Power supply state fault positioning method and system, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a power supply state fault positioning method and system, electronic equipment and a storage medium, and belongs to the technical field of fault positioning. The method comprises the following steps: carrying out residual calculation according to an original time domain electromagnetic signal and a reference frequency spectrum template to obtain a frequency spectrum residual matrix; and performing cross-frame disturbance integral scoring on the spectrum residual vector according to the integral residual scoring function to obtain disturbance scoring data. And obtaining a disturbance event set according to the disturbance score data and the disturbance threshold. And fusing the disturbance event set and the operation log to obtain a state event sequence. And constructing a state semantic tree according to the state event sequence and calculating an edge weight. And scoring each path according to the path scoring function and the edge weight to obtain path scoring data. And selecting a path corresponding to the maximum path score data as a main cause path, wherein a head node of the path is a fault root cause node. And entering a preset execution output process according to the main cause path and the fault root cause node. And the accuracy of power supply fault positioning is improved.

Description

Technical Field

[0001] This application relates to the field of fault location technology, and in particular to a power state fault location method, system, electronic device and storage medium. Background Technology

[0002] With the increasing demands on power supply stability from next-generation industrial control systems, data centers, and rail transportation, intelligent fault location capabilities in power systems have become a core technology for ensuring reliable system operation. In practical applications, power systems typically operate in complex states, involving multiple power supply paths, load-side disturbances, relay protection linkages, and coupling interactions between controllers. Once a system anomaly occurs, traditional methods relying on voltage and current signal thresholds are insufficient to accurately identify fault paths, especially in scenarios involving latent faults, gradual degradation, and chain propagation. Therefore, improving the accuracy of power supply fault location has become a pressing technical challenge. Summary of the Invention

[0003] The main objective of this application is to provide a power state fault location method, system, electronic device, and storage medium, aiming to improve the accuracy of power fault location.

[0004] To achieve the above objectives, a first aspect of this application proposes a power state fault location method, the method comprising:

[0005] Acquire raw time-domain electromagnetic signals and runtime logs;

[0006] The residual is calculated based on the original time-domain electromagnetic signal and the preset reference spectrum template to obtain the spectrum residual matrix; wherein, the spectrum residual matrix includes multiple spectrum residual vectors;

[0007] The spectral residual vector is subjected to cross-frame perturbation integral scoring according to a preset integral residual scoring function to obtain perturbation score data;

[0008] Based on the disturbance score data and the preset disturbance threshold, a set of disturbance events is obtained;

[0009] The set of disturbance events and the operation log are fused to obtain a sequence of state events;

[0010] A state semantic tree is constructed based on the state event sequence, and the edge weights of the state semantic tree are calculated according to a preset edge weight function to obtain an edge weight set; wherein, the state semantic tree includes multiple paths;

[0011] Each path is scored according to a preset path scoring function and the set of edge weights to obtain path score data;

[0012] The path corresponding to the highest path score is selected as the primary cause path, and the first node of the primary cause path is the root cause node of the fault.

[0013] The preset execution output process is entered based on the main cause path and the root cause node of the fault.

[0014] In some embodiments, the state semantic tree includes multiple nodes. After constructing the state semantic tree based on the state event sequence and calculating the edge weights of the state semantic tree according to a preset edge weight function to obtain a set of edge weights, the method further includes:

[0015] Each node is scored according to a preset causal influence scoring function to obtain causal score data;

[0016] The edge weights of the state semantic tree are calculated based on the causal scoring data and the edge weight set to obtain the updated edge weight set.

[0017] In some embodiments, the step of calculating the residual based on the original time-domain electromagnetic signal and a preset reference spectrum template to obtain a spectral residual matrix includes:

[0018] A fast Fourier transform is performed on the original time-domain electromagnetic signal of each frame to obtain the current spectrum matrix; wherein the current spectrum matrix includes multiple current spectrum vectors;

[0019] Obtain a reference spectrum template; wherein the reference spectrum template is the average of the spectrum vectors of multiple health frames;

[0020] The spectral residual matrix is ​​obtained by performing residual calculation based on the current spectral vector and the reference spectral template; wherein the spectral residual matrix includes multiple spectral residual vectors.

[0021] In some embodiments, the step of calculating the spectral residual matrix based on the current spectral vector and the reference spectral template includes:

[0022] The formula for calculating residuals is:

[0023] ;

[0024] in, Represents the spectral residual vector. Indicates the first The current spectral vector of the frame, Indicates the reference spectrum template. This represents a very small constant.

[0025] In some embodiments, the step of performing cross-frame perturbation integral scoring on the spectral residual vector according to a preset integral residual scoring function to obtain perturbation score data includes:

[0026] The integral residual scoring function is:

[0027] ;

[0028] in, Represents a node The perturbation score data, Indicates frequency band range, Represents the spectral residual vector. Denotes the coefficients of the second-order regularization term. This represents the second-order difference of the spectrum.

[0029] In some embodiments, the step of constructing a state semantic tree based on the state event sequence and calculating the edge weights of the state semantic tree according to a preset edge weight function to obtain an edge weight set includes:

[0030] The edge weight function is:

[0031] ;

[0032] in, Represents a node To the node The right to the border, This represents the Sigmoid function. , , These represent the weighting coefficients, Represents the coefficient of the regularization term. Standard cosine similarity represents the characteristics of event states. Indicates whether two events are reachable in the system control topology. Indicates the time difference between two events. Indicates the heterogeneity of the characteristic structure.

[0033] In some embodiments, scoring each path according to a preset path scoring function and the set of edge weights to obtain path score data includes:

[0034] The path scoring function is:

[0035] ;

[0036] in, This represents path rating data. Indicates a path. Indicates the edge weight, Indicates the path length.

[0037] To achieve the above objectives, a second aspect of this application provides a power state fault location system, the system comprising:

[0038] The acquisition module is used to acquire raw time-domain electromagnetic signals and operation logs;

[0039] The residual module is used to perform residual calculation based on the original time-domain electromagnetic signal and a preset reference spectrum template to obtain a spectral residual matrix; wherein, the spectral residual matrix includes multiple spectral residual vectors;

[0040] The first scoring module is used to perform cross-frame perturbation integral scoring on the spectral residual vector according to a preset integral residual scoring function to obtain perturbation scoring data.

[0041] The disturbance module is used to obtain a set of disturbance events based on the disturbance score data and a preset disturbance threshold.

[0042] The fusion module is used to fuse the set of disturbance events and the operation log to obtain a sequence of state events;

[0043] A construction module is used to construct a state semantic tree based on the state event sequence and calculate the edge weights of the state semantic tree according to a preset edge weight function to obtain an edge weight set; wherein, the state semantic tree includes multiple paths;

[0044] The second scoring module is used to score each path according to a preset path scoring function and the edge weight set to obtain path scoring data.

[0045] The selection module is used to select the path corresponding to the largest path score data as the root cause path, and the first node of the root cause path is the root cause node of the fault.

[0046] The execution module is used to enter a preset execution output process based on the main cause path and the root cause node of the fault.

[0047] To achieve the above objectives, a third aspect of the present application provides an electronic device, the electronic device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method described in the first aspect.

[0048] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0049] The power state fault location method, system, electronic device, and storage medium proposed in this application acquire raw time-domain electromagnetic signals and operation logs. Residual calculation is performed based on the raw time-domain electromagnetic signals and a preset reference spectrum template to obtain a spectral residual matrix, which includes multiple spectral residual vectors. Cross-frame perturbation integral scoring is performed on the spectral residual vectors according to a preset integral residual scoring function to obtain perturbation score data. A set of perturbation events is obtained based on the perturbation score data and a preset perturbation threshold. The set of perturbation events and the operation logs are fused to obtain a state event sequence. A state semantic tree is constructed based on the state event sequence, and the edge weights of the state semantic tree are calculated according to a preset edge weight function to obtain a set of edge weights, where the state semantic tree includes multiple paths. Each path is scored according to a preset path scoring function and the set of edge weights to obtain path score data. The path corresponding to the path with the largest path score is selected as the primary cause path, and the first node of the primary cause path is the root cause node. A preset execution output process is then initiated based on the primary cause path and the root cause node. It achieves closed-loop processing across the entire chain from perception, modeling, reasoning to execution, thereby improving the accuracy of power fault location. Attached Figure Description

[0050] Figure 1 This is a flowchart of the power state fault location method provided in the embodiments of this application;

[0051] Figure 2 yes Figure 1 The flowchart of step S102 in the document;

[0052] Figure 3 This is a flowchart of a power state fault location method provided in another embodiment of this application;

[0053] Figure 4 This is a schematic diagram of the power status fault location system provided in the embodiments of this application;

[0054] Figure 5 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] It should be noted that although functional modules are divided in the system diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0058] With the increasing demands on power supply stability from next-generation industrial control systems, data centers, and rail transportation, intelligent fault location capabilities in power systems have become a core technology for ensuring reliable system operation. In practical applications, power systems typically operate in complex states, involving multiple power supply paths, load-side disturbances, relay protection linkages, and coupling interactions between controllers. Once a system anomaly occurs, traditional methods relying on voltage and current signal thresholds are insufficient to accurately identify fault paths, especially in scenarios involving latent faults, gradual degradation, and chain propagation.

[0059] Based on this, embodiments of this application provide a power state fault location method, system, electronic device, and storage medium, aiming to automatically identify perturbation events from spectrum signals and accurately extract potential abnormal signals in non-alarm states through frequency domain residual calculation and integral scoring mechanisms. Subsequently, these signals are fused with the operation logs generated by the control system to construct a multi-source state event sequence with a unified time axis, thereby establishing a semantic tree structure capable of expressing the state evolution path to reconstruct the logical evolution process from early anomalies to manifest faults. Furthermore, a causal graph modeling mechanism based on path penetration scoring is designed. Through structural node influence assessment, causal weight modulation is applied to the state evolution path to improve the accuracy and interpretability of principal node identification. Ultimately, the system can not only identify the root cause node with the highest causal weight but also output the fault path and location results as structured, directly executable suggested actions and alarm signals from the control system, achieving a closed-loop processing across the entire chain from perception, modeling, reasoning to execution, thereby improving the accuracy of power fault location.

[0060] The power state fault location method, system, electronic device and storage medium provided in the embodiments of this application are specifically described through the following embodiments. First, the power state fault location method in the embodiments of this application is described.

[0061] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0062] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0063] The power status fault location method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the power status fault location method, but is not limited to the above forms.

[0064] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0065] Please refer to Figure 1 , Figure 1 This is a flowchart of the power state fault location method provided in the embodiments of this application. Figure 1The method may include, but is not limited to, steps S101 to S109.

[0066] Step S101: Obtain the original time-domain electromagnetic signal and operation log;

[0067] Step S102: Perform residual calculation based on the original time-domain electromagnetic signal and the preset reference spectrum template to obtain the spectrum residual matrix; wherein, the spectrum residual matrix includes multiple spectrum residual vectors;

[0068] Step S103: Perform cross-frame perturbation integral scoring on the spectral residual vector according to the preset integral residual scoring function to obtain perturbation scoring data;

[0069] Step S104: Obtain a set of disturbance events based on the disturbance score data and the preset disturbance threshold;

[0070] Step S105: Merge the set of disturbance events and the running log to obtain a sequence of state events;

[0071] Step S106: Construct a state semantic tree based on the state event sequence and calculate the edge weights of the state semantic tree according to the preset edge weight function to obtain the edge weight set; wherein, the state semantic tree includes multiple paths;

[0072] Step S107: Score each path according to the preset path scoring function and edge weight set to obtain path score data;

[0073] Step S108: Select the path corresponding to the largest path score data as the main cause path, and the first node of the main cause path is the root cause node of the fault.

[0074] Step S109: Enter the preset execution output process based on the main cause path and the root cause node of the fault.

[0075] In step S101 of some embodiments, a near-field electromagnetic interference sensor deployed near the housing of the power controller, output bus, or critical load module is used to sample the signal and obtain the original time-domain electromagnetic signal. The signal is connected to a high-speed analog-to-digital converter (such as the TI ADS8881) via a shielded coaxial cable. The sampling parameters are: 2MHz sampling rate, 2048-point window length, Hamming window function, and 50% inter-frame overlap. Additionally, the power system's operating logs need to be acquired. .

[0076] Please see Figure 2 In some embodiments, step S102 may include, but is not limited to, steps S201 to S203:

[0077] Step S201: Perform a fast Fourier transform on the original time-domain electromagnetic signal of each frame to obtain the current spectrum matrix; wherein, the current spectrum matrix includes multiple current spectrum vectors;

[0078] Step S202: Obtain a reference spectrum template; wherein, the reference spectrum template is the average of the spectrum vectors of multiple health frames;

[0079] Step S203: Perform residual calculation based on the current spectrum vector and the reference spectrum template to obtain the spectrum residual matrix; wherein, the spectrum residual matrix includes multiple spectrum residual vectors.

[0080] In step S201 of some embodiments, the system continuously acquires raw time-domain electromagnetic signals at a set sampling rate. Each frame consists of 2048 points, with a 1024-point overlap between frames. After windowing, each frame undergoes a Fast Fourier Transform to obtain the current spectrum vector. As shown in the following formula (1):

[0081] (1)

[0082] in, This represents the current spectrum vector, indicating the energy distribution of this frame in the frequency domain. This represents the Fast Fourier Transform operation, implemented using an embedded FFT library. This represents a sequence of window functions (Hamming window, 2048 points). Indicates the first The original time-domain electromagnetic signal of the frame.

[0083] The current spectrum matrix comprises multiple current spectrum vectors. The current spectrum matrix is ​​obtained by performing the above transformation on all original time-domain electromagnetic signals, as shown in formula (2) below:

[0084] (2)

[0085] in, This represents the current spectrum matrix. Represents the current spectrum vector. For frame number, It is a frequency index, with each column representing the spectral amplitude of a certain frame, forming... A two-dimensional structure.

[0086] In step S202 of some embodiments, a normal operating state is marked. The data acquisition task is synchronously written by the main controller (such as STM32 or PLC) via a communication bus. During the template construction phase, the system uses... Select the time period marked as "normal". Health frame spectrum vector The mean value at each frequency point is calculated as a reference spectrum template, as shown in formula (3) below:

[0087] (3)

[0088] in, Indicates the reference spectrum template. This indicates the number of sampled frames, typically set to 300–1000 frames. Indicates the first Each healthy frame's spectral vector. After being constructed using a reference spectral template, it is normalized to... and The spectral amplitudes are within the same scale range, which facilitates subsequent residual analysis.

[0089] In step S203 of some embodiments, the current spectrum matrix already obtained in the previous step... With reference spectrum template Based on this, the difference residual between the two is calculated, and potential disturbance events are mined in the spectral time series. These disturbance events represent "physical layer signal precursors" of early system anomalies, which may not be visible in traditional voltage and current monitoring and are characterized by high frequency, low intensity, and short duration.

[0090] In response to the background of chain-propagation faults and inconspicuous early fluctuations in power systems, this step introduces a local peak normalization regularization term and a disturbance delay feedback fitting term of the frequency domain residual to enhance the ability to identify minor anomalies in multiple frequency bands and avoid misjudging instantaneous unstructured fluctuations.

[0091] The system first processes the current spectrum vector for each time window. With reference spectrum template The residual is calculated to obtain the amplitude difference in the frequency domain at that time point. To enhance the sensitivity to local anomalous frequency points, a normalization term is introduced into the calculation so that the offset metric is not suppressed by the amplitude difference in the frequency band. As shown in the following formula (4):

[0092] (4)

[0093] in, This represents the spectral residual vector, reflecting the degree of deviation at each frequency point. Indicates the first The current spectral vector of the frame, from Take it out. Indicates the reference spectrum template. To represent a very small constant and prevent division by zero, a typical value is... The spectral residual matrix comprises multiple spectral residual vectors.

[0094] Through steps S201 to S203, the spectral residual vector can be obtained. Then, cross-frame perturbation integration scoring is performed on the spectral residual vector to obtain perturbation score data. Based on the perturbation score data, it is determined whether a micro-perturbation event has occurred.

[0095] In step S103 of some embodiments, in order to further improve the ability to identify "discontinuous weak abnormal disturbances" in the power supply system, a cross-frame disturbance integral scoring mechanism is introduced on the basis of frame level, and a regularization suppression term is set to avoid false alarms caused by isolated abrupt changes in a certain frame. The integral residual scoring function is shown in the following formula (5):

[0096] (5)

[0097] in, Represents a node The disturbance score data is used to determine whether a disturbance event has occurred. This indicates a frequency band range, which in practice is set as a centralized monitoring frequency band, such as the power control switch frequency band or a typical EMI resonant frequency point. Represents the spectral residual vector. This represents the coefficient of the second-order regularization term, which controls the impact of the rate of change of the disturbance on the overall score; a typical value is 0.3. This represents the second-order difference of the spectrum, used to suppress gradual trends in the background and highlight peak disturbances (i.e., high-frequency vibration rate of change).

[0098] The integral residual scoring function not only focuses on the absolute deviation value but also considers the sharpness of the perturbation curve in the spectrum. This makes the system more inclined to detect local abrupt signals with structured, non-smooth peaks, thus distinguishing them from ordinary spectral perturbation background. This strategy is particularly suitable for local band surge behavior in power supply systems caused by MOSFET breakdown, relay jitter, and capacitor degradation, which is difficult to identify using traditional frequency domain thresholding methods.

[0099] In step S104 of some embodiments, the perturbation threshold can be set to an adaptive multiple. , This is the average disturbance score for the reference interval. Once... Adaptive factor exceeding background baseline score The system then marks this frame as the center frame of the disturbance and searches for frames with continuously exceeding residual limits in the same frequency band within the consecutive windows before and after it, ultimately merging them to form a disturbance event. The disturbance event fields include: start timestamp. Center frequency point Average residual amplitude and duration of frames .

[0100] The above process ultimately generates a set of disturbance events. Each event is a structured object composed of the above fields, which can be directly used in subsequent steps for constructing state event sequences and activating causal paths.

[0101] In step S105 of some embodiments, state events from different sources are integrated into a unified state event sequence. This provides standardized input for subsequent state evolution modeling and semantic path generation. (Set of perturbation events) and spectral residual matrix It is a direct expression of the "physical signal layer," while the operating logs come from power supply operating devices (such as relay actions and trip logs). This is an expression of the "control behavior layer". Integrating these two into a unified state event sequence is a necessary prerequisite for realizing the subsequent "power state path modeling".

[0102] First of all, Each perturbation event in Represent it in a structured way as state event nodes .in: Use the timestamp of the disturbance event directly. Fixed as "perturb," "perturb" indicates that the state event originates from spectral perturbation detection and belongs to the physical layer's micro-perturbation signal event. This type identifier is used to distinguish state nodes from different sources, ensuring accurate identification of the perturbation origin during modeling. Feature Vector Construct as shown in the following formula (6):

[0103] (6)

[0104] in, This represents the mean residual value in the perturbation frequency band. , express The spectral residual vector corresponding to the frame is actually taken from the spectral residual matrix. Zhongyu The closest frame. This indicates the perturbation bandwidth. The perturbation shape compactness index is defined as the coefficient of variation of the residual within the frequency band. This measures whether the disturbance is concentrated at certain frequencies.

[0105] Next, check the runtime log. Each event in The device uses a pre-built event encoding dictionary to convert it into a fixed-length sparse state vector. Taking a UPS control system as an example:

[0106] "relay_closed" is mapped to [1,0,0,0,...];

[0107] "breaker_trip" is mapped to [0,1,0,0,...];

[0108] "overvoltage" is mapped to [0,0,1,0,...];

[0109] Here, "relay_closed" indicates relay closure, "breaker_trip" indicates circuit breaker tripping, and "overvoltage" indicates overvoltage. This mapping process uses an event source mapping table, retaining the original event index for subsequent traceability retrieval. The event source mapping table is typically deployed in the state parsing module of an embedded system, and is completed through table lookup, offering high real-time performance and engineering feasibility.

[0110] Finally, the system merges all events (from disturbances and logs) into a unified timeline and arranges them in ascending chronological order, forming a triplet structure. State event sequence .in, For the time of the event, The event type is identified (e.g., "perturb", "breaker", "relay"). "perturb" indicates a spectral perturbation event, "breaker" indicates a circuit breaker operation event, and "relay" indicates a relay state change event. This is a feature representation vector for the event.

[0111] In one example, suppose a power system detects a spectral perturbation at 10:02:03.500, with a frequency band of [1.5MHz, 1.8MHz], a mean residual of 0.27, and a coefficient of variation of 0.12; and the controller records a relay activation at 10:02:03.600. These two events in the final state event sequence are represented as: $(10:02:03.500,"perturb",[0.27,0.3,0.12])$; $(10:02:03.600,"relay",[0, 0,0,1,0,...,0])$.

[0112] In step S106 of some embodiments, in the state event sequence Based on this, a state semantic tree that conforms to the actual operating logic of a power system is constructed. This structure should be able to describe the entire chain propagation process from physical disturbances (such as spectral perturbations) to control behaviors (such as relay actions), alarm events, and finally manifest faults, reflecting the evolution and response relationships between states.

[0113] Considering that complex control logic power devices such as UPS systems and industrial power control systems often exhibit multi-stage, cross-module, and weakly coupled propagation characteristics, traditional rule-chain-based or fixed-flowchart methods are insufficient to establish stable state evolution paths. Therefore, this step designs an innovative construction method based on heterogeneous event topology mapping, polynomial regularized edge weight inference, and anomaly-driven sub-path extraction. This method enables the state semantic tree to express real physical behavior chains and adapt to situations where early signals serve as propagation starting points.

[0114] A state event sequence includes multiple state events. Building a state semantic tree first requires modeling the relationships between state events as an initial candidate graph structure. , where the set of nodes For all state events, the edge set This needs to be calculated from the correlation between state events. Let the two state event nodes be... and ,like Furthermore, if there are potential logical / physical / topological dependencies between the two state events, then a directed edge is constructed. .

[0115] Unlike the common "cosine similarity + time window" edge construction method, this scheme introduces a comprehensive weight function that combines the device control logic topology, physical adjacency, and state feature structure differences, and uses a nonlinear weighting function to suppress the dominance of a single feature, preventing "over-connection" of physically unrelated events. That is, the edge weight function is shown in the following formula (7):

[0116] (7)

[0117] in, Represents a node To the node The edge weights, each representing a structural weight, indicate semantic similarity or the connection strength controlling the path. All edge weights constitute the edge weight set. . This represents the Sigmoid function, ensuring that the edge weights are within a certain range. . , , These represent the weighting coefficients, and their typical values ​​can be set to... . This represents the regularization coefficient, used to penalize the risk of direct connections between highly similar but heterogeneous feature events. Standard cosine similarity, representing the characteristics of an event's state, reflects the similarity of the event's content. This indicates whether two state events in the system control topology are reachable, which is obtained by looking up a table (e.g., whether a relay action may affect the downstream voltage). This represents the time difference between two state events. The heterogeneity of the feature structure is defined as follows: This reflects whether two state events belong to similar categories.

[0118] Edge weight function by introducing The regularization term classifies connections between state events with high "content differences" as low-confidence edges, making it more suitable for chain-like expressions of hierarchical jumps in power supply systems, such as "physical disturbance → control action → alarm". The time term uses a squared penalty to suppress long-distance virtual connections, while leveraging the system's internal control graph topology for discrimination to enhance robustness.

[0119] After obtaining all edge weights, the system processes each state event. If its type is "perturb", it is considered the root node of the latent semantic tree. Starting from it, a weighted shortest path algorithm (such as Dijkstra's algorithm) is used to select the set of sub-paths with the largest cumulative weights. The maximum depth of each tree is then limited. (e.g., 3 or 5), remove weak edges with low relevance, and finally form a state semantic tree with consistent direction and tree structure.

[0120] For example, if the sequence of state events is as follows:

[0121] : "perturb" ;

[0122] : "relay_open", ;

[0123] : "voltage_alarm", ;

[0124] Then the system build path And set it as a main branch of the state semantic tree.

[0125] It should be noted that, since the state propagation path of a power system may be nonlinear and weakly coupled, the connections between state events are not based on necessary logical rules, but rather on a "semantic path" constructed through the combination and association of multi-dimensional information. Therefore, in order to identify the event nodes that truly play a dominant role in propagation, a causal modeling mechanism based on structural propagation effects needs to be designed without changing the node / edge structure, thus elevating semantic strength to causal strength. To this end, this step uses the edge weight set generated in step S106. As a foundation, each edge The original weights are denoted as Based on this, node causal scoring is introduced for weighted modulation to construct new edge weights. This is to demonstrate the influence of cause and effect.

[0126] For details, please refer to Figure 3 In some embodiments, after step S106, the state semantic tree includes multiple nodes, and the power state fault location method may also include, but is not limited to, steps S301 to S302:

[0127] Step S301: Score each node according to the preset causal influence scoring function to obtain causal score data;

[0128] Step S302: Calculate the edge weights of the state semantic tree based on the causal scoring data and the edge weight set to obtain the updated edge weight set.

[0129] In step S301 of some embodiments, a causal influence scoring function for the propagation node is defined based on the "information transmission capability" of the path relay node. In a power system, if a certain node... The existence of a node allows a large number of subsequent nodes to be "activated," meaning that the node is a "main point" on a path, and therefore has a strong causal influence. Based on path penetration as the causal scoring benchmark, the causal influence scoring function is shown in the following formula (8):

[0130] (8)

[0131] in, Represents a node Causal scoring data, causal scoring table . Indicates from node Starting point, the set of reachable downstream nodes. This indicates the size of the set, i.e., the propagation depth. This represents the number of nodes reachable from the node with the longest propagation path among all nodes. This represents a small constant and prevents division by zero.

[0132] The causal influence scoring function indicates that the longer the propagation chain controlled by a node on the path, the more likely that node is to be a "critical influence node". For example, in a UPS system, a spectrum disturbance node can be connected to three subsequent control actions and two alarm events, while another intermediate transition event is only connected to one voltage drop event. In this case, the former is more likely to be the primary cause of the failure.

[0133] In step S302 of some embodiments, the above-mentioned causal scoring data is used as the modulation factor of the edge to construct the edge weight function with causal scoring, as shown in the following formula (9):

[0134] (9)

[0135] in, This represents the updated edge weights, and all updated edge weights constitute the updated edge weight set. This indicates the edge weights before the update. Represents a node Causal scoring data, Represents a node Causal scoring data.

[0136] Formula (9) avoids introducing new modeling modules and quantifies the importance of state events solely through "reachability logic on existing paths". This approach is naturally suited to the structural characteristics of power systems, which are characterized by "chain failures and multi-cause triggering". It is particularly suitable for modeling causal paths from the perspective of "information flow" between events in real-world scenarios where there is no explicit control logic.

[0137] In calculation Then, a causal structure diagram with causal scores was constructed. Structure and State Semantic Tree The results are the same, but the edge weights now possess causal interpretability. In practical deployments, this model can be implemented using a graph analysis library (such as using an embedded graph traversal algorithm) or represented as an adjacency matrix in an embedded controller for real-time evaluation. In one example, assume a certain state semantic tree is as follows:

[0138] : Spectrum disturbance node. ;

[0139] Relay operation, ;

[0140] Overvoltage alarm, endpoint.

[0141] The edges are connected as follows: The system then calculates sequentially:

[0142] ;

[0143] ;

[0144] set up , ;

[0145] have to ;

[0146] (The endpoint cannot be propagated).

[0147] Through the above steps S301 to S302, the updated edge weight function is obtained, and the accuracy of power supply fault location based on the updated edge weight function is higher.

[0148] In step S107 of some embodiments, the system traverses the causal structure graph. All paths, starting from physical disturbances or control operations, ultimately lead to the complete set of paths that result in an "obvious fault" (such as power outage or main unit tripping). The data is then sorted according to its path score. The path score function is shown in the following formula (10):

[0149] (10)

[0150] in, This represents path rating data. Indicates a path. Indicates the boundary rights, Indicates the path length.

[0151] Furthermore, when scoring a path using the updated set of edge weights, the path scoring function is shown in the following formula (11):

[0152] (11)

[0153] in, This represents path rating data. Indicates a path. This indicates the updated edge weights. Indicates the path length.

[0154] In some embodiments, in steps S108 to S109, the system selects Maximum path As the main cause path The first node of the main path is the root cause node of the failure. .

[0155] After this, the system enters the execution output process, which will... and The output is a structured result, including but not limited to:

[0156] Root cause event ID, type (e.g., "spectral disturbance", "relay malfunction"), and timestamp of occurrence;

[0157] The physical location (such as UPS branch, module number) is extracted from the front-end data collection / logs;

[0158] The inferred propagation chain event sequence and timeline;

[0159] Action suggestions, such as:

[0160] If the cause is spectral disturbance: generate "EMI suppression circuit detection suggestions";

[0161] If the relay is faulty: generate a "Relay K3 Status Maintenance Request";

[0162] If there is a sudden change in parameters: trigger the "automatic load reduction" or "enable bypass" policy command.

[0163] Simultaneously output the following actually available system interface signaling:

[0164] Submit to the fault alarm module ;

[0165] Submit to the maintenance work order system ;

[0166] Write back to the master control strategy system .

[0167] For example, a certain path is:

[0168] (Spectrum disturbance @10:01:00) → (Relay switching) → (Voltage drop) → (Power outage);

[0169] The system will Identified as ,path → for The system outputs structured content including time, location, suggested inspection actions, work order suggestions, etc. The system immediately sends a "prioritize detection of EMI interference sources in UPS 3 modules" command to the main controller.

[0170] Steps S101 to S109 of this embodiment involve acquiring the original time-domain electromagnetic signal and the operation log. Residual calculation is performed based on the original time-domain electromagnetic signal and a preset reference spectrum template to obtain a spectral residual matrix, which includes multiple spectral residual vectors. Cross-frame perturbation integral scoring is performed on the spectral residual vectors according to a preset integral residual scoring function to obtain perturbation scoring data. A perturbation event set is obtained based on the perturbation scoring data and a preset perturbation threshold. The perturbation event set and the operation log are fused to obtain a state event sequence. A state semantic tree is constructed based on the state event sequence, and the edge weights of the state semantic tree are calculated according to a preset edge weight function to obtain an edge weight set, where the state semantic tree includes multiple paths. Each path is scored according to a preset path scoring function and the edge weight set to obtain path scoring data. The path corresponding to the path with the highest path score is selected as the primary cause path, and the first node of the primary cause path is the root cause node. A preset execution output process is then initiated based on the primary cause path and the root cause node. It achieves closed-loop processing across the entire chain from perception, modeling, reasoning to execution, thereby improving the accuracy of power fault location.

[0171] Please see Figure 4 This application also provides a power state fault location system, which can implement the above-mentioned power state fault location method. The system includes:

[0172] Acquisition module 401 is used to acquire raw time-domain electromagnetic signals and operation logs;

[0173] The residual module 402 is used to perform residual calculation based on the original time-domain electromagnetic signal and the preset reference spectrum template to obtain the spectrum residual matrix; wherein, the spectrum residual matrix includes multiple spectrum residual vectors;

[0174] The first scoring module 403 is used to perform cross-frame perturbation integral scoring on the spectral residual vector according to the preset integral residual scoring function to obtain perturbation scoring data.

[0175] The disturbance module 404 is used to obtain a set of disturbance events based on the disturbance score data and the preset disturbance threshold.

[0176] The fusion module 405 is used to fuse the set of disturbance events and the operation log to obtain a sequence of state events.

[0177] The construction module 406 is used to construct a state semantic tree based on the state event sequence and calculate the edge weights of the state semantic tree according to the preset edge weight function to obtain the edge weight set; wherein, the state semantic tree includes multiple paths;

[0178] The second scoring module 407 is used to score each path according to a preset path scoring function and edge weight set to obtain path scoring data.

[0179] The selection module 408 is used to select the path corresponding to the largest path score data as the main cause path, and the first node of the main cause path is the root cause node of the fault.

[0180] The execution module 409 is used to enter a preset execution output process based on the main cause path and the root cause node of the fault.

[0181] The specific implementation of this power state fault location system is basically the same as the specific implementation of the power state fault location method described above, and will not be repeated here.

[0182] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned power state fault location method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0183] Please see Figure 5 , Figure 5 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0184] The processor 501 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0185] The memory 502 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 502 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 502 and is called and executed by the processor 501 to execute the power state fault location method of the embodiments of this application.

[0186] The input / output interface 503 is used to implement information input and output;

[0187] The communication interface 504 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0188] Bus 505 transmits information between various components of the device (e.g., processor 501, memory 502, input / output interface 503, and communication interface 504);

[0189] The processor 501, memory 502, input / output interface 503, and communication interface 504 are connected to each other within the device via bus 505.

[0190] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described power state fault location method.

[0191] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0192] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0193] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0194] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0195] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0196] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application 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 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.

[0197] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0198] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of the units described above 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 coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0199] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0200] Furthermore, the functional units in the various embodiments of this application 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.

[0201] 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, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0202] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for locating power supply status faults, characterized in that, The method includes: Acquire raw time-domain electromagnetic signals and runtime logs; The residual is calculated based on the original time-domain electromagnetic signal and the preset reference spectrum template to obtain the spectrum residual matrix; wherein, the spectrum residual matrix includes multiple spectrum residual vectors; The spectral residual vector is subjected to cross-frame perturbation integral scoring according to a preset integral residual scoring function to obtain perturbation score data; Based on the disturbance score data and the preset disturbance threshold, a set of disturbance events is obtained; The set of disturbance events and the operation log are fused to obtain a sequence of state events; A state semantic tree is constructed based on the state event sequence, and the edge weights of the state semantic tree are calculated according to a preset edge weight function to obtain an edge weight set; wherein, the state semantic tree includes multiple paths; Each path is scored according to a preset path scoring function and the set of edge weights to obtain path score data; The path corresponding to the highest path score is selected as the primary cause path, and the first node of the primary cause path is the root cause node of the fault. The preset execution output process is entered based on the main cause path and the root cause node of the fault.

2. The method according to claim 1, characterized in that, The state semantic tree includes multiple nodes. After constructing the state semantic tree based on the state event sequence and calculating the edge weights of the state semantic tree according to a preset edge weight function to obtain the edge weight set, the method further includes: Each node is scored according to a preset causal influence scoring function to obtain causal score data; The edge weights of the state semantic tree are calculated based on the causal scoring data and the edge weight set to obtain the updated edge weight set.

3. The method according to claim 1, characterized in that, The step of calculating the residual based on the original time-domain electromagnetic signal and a preset reference spectrum template to obtain the spectral residual matrix includes: A fast Fourier transform is performed on the original time-domain electromagnetic signal of each frame to obtain the current spectrum matrix; wherein the current spectrum matrix includes multiple current spectrum vectors; Obtain a reference spectrum template; wherein the reference spectrum template is the average of the spectrum vectors of multiple health frames; The spectral residual matrix is ​​obtained by performing residual calculation based on the current spectral vector and the reference spectral template; wherein the spectral residual matrix includes multiple spectral residual vectors.

4. The method according to claim 3, characterized in that, The step of calculating the spectral residual matrix based on the current spectral vector and the reference spectral template includes: The formula for calculating residuals is: ； in, Represents the spectral residual vector. Indicates the first The current spectral vector of the frame, Indicates the reference spectrum template. This represents a very small constant.

5. The method according to claim 1, characterized in that, The step of performing cross-frame perturbation integral scoring on the spectral residual vector according to a preset integral residual scoring function to obtain perturbation score data includes: The integral residual scoring function is: ； in, Represents a node The perturbation score data, Indicates frequency band range, Represents the spectral residual vector. Denotes the coefficients of the second-order regularization term. This represents the second-order difference of the spectrum.

6. The method according to claim 1, characterized in that, The step of constructing a state semantic tree based on the state event sequence and calculating the edge weights of the state semantic tree according to a preset edge weight function to obtain an edge weight set includes: The edge weight function is: ； in, Represents a node To the node The right to the border, This represents the Sigmoid function. , , These represent the weighting coefficients, Represents the coefficient of the regularization term. Standard cosine similarity represents the characteristics of event states. Indicates whether two events are reachable in the system control topology. Indicates the time difference between two events. Indicates the heterogeneity of the characteristic structure.

7. The method according to claim 1, characterized in that, The step of scoring each path according to a preset path scoring function and the edge weight set to obtain path scoring data includes: The path scoring function is: ； in, This represents path rating data. Indicates a path. Indicates the boundary rights, Indicates the path length.

8. A power supply status fault location system, characterized in that, The system includes: The acquisition module is used to acquire raw time-domain electromagnetic signals and operation logs; The residual module is used to perform residual calculations based on the original time-domain electromagnetic signal and a preset reference spectrum template to obtain a spectral residual matrix; wherein, the spectral residual matrix includes multiple spectral residual vectors; The first scoring module is used to perform cross-frame perturbation integral scoring on the spectral residual vector according to a preset integral residual scoring function to obtain perturbation scoring data. The disturbance module is used to obtain a set of disturbance events based on the disturbance score data and a preset disturbance threshold. The fusion module is used to fuse the set of disturbance events and the operation log to obtain a sequence of state events; A construction module is used to construct a state semantic tree based on the state event sequence and calculate the edge weights of the state semantic tree according to a preset edge weight function to obtain an edge weight set; wherein, the state semantic tree includes multiple paths; The second scoring module is used to score each path according to a preset path scoring function and the edge weight set to obtain path scoring data. The selection module is used to select the path corresponding to the largest path score data as the root cause path, and the first node of the root cause path is the root cause node of the fault. The execution module is used to enter a preset execution output process based on the main cause path and the root cause node of the fault.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the power state fault location method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the power state fault location method according to any one of claims 1 to 7.

Images