certain data source is higher than the mean for a long time or its weight fluctuates greatly, the upper and lower limits of the participation weight of the corresponding channel will be automatically adjusted or the "soft elimination" mechanism will be enabled to temporarily freeze the data source and replace it with the historical average.

[0088] At the same time, the entropy fluctuation trend is also used to drive subsequent window step size adjustments. For example, when the fusion entropy jitter increases, the system automatically shortens the time window, improving time series resolution and enhancing diagnostic sensitivity.

[0089] This system realizes for the first time a closed-loop control path of information uncertainty, weight adjustment, confidence judgment, and feedback callback. It not only has static fusion capabilities, but also has self-evolutionary adaptability.

[0090] The fusion reasoning module further includes:

[0091] A feature compression unit is used to perform dimensionality compression processing on the received feature tensor before performing uncertainty quantization. The compression processing is based on the principal component analysis (PCA) algorithm. After constructing the sample covariance matrix, the first several principal components are extracted and low-order feature components whose cumulative contribution rate is lower than a preset threshold are eliminated.

[0092] The compressed tensor output by the feature compression unit serves as the input basis for subsequent information entropy value calculation and weight generation, and is used to improve the feature alignment and fusion consistency between multi-source data. The compression dimension can be dynamically adjusted according to the fusion feedback results.

[0093] In practice, to enhance the fusion inference module's adaptability to heterogeneous data from multiple scenarios and with varying qualities, the system introduces a dynamic weight adjustment window mechanism driven by information entropy. This allows for temporary adjustments to the fusion contribution of the current data source based on changes in fault burst intensity or data loss. Specifically, if a data channel exhibits significant drift, distortion, or mutation, its corresponding entropy value is explicitly suppressed in the weight calculation, ensuring that the fusion result is not affected by local distortion.

[0094] In addition, the entropy normalization weight calculation unit supports an interpolation and backtracking correction mechanism. Even if a certain type of data is missing or unreliable in the current time window, it can be estimated through the historical entropy weight trend of adjacent time windows, thereby improving the system's continuity and fault tolerance.

[0095] The fusion input interface utilizes a tensor formatting cache, allowing the compressed principal component tensor to be directly input into the subsequent graph structure diagnosis module, avoiding redundant intermediate processing steps and reducing system response latency. This structure also facilitates rapid migration and deployment of the system across different renewable energy distribution edge node platforms, enhancing the feasibility of this invention in real-world industrial environments.

[0096] (3) Graph structure diagnosis module

[0097] The system is used to receive the fusion confidence judgment result and construct a graph structure model in combination with the topology of the current power distribution network. The graph structure model includes nodes representing power distribution equipment and edges or structural attribute fields representing electrical connection relationships. The fusion result is mapped to the embedded features of the nodes in the graph, and node-level fault identification and location are performed through a graph neural network model including an attention mechanism, and the fault diagnosis result is output;

[0098] The graph structure diagnosis module includes:

[0099] Graph model building unit, used to build a graph structure model based on the topology of the current new energy distribution system and equipment connection information

[0100] G = (V, E), where the node set V represents the electrical equipment unit, and the edge set E represents the physical connection relationship. Each node and edge in the graph structure model is configured with a structural attribute field containing a state flag, device type, and control attribute;

[0101] A feature mapping unit is used to map the fusion confidence judgment result to the corresponding node of the graph structure model according to the node device number, and splice it with the historical state feature in the graph structure model to form an input feature vector;

[0102] A graph neural reasoning unit, configured to execute a graph neural network model with attention based on the graph structure model and node feature input to identify the fault and fault type label of each node;

[0103] The graph embedding output unit is used to jointly output the intermediate state embedding results of the graph neural network and the fault discrimination results to form a structured diagnosis tensor.

[0104] The graph neural reasoning unit adopts a structural attribute-aware attention mechanism, and the attention weight coefficient between nodes is calculated according to the following formula:

[0105]

[0106] Where: α ij represents the attention weight coefficient assigned by the i-th node to its neighboring node j; h i , h j are the input feature vectors of nodes i and j respectively; W1 and W2 are trainable feature mapping matrices used to embed node features into a unified representation space; s i , s j are the structural state attribute vectors of nodes i and j, including device type, current fault status, and historical control records; φ(s i ,s j ) is the joint embedding function between node attributes, which is used to reflect the degree of association between structural states; is a trainable attention parameter vector; σ(·) is the activation mapping function; ‖ represents the vector concatenation operation; h k Represents adjacent nodes The input feature vector of is the set of adjacent nodes of node i.

[0107] The attention graph neural network model further includes the following technical improvement modules:

[0108] The attribute-driven fine-tuning module is used to dynamically fine-tune the attention weight according to the differences in the node state attribute vectors in the graph structure during the attention coefficient calculation process;

[0109] The policy control module is used to adjust the scale factor in the normalization process according to the degree distribution characteristics of the nodes in the current graph structure;

[0110] The connection suppression module is used to prune the edge connections corresponding to the attention weights below the set threshold.

[0111] In a preferred embodiment, to further enhance the upstream and downstream perception linkage of fault location, the graph structure model supports the dynamic evolution of structural attributes. Specifically, after each diagnostic cycle, the system updates node attribute states based on fusion confidence results, automatically triggering a local diffusion propagation mechanism for the corresponding graph edge attributes. This mechanism uses a propagation coefficient adjustment method based on edge weight control, allowing confidence information to propagate only within areas of strong fusion consistency, thus preventing error amplification caused by invalid edges.

[0112] In addition, to support distributed deployment scenarios, the internal neural network parameters of this module support heterogeneous graph partition training. The embedded features and weight parameters in each local subgraph can be synchronized to the main graph model after local convergence, forming an efficient edge deployment architecture; the system also has a graph caching mechanism. In scenarios where the topology changes slowly, there is no need to reconstruct the graph model in each round. Only the dynamically updated attribute fields in the graph are differentially updated, thereby significantly reducing the overall graph modeling overhead of the diagnosis cycle.

[0113] (4) Control response module

[0114] It is used to receive the fault diagnosis result, generate corresponding isolation control instructions, and receive control execution feedback, and update the feedback information to the structural attribute field of the node or edge in the graph structure model for dynamic correction of the graph structure in subsequent diagnosis cycles.

[0115] The control response module includes:

[0116] A response instruction generating unit is used to extract the control attributes of the target node and the associated edges according to the fault diagnosis result, and generate an isolation control instruction including a control target identifier, an operation type, a priority, and execution strategy parameters;

[0117] An instruction issuing and executing unit, configured to issue the isolation control instruction to an execution terminal of a corresponding power distribution device and trigger corresponding switchgear actions, including opening and closing, load switching or feeder isolation control operations;

[0118] An execution status acquisition unit is used to collect operation feedback information of the execution terminal in real time, the feedback information including the actual status of the switch, the action completion flag, the response time and the abnormal execution flag;

[0119] The graph attribute update unit is used to map and update the operation feedback information to the state structure attribute field of the corresponding node or edge in the graph structure model, including the row status mark, execution success flag and response delay value, for dynamic adjustment and structural correction of the fusion strategy in the subsequent graph neural network diagnosis process.

[0120] The control response module further includes:

[0121] A strategy adjustment unit is used to dynamically adjust the information entropy value threshold parameter, fusion weight generation strategy or node state confidence threshold in the fusion reasoning module according to the state parameters, response delay or execution abnormality flag contained in the control execution feedback;

[0122] The strategy adjustment unit constructs a fusion mechanism adaptive path driven by feedback information, so that the system can strategically optimize the fusion accuracy according to historical control behavior in the next diagnosis cycle.

[0123] In this embodiment of the present invention, the control response module serves as the terminal decision-making execution layer of the diagnostic chain. Its function is not only to responsively execute isolation operations but also to continuously optimize system integration strategies through feedback information. In actual deployment, a clear data transmission and status linkage mechanism is established between response instruction generation, execution, and feedback collection. Each unit achieves near-real-time closed-loop operation through a message queue or control bus.

[0124] After being mapped to the graph structure attribute field, the feedback information will be embedded as part of the input of the graph neural network in the next diagnostic cycle, thereby affecting the node representation learning process; after the feedback is triggered, the strategy adjustment unit indirectly affects the weight generation and confidence calculation path in the fusion reasoning module by modifying the form of the fusion parameters, so that the system fusion results can gradually adapt to environmental disturbances and control behavior deviations.

[0125] Through the dynamic coordination of the above-mentioned structure and data flow, the present invention realizes the multi-stage linkage logic of fusion judgment - graph structure mapping - control response - strategy adjustment, builds a stable and highly responsive self-regulating mechanism for the diagnostic system, and ensures that fault identification and control execution form a logical closed loop.

[0126] like Figure 2 FIG. 1 is another embodiment of the present invention, which provides a method for diagnosing a new energy distribution fault based on multi-source data fusion. The method uses the above-mentioned new energy distribution fault diagnosis system based on multi-source data fusion, and includes the following steps:

[0127] Step S1: receiving multi-source operating data from a new energy distribution system, including voltage, current, load power, and environmental state parameters; performing structured sampling on the multi-source operating data in a time series, completing unified format packaging, and generating a multi-source feature tensor through normalization, standardization, and feature extraction;

[0128] Step S2: performing principal component analysis on the feature tensor to remove redundant feature components whose cumulative contribution rate is lower than a preset threshold, thereby forming a compressed feature tensor;

[0129] Step S3: Construct the fault classification probability distribution within the time window corresponding to each data source, calculate the information entropy value, and construct the dynamic fusion weight based on the information entropy value. The fusion reasoning module performs weighted synthesis processing of the confidence information according to the weight and outputs the fusion confidence judgment result;

[0130] Step S4: Build a graph structure model by combining the topological structure relationship of the new energy distribution network and the device connection information, map the fusion confidence result into the feature vector of the node in the graph structure, perform node-level aggregation and reasoning through the structural attribute-aware attention graph neural network model, and generate the node embedding vector and the corresponding fault type label;

[0131] Step S5: generating an isolation control instruction including a control target identifier, an operation type, a priority, and policy parameters based on the diagnostic tag and the node control attribute, and sending the isolation control instruction to the corresponding power distribution equipment terminal for execution;

[0132] Step S6: collecting the terminal execution feedback information, including the action completion status, response time and abnormal flag, and writing the feedback result into the structural attribute field of the corresponding node or edge in the graph structure model;

[0133] Step S7: According to the updated graph structure state, dynamically adjust the information entropy value threshold parameter, fusion weight generation strategy or confidence judgment threshold to achieve adaptive evolution of the diagnosis strategy and form a closed-loop intelligent fault diagnosis and control mechanism.

[0134] In summary, the present invention constructs a new energy distribution fault diagnosis system based on multi-source data fusion, integrating and optimizing multiple key technical links, including data acquisition standardization, fusion reasoning adaptation, graph structure perception diagnosis, and response control closed loop, thus forming a dynamic closed-loop system from data processing to diagnosis execution and strategy feedback. This system not only improves the accuracy of fault identification and response efficiency in a multi-source data environment, but also enhances the system's intelligent regulation capabilities and operational adaptability. It is suitable for a variety of practical application scenarios, such as distributed energy stations, smart distribution networks, and power edge nodes, and has good engineering feasibility and broad prospects for industrial promotion.

[0135] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0136] Any process or method description in the flowchart or otherwise described herein can be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process. The scope of the preferred embodiments of the present application includes additional implementations in which the functions may be performed in a different order than shown or discussed, including in a substantially simultaneous manner or in a reverse order depending on the functions involved.

[0137] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various modifications or substitutions within the technical scope disclosed in this application, and such modifications or substitutions should be included within the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A new energy distribution fault diagnosis system based on multi-source data fusion, characterized in that: include: A data processing module is used to receive multi-source operating data from the new energy distribution system and perform unified structured processing on the multi-source operating data, including normalization, standardization and multi-dimensional feature extraction, to generate corresponding feature tensors; A fusion reasoning module is configured to receive the feature tensor, calculate an information entropy value based on the fault classification probability distribution of each data source within a preset time window, and construct a dynamic fusion weight vector based on the information entropy value. The dynamic fusion weight vector is used to guide the weighted fusion processing of multi-source confidence information and output a fusion confidence judgment result for each type of fault; A graph structure diagnosis module is configured to receive the fusion confidence judgment result and construct a graph structure model based on the topology of the current power distribution network. The graph structure model includes nodes representing power distribution equipment and edges or structural attribute fields representing electrical connection relationships. The fusion result is mapped into embedded features of nodes in the graph, and node-level fault identification and location are performed through a graph neural network model including an attention mechanism, and the fault diagnosis result is output; A control response module is used to receive the fault diagnosis results, generate corresponding isolation control instructions, and receive control execution feedback, and update the feedback information to the structural attribute fields of the nodes or edges in the graph structure model for dynamic correction of the graph structure in subsequent diagnosis cycles.

2. A new energy distribution fault diagnosis system based on multi-source data fusion according to claim 1, characterized in that: The data processing module includes: A data receiving unit is used to receive operating data from multiple edge acquisition terminals, including electrical quantities, state quantities, and environmental parameters, and align and encapsulate the operating data according to timestamps to construct a structured data input format; A processing unit is used to perform normalization processing based on Box-Cox transformation and Z-score standardization on the structured data; a feature extraction unit is used to extract multidimensional features including RMS value, kurtosis, wavelet energy density, spectral center of gravity and frequency offset from each data channel within a sliding time window, and generate a feature tensor for subsequent processing.

3. The new energy distribution fault diagnosis system based on multi-source data fusion according to claim 1 is characterized in that: The fusion reasoning module includes: A feature archiving unit, configured to receive the feature tensors, classify and archive the feature tensors according to data source identifiers, and construct a fusion input set; Uncertainty quantification unit, used to determine the probability distribution P of faults based on each data source within the target time window i ={p i1 ,p i2 ,…,p iK }, calculate the information entropy value, the calculation formula is: Among them: H i is the information entropy value, i represents the data source number, k represents the fault type number, K is the total number of fault categories, and p ik represents the probability of the i-th data source distinguishing the k-th type of fault; Entropy weight generation unit, used to construct dynamic fusion weight w according to the information entropy value i , the calculation formula is: Where: N is the total number of data sources involved in the fusion, j is the data source number in the fusion, H j The information entropy value calculated for the j-th data source; Confidence fusion unit, used to combine the confidence evidence of each data source with the dynamic fusion weight w i Perform weighted synthesis and output the fusion confidence result corresponding to the fault type; The fusion feedback adjustment unit is used to adaptively update the weight generation strategy or confidence threshold parameters in the subsequent time window based on the fusion confidence result.

4. A new energy distribution fault diagnosis system based on multi-source data fusion according to claim 3, characterized in that: The fusion reasoning module further includes: A feature compression unit is used to perform dimensionality compression processing on the received feature tensor before performing uncertainty quantization. The compression processing is based on the principal component analysis (PCA) algorithm. After constructing the sample covariance matrix, the first several principal components are extracted and low-order feature components whose cumulative contribution rate is lower than a preset threshold are eliminated. The compressed tensor output by the feature compression unit serves as the input basis for subsequent information entropy value calculation and weight generation, and is used to improve the feature alignment and fusion consistency between multi-source data. The compression dimension can be dynamically adjusted according to the fusion feedback results.

5. The new energy distribution fault diagnosis system based on multi-source data fusion according to claim 1 is characterized in that: The graph structure diagnosis module includes: A graph model construction unit is configured to construct a graph structure model G = (V, E) based on the topology of the current new energy distribution system and the device connection information, wherein the node set V represents the electrical device unit, and the edge set E represents the physical connection relationship. Each node and edge in the graph structure model is configured with a structural attribute field including a status flag, a device type, and a control attribute; A feature mapping unit is used to map the fusion confidence judgment result to the corresponding node of the graph structure model according to the node device number, and splice it with the historical state feature in the graph structure model to form an input feature vector; A graph neural reasoning unit, configured to execute a graph neural network model with attention based on the graph structure model and node feature input to identify the fault and fault type label of each node; The graph embedding output unit is used to jointly output the intermediate state embedding results of the graph neural network and the fault discrimination results to form a structured diagnosis tensor.

6. A new energy distribution fault diagnosis system based on multi-source data fusion according to claim 5, characterized in that: The graph neural reasoning unit adopts a structural attribute-aware attention mechanism, and the attention weight coefficient between nodes is calculated according to the following formula: Where: α ij represents the attention weight coefficient assigned by the i-th node to its neighboring node j; h i , h j are the input feature vectors of nodes i and j respectively; W1 and W2 are trainable feature mapping matrices used to embed node features into a unified representation space; s i , s j are the structural state attribute vectors of nodes i and j, including device type, current fault status, and historical control records; φ(s i ,s j ) is the joint embedding function between node attributes, which is used to reflect the degree of association between structural states; is a trainable attention parameter vector; σ(·) is the activation mapping function; ‖ represents the vector concatenation operation; h k Represents adjacent nodes The input feature vector of is the set of adjacent nodes of node i.

7. A new energy distribution fault diagnosis system based on multi-source data fusion according to claim 6, characterized in that: The attention graph neural network model further includes the following technical improvement modules: The attribute-driven fine-tuning module is used to dynamically fine-tune the attention weight according to the differences in the node state attribute vectors in the graph structure during the attention coefficient calculation process; The policy control module is used to adjust the scale factor in the normalization process according to the degree distribution characteristics of the nodes in the current graph structure; The connection suppression module is used to prune the edge connections corresponding to the attention weights below the set threshold.

8. The new energy distribution fault diagnosis system based on multi-source data fusion according to claim 1 is characterized in that: The control response module includes: A response instruction generating unit is used to extract the control attributes of the target node and the associated edges according to the fault diagnosis result, and generate an isolation control instruction including a control target identifier, an operation type, a priority, and execution strategy parameters; An instruction issuing and executing unit, configured to issue the isolation control instruction to an execution terminal of a corresponding power distribution device and trigger corresponding switchgear actions, including opening and closing, load switching or feeder isolation control operations; An execution status acquisition unit is used to collect operation feedback information of the execution terminal in real time, the feedback information including the actual status of the switch, the action completion flag, the response time and the abnormal execution flag; The graph attribute update unit is used to map and update the operation feedback information to the state structure attribute field of the corresponding node or edge in the graph structure model, including the row status mark, execution success flag and response delay value, for dynamic adjustment and structural correction of the fusion strategy in the subsequent graph neural network diagnosis process.

9. The new energy distribution fault diagnosis system based on multi-source data fusion according to claim 8 is characterized in that: The control response module further includes: A strategy adjustment unit is used to dynamically adjust the information entropy value threshold parameter, fusion weight generation strategy or node state confidence threshold in the fusion reasoning module according to the state parameters, response delay or execution abnormality flag contained in the control execution feedback; The strategy adjustment unit constructs a fusion mechanism adaptive path driven by feedback information, so that the system can strategically optimize the fusion accuracy according to historical control behavior in the next diagnosis cycle.

10. A new energy distribution fault diagnosis method based on multi-source data fusion, using a new energy distribution fault diagnosis system based on multi-source data fusion as described in any one of claims 1 to 9, characterized in that: The method comprises the following steps: Step S1: receiving multi-source operating data from a new energy distribution system, including voltage, current, load power, and environmental state parameters; performing structured sampling on the multi-source operating data in a time series, completing unified format packaging, and generating a multi-source feature tensor through normalization, standardization, and feature extraction; Step S2: performing principal component analysis on the feature tensor to remove redundant feature components whose cumulative contribution rate is lower than a preset threshold, thereby forming a compressed feature tensor; Step S3: Construct the fault classification probability distribution within the time window corresponding to each data source, calculate the information entropy value, and construct the dynamic fusion weight based on the information entropy value. The fusion reasoning module performs weighted synthesis processing of the confidence information according to the weight and outputs the fusion confidence judgment result; Step S4: Build a graph structure model by combining the topological structure relationship of the new energy distribution network and the device connection information, map the fusion confidence result into the feature vector of the node in the graph structure, perform node-level aggregation and reasoning through the structural attribute-aware attention graph neural network model, and generate the node embedding vector and the corresponding fault type label; Step S5: generating an isolation control instruction including a control target identifier, an operation type, a priority, and policy parameters based on the diagnostic tag and the node control attribute, and sending the isolation control instruction to the corresponding power distribution equipment terminal for execution; Step S6: collecting the terminal execution feedback information, including the action completion status, response time and abnormal flag, and writing the feedback result into the structural attribute field of the corresponding node or edge in the graph structure model; Step S7: According to the updated graph structure state, dynamically adjust the information entropy value threshold parameter, fusion weight generation strategy or confidence judgment threshold to achieve adaptive evolution of the diagnosis strategy and form a closed-loop intelligent fault diagnosis and control mechanism.

Images