A weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction

Abstract

The present invention provides a multi-mode networking adaptive monitoring system for weak signals in power transmission and transformation construction, relating to the technical field of multi-mode networking. The system comprises a multi-modal module for deploying multi-modal sensing units at key monitoring points on power transmission and transformation equipment to collect vibration, partial discharge, leakage current, and environmental parameter signals. The system then processes these signals through an electromagnetic shielding enclosure and a built-in harmonic filtering circuit to obtain raw monitoring signal data. A detection module inputs the raw monitoring signal data into a chaos-enhanced weak signal detection device, extracting fault feature vectors by capturing sudden changes in the attractor phase diagram and time-domain waveform thresholds. By utilizing multi-modal sensing and adaptive networking technologies, the present invention enables monitoring of weak signals in power transmission and transformation and timely extraction of fault features, thereby improving the accuracy and adaptive capabilities of fault warnings.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-module networking, and in particular to a weak-signal multi-module networking adaptive monitoring system for power transmission and transformation construction. Background Art

[0002] In power transmission and transformation projects, traditional technologies have limitations when dealing with weak signals in complex environments. For example, during the construction of new substations in mountainous areas, due to complex terrain and insufficient base station coverage, traditional single-mode signal transmission technology is prone to signal interruption or transmission delays. During the critical installation of high-voltage transmission towers, when construction workers used traditional monitoring equipment to record tower foundation pouring data, unstable signals resulted in incomplete transmission of some data, preventing construction managers from obtaining timely and accurate information on construction progress and quality.

[0003] Traditional systems also lack networking flexibility. In mountainous substation construction scenarios, when the construction area expands or equipment locations change, traditional fixed networking models struggle to adapt quickly. Adding a temporary cable-laying monitoring point requires re-laying numerous lines and performing complex equipment commissioning, consuming significant time and labor. This can result in the new monitoring point being unable to connect to the system in a timely manner, creating blind spots and preventing comprehensive, real-time monitoring of the entire construction process. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a weak signal multi-module networking adaptive monitoring system for power transmission and transformation construction, which realizes stable transmission in weak signal environment and dynamic adaptive adjustment of networking topology through multi-module collaboration, thereby improving the real-time performance of full-process monitoring of power transmission and transformation construction.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0006] In a first aspect, a weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction includes:

[0007] Multimodal module, used to deploy multimodal sensing units at key monitoring points of power transmission and transformation equipment, collect vibration, partial discharge, leakage current and environmental parameter signals, and process the signals through electromagnetic shielding enclosure and built-in harmonic filtering circuit to obtain original monitoring signal data;

[0008] The detection module is used to input the original monitoring signal data into the chaos-enhanced weak signal detection device, and extract the fault feature vector by capturing the attractor phase diagram mutation and the time domain waveform threshold;

[0009] The transmission module is used to transmit the fault feature vector through a wired and wireless dual-channel hot standby transmission device. When the signal attenuation reaches a preset threshold, the transmission device automatically switches the transmission channel and triggers the relay node to perform gain compensation for the fault feature vector.

[0010] The planning module is used to plan the transmission path of the fault feature vector data using a clustered multi-hop self-organizing network architecture. The cluster head node implements autonomous repair of wireless mesh network breakpoints based on a dynamic pathfinding strategy and transmits the fault feature vector data to the sink node.

[0011] The diagnosis module is used to perform spatiotemporal correlation fusion of multi-source fault feature vectors based on the aggregation node, match the diagnosis results based on the pre-trained fault pattern library, dynamically adjust the sampling rate, transmission power and filtering parameters based on the diagnosis results, and generate alarm events;

[0012] The control module is used to automatically reconstruct the multi-module network topology based on fault diagnosis results and network status feedback, dynamically adjust the working mode of each sensor unit, and realize adaptive maintenance and adjustment of multi-module network monitoring.

[0013] Furthermore, the original monitoring signal data is input into the chaos-enhanced weak signal detection device, and the fault feature vector is extracted by capturing the attractor phase diagram mutation and the time domain waveform threshold, including:

[0014] Normalize the original monitoring signal data to eliminate the dimensional differences between different sensor signals and obtain standardized signal data with unified dimensions. The standardized signal data is then used as an external driving signal and input into a chaos detection structure with preset parameters.

[0015] Calculate the dynamic stability index of the chaos detection structure in real time, judge the changing trend of the current state based on the dynamic stability index, and track the attractor trajectory formed in the state space;

[0016] When it is detected that the attractor trajectory suddenly changes from a chaotic state to a regular periodic state, the moment of state transition is recorded. At the same time, the changes in signal amplitude and frequency before and after the transition are analyzed in the time domain waveform of the standardized signal.

[0017] The waveform segments that meet the preset amplitude threshold and frequency change characteristics are identified as signal segments with faults, and the potential fault signal segments are decomposed by wavelet packets to obtain the energy distribution in different frequency bands;

[0018] Based on the energy proportion of each frequency band, a frequency domain energy distribution vector is constructed to characterize the signal characteristics, and the frequency domain energy distribution vector is used as a fault feature vector reflecting the operating status of the equipment.

[0019] Furthermore, the fault feature vector is transmitted through a wired and wireless dual-channel hot standby transmission device. When the signal attenuation reaches a preset threshold, the transmission device automatically switches the transmission channel and triggers the relay node to perform gain compensation for the fault feature vector, including:

[0020] The main controller monitors the quality index of the currently activated main transmission channel. If the main channel is a wired channel, it monitors the bit error rate; if the main channel is a wireless channel, it monitors the received signal strength indicator value. When the quality index of the main channel is less than the corresponding preset threshold, it determines that the channel signal attenuation has reached the preset threshold and generates a channel switching instruction and a gain compensation trigger signal.

[0021] Based on the channel switching instruction and the gain compensation trigger signal, the transmission device performs a switching operation from the current active channel to the standby channel. At the same time, the gain compensation trigger signal activates the designated relay node on the transmission path to obtain the fault feature vector data packet of the transmission interruption caused by the disconnection of the active channel;

[0022] For the fault feature vector data packet, the signal attenuation trajectory of the current transmission link is predicted based on the signal attenuation data, and the compensation gain value required for power amplification of the fault feature vector data packet is dynamically calculated according to the signal attenuation rate.

[0023] Furthermore, for the fault feature vector data packet, the signal attenuation trajectory of the current transmission link is predicted based on the signal attenuation data, and the compensation gain value required for power amplification of the fault feature vector data packet is dynamically calculated according to the signal attenuation rate, including:

[0024] Obtain the channel attenuation data and environmental interference characteristics of the current transmission link. Based on the correlation between the time characteristics of the channel attenuation data and the environmental interference characteristics, predict the trend of signal attenuation amplitude changing with transmission time and distance.

[0025] The signal attenuation trend is combined with the real-time collected link impedance fluctuation parameters and instantaneous electromagnetic interference intensity values ​​to dynamically generate real-time signal attenuation rate prediction values ​​for each discrete location in the transmission path;

[0026] Based on the distribution of the predicted values ​​of the real-time signal attenuation rate, the transmission path of the fault feature vector data packet is divided into continuous segments according to the attenuation rate mutation points, and the corresponding theoretical signal attenuation degree at each segment node is calculated;

[0027] Compare the theoretical signal attenuation degree of the segmented node with the actual attenuation degree monitored by the sensor in real time, and obtain the attenuation degree deviation at each node;

[0028] According to the attenuation degree deviation and the ratio of the influence of the transmission distance of the corresponding section on the signal attenuation, the compensation gain value required for power amplification of the fault feature vector data packet is dynamically generated.

[0029] Furthermore, for the fault feature vector data, a clustered multi-hop self-organizing network architecture is used to plan the transmission path. The cluster head node implements autonomous repair of wireless mesh network breakpoints based on a dynamic pathfinding strategy and transmits the fault feature vector data to the sink node, including:

[0030] Receive the fault feature vector data packet after relay node gain compensation, and establish the topological connection structure between nodes in the cluster in the clustered multi-hop ad hoc network based on the node geographic location information and real-time link quality data required for data packet transmission to form an initial transmission path;

[0031] Based on the topological connection structure, the cluster head node periodically detects the connectivity status of each neighboring node and updates the network routing table in real time based on the detection results. For each path recorded in the table, the performance is evaluated based on the comprehensive signal strength and node residual energy.

[0032] When connectivity status monitoring indicates that there is a breakpoint in the current transmission path, the cluster head node initiates a dynamic pathfinding strategy to perform autonomous repair;

[0033] After the path repair is completed, the cluster head node relays the fault feature vector data to the sink node through the repaired multi-hop path.

[0034] Furthermore, based on the aggregation node, the multi-source fault feature vectors are temporally and spatially correlated and fused, and the diagnosis results are matched based on the pre-trained fault pattern library. The sampling rate, transmission power, and filtering parameters are dynamically adjusted according to the diagnosis results, and alarm events are generated at the same time, including:

[0035] The aggregation node extracts the timestamp and device location tag of each fault feature vector and aligns the signals from different devices using a sliding time window based on the timestamp and location tags. After alignment, the weighted evidence theory is used to fuse the fault features associated in the temporal and spatial dimensions to generate a fused feature vector.

[0036] The fused feature vector is input into the pre-trained fault pattern library for similarity matching. Based on the matching results, the equipment fault type and probability value are obtained. When the probability value is greater than the preset diagnostic threshold, a deterministic diagnostic result is generated.

[0037] Dynamically adjust parameters based on the diagnosis results. If the fault probability in the diagnosis results is greater than the preset probability threshold, the signal sampling rate of the relevant equipment is increased; if the current link quality index is less than the preset quality threshold, the node transmission power is increased; if a frequency band interference signal is detected in the fusion feature, the suppression parameter of the filter is adaptively enhanced;

[0038] The voltage signal in the fusion feature is monitored in real time. When the amplitude of the transient voltage mutation is detected to be greater than the safety limit, the hierarchical alarm event generation mechanism is immediately triggered.

[0039] Furthermore, the fused feature vector is input into the pre-trained fault pattern library for similarity matching. Based on the matching results, the equipment fault type and probability value are obtained. When the probability value is greater than the preset diagnostic threshold, a deterministic diagnostic result is generated, including:

[0040] For each fault mode feature vector in the pre-trained fault mode library, the Euclidean distance between the feature vector and the fused feature vector in the feature space is calculated to generate a distance measurement value set including the distance values ​​corresponding to all fault modes. The distance measurement value set is then normalized to obtain a normalized distance value set.

[0041] The standardized distance value set is input into the preset similarity converter. Through the built-in distance-probability inverse relationship of the converter, the matching probability value of the fused feature vector corresponding to each fault mode is dynamically generated;

[0042] Arrange the matching probability values ​​in descending order and compare them with the preset diagnostic thresholds one by one to determine the probability values ​​greater than the thresholds and the associated fault modes to form a set of candidate fault modes;

[0043] Extract the top-ranked fault mode in the candidate fault mode set, and analyze the equipment fault type code and affected component identification corresponding to the mode;

[0044] Based on the fault type code and the identification of the affected components, combined with the preset confidence grading rules, a deterministic diagnostic result is generated, including the specific fault location identification, fault type description and confidence rating.

[0045] Furthermore, based on fault diagnosis results and network status feedback, the multi-module network topology is automatically reconstructed, and the working mode of each sensor unit is dynamically adjusted to achieve adaptive maintenance and adjustment of multi-module network monitoring, including:

[0046] Based on the fault location identification and confidence rating in the deterministic diagnosis results, combined with real-time network status feedback data, including node connectivity status, link quality index, and node remaining energy, multi-mode network topology reconstruction instructions are generated;

[0047] According to the topology reconstruction instructions, cluster areas are re-divided and cluster head nodes are determined in the clustered multi-hop ad hoc network. Fault areas with confidence ratings exceeding a preset threshold are isolated, and redundant transmission paths are constructed to bypass the fault areas.

[0048] Based on the reconstructed topology and the impact range of the fault type in the deterministic diagnosis results, the working mode of the sensor units in the associated area is dynamically adjusted;

[0049] According to the link quality change trend in the network status feedback, the effectiveness of the reconstructed topology is periodically verified, and the adaptive operation status of the monitoring network is realized based on the verification results.

[0050] In a second aspect, a computing device includes:

[0051] one or more processors;

[0052] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the system.

[0053] According to a third aspect, a computer-readable storage medium stores a program, which implements the system when executed by a processor.

[0054] The above solution of the present invention includes at least the following beneficial effects:

[0055] By deploying multiple sensors in a multimodal manner to collect vibration and partial discharge signals, and utilizing a chaos-enhanced weak signal detection device to capture sudden changes in attractor phase diagrams and time-domain waveform thresholds, the system can accurately extract weak fault signatures and effectively identify potential faults at an early stage, reducing the probability of accidents compared to traditional monitoring methods. Dual-channel transmission utilizes a wired and wireless hot standby transmission mechanism. When signal attenuation reaches a preset threshold, the channel is automatically switched and relay nodes are triggered to perform gain compensation. Compensation gain is dynamically calculated based on signal attenuation predictions to ensure continuous and accurate data transmission, effectively preventing monitoring failures caused by transmission interruptions. Network planning utilizes a clustered multi-hop self-organizing architecture, with cluster head nodes autonomously repairing wireless mesh network outages using a dynamic pathfinding strategy. This system rapidly responds to network topology changes, automatically reconfiguring transmission paths around faulty areas, and reducing network recovery time to seconds, improving network stability and fault resilience. Fault diagnosis integrates spatiotemporal correlations between multi-source fault feature vectors, matches diagnostic results with a pre-trained fault pattern library, and generates a deterministic diagnostic report using Euclidean distance metric and probability conversion. Adaptive control automatically reconfigures the multi-module network topology and dynamically adjusts the operating mode of sensor units based on fault diagnosis and network status feedback. It can quickly adapt to equipment failures and environmental changes, improving the intelligence level and operation and maintenance efficiency of power transmission and transformation construction monitoring systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of a weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction provided by an embodiment of the present invention.

[0057] Figure 2It is a flowchart that inputs the fused feature vector into the pre-trained fault pattern library for similarity matching, and obtains the equipment fault type and probability value based on the matching results. When the probability value is greater than the preset diagnostic threshold, a deterministic diagnostic result is generated. DETAILED DESCRIPTION

[0058] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0059] like Figure 1 As shown, an embodiment of the present invention provides a weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction, comprising:

[0060] Multimodal module, used to deploy multimodal sensing units at key monitoring points of power transmission and transformation equipment, collect vibration, partial discharge, leakage current and environmental parameter signals, and perform anti-interference processing on the signals through electromagnetic shielding enclosure and built-in harmonic filtering circuit to obtain the original monitoring signal data;

[0061] The detection module is used to input the original monitoring signal data into the chaos-enhanced weak signal detection device, and extract the fault feature vector by capturing the attractor phase diagram mutation and the time domain waveform threshold;

[0062] The transmission module is used to transmit the fault feature vector through a wired and wireless dual-channel hot standby transmission device. When the signal attenuation reaches a preset threshold, the transmission device automatically switches the transmission channel and triggers the relay node to perform gain compensation for the fault feature vector.

[0063] The planning module is used to plan the transmission path of the fault feature vector data using a clustered multi-hop self-organizing network architecture. The cluster head node implements autonomous repair of wireless mesh network breakpoints based on a dynamic pathfinding strategy and transmits the fault feature vector data to the sink node.

[0064] The diagnosis module is used to perform spatiotemporal correlation fusion of multi-source fault feature vectors based on the aggregation node, match the diagnosis results based on the pre-trained fault pattern library, dynamically adjust the sampling rate, transmission power and filtering parameters based on the diagnosis results, and generate alarm events;

[0065] The control module is used to automatically reconstruct the multi-module network topology based on fault diagnosis results and network status feedback, dynamically adjust the working mode of each sensor unit, and realize adaptive maintenance and adjustment of multi-module network monitoring.

[0066] In an embodiment of the present invention, multimodal sensing units are deployed at key monitoring points of power transmission and transformation equipment to achieve synchronous acquisition of multi-dimensional signals such as vibration, partial discharge, leakage current and environmental parameters, comprehensively covering the key indicators of the equipment's operating status and avoiding the limitations of single signal monitoring. Anti-interference processing is performed through electromagnetic shielding shell packaging and built-in harmonic filtering circuits, reducing the influence of external electromagnetic interference and harmonic noise from the source of signal acquisition, and improving the purity and reliability of the original monitoring signal data. A chaos-enhanced weak signal detection device is adopted, utilizing the sensitivity of the chaotic system to weak signal changes, breaking through the accuracy bottleneck of traditional detection methods, and being able to capture extremely small abnormal signal changes during equipment operation. By combining the attractor phase diagram mutation with the time domain waveform threshold criterion, the fault feature vector can be effectively extracted. Even in the early stage of equipment failure, weak fault signals can be accurately identified, improving the timeliness and accuracy of fault warning.

[0067] A dual-channel hot standby transmission mechanism, combining wired and wireless channels, establishes redundant data transmission paths. When the quality of the primary transmission channel drops below a preset threshold due to signal attenuation or environmental interference, the transmission device automatically and rapidly switches to the backup channel, ensuring uninterrupted transmission of the fault signature vector. Simultaneously, relay nodes are triggered to perform data gain compensation, dynamically adjusting transmission power based on signal attenuation. This effectively overcomes distance limitations and attenuation during signal transmission, ensuring data transmission stability and integrity, and preventing monitoring failures caused by transmission interruptions. Transmission paths are planned based on a clustered multi-hop ad hoc network architecture, fully adapting to the widespread distribution and complex environments of power transmission and transformation equipment, allowing for flexible network topology construction. Cluster head nodes employ a dynamic routing strategy to monitor network node connectivity in real time. When a wireless mesh network breakpoint occurs, they rapidly and autonomously repair the path and automatically reroute data around the faulty area, reducing manual troubleshooting and intervention, improving the network's fault resilience and self-healing capabilities, and ensuring efficient and stable transmission of fault signature vector data to the sink node.

[0068] The multi-source fault feature vectors collected by the sink node are fused through temporal and spatial correlation, comprehensively considering signal information from different devices and across different temporal and spatial dimensions. This system deeply explores potential correlations between signals and avoids misjudgments and missed detections based on single-signal diagnosis. Combined with a pre-trained fault pattern library for matching diagnosis, the system can quickly and accurately determine the type, probability, and location of device faults. Based on the diagnostic results, the system dynamically adjusts the sampling rate, transmission power, and filtering parameters, achieving intelligent optimization of monitoring resources. Furthermore, it generates timely alarms to help maintenance personnel quickly locate faults and improve troubleshooting efficiency. Based on fault diagnosis results and network status feedback, the system automatically reconstructs the multi-module network topology, isolates faulty areas, re-divides clusters, identifies cluster heads, and constructs redundant transmission paths around faulty areas, enhancing network reliability and fault tolerance. The system dynamically adjusts the operating mode of each sensor unit, such as increasing the sensor sampling frequency in high-fault areas and optimizing transmission strategies during network congestion. This enables the entire multi-module network monitoring system to rapidly adapt to device failures, environmental changes, and network status fluctuations, achieving adaptive maintenance and adjustment.

[0069] In a preferred embodiment of the present invention, the original monitoring signal data is input into the chaos-enhanced weak signal detection device, and the fault feature vector is extracted by capturing the attractor phase diagram mutation and the time domain waveform threshold, which may include:

[0070] Normalize the original monitoring signal data to eliminate the dimensional differences between different sensor signals and obtain standardized signal data with unified dimensions. The standardized signal data is then used as an external driving signal and input into a chaos detection structure with preset parameters.

[0071] Calculate the dynamic stability index of the chaos detection structure in real time, judge the changing trend of the current state based on the dynamic stability index, and track the attractor trajectory formed in the state space;

[0072] When it is detected that the attractor trajectory suddenly changes from a chaotic state to a regular periodic state, the moment of state transition is recorded. At the same time, the changes in signal amplitude and frequency before and after the transition are analyzed in the time domain waveform of the standardized signal.

[0073] The waveform segments that meet the preset amplitude threshold and frequency change characteristics are identified as signal segments with faults, and the potential fault signal segments are decomposed by wavelet packets to obtain the energy distribution in different frequency bands;

[0074] Based on the energy proportion of each frequency band, a frequency domain energy distribution vector is constructed to characterize the signal characteristics, and the frequency domain energy distribution vector is used as a fault feature vector reflecting the operating status of the equipment.

[0075] In the embodiment of the present invention, the original monitoring signal data is collected by vibration, partial discharge, and leakage current multimodal sensors. The signals have different dimensions due to different physical quantities (such as vibration is acceleration m / s², current is ampere A). In order to unify the processing, the linear normalization method is used to process the original signal of each sensor. Calculate. Assume that the minimum value of a sensor signal is , the maximum value is , the normalized formula is , mapping the signal to the interval [0, 1]. Through calculation, dimensionality effects are eliminated, making signals from different sensors comparable. Standardized signal data is generated, which serves as the external driving signal for the chaos detection structure. The chaos detection structure with preset parameters has a specific dynamic equation (such as the Lorentz equation, a chaotic system equation), and its state changes over time. To determine the trend of state changes, dynamic stability indicators are calculated in real time. For example, the Lyapunov exponent describes the average exponential divergence or convergence rate of adjacent trajectories in phase space. During the calculation, the dynamic equation of the chaos detection structure is linearized, and the Jacobian matrix is ​​iteratively calculated to obtain the Lyapunov exponent. A positive exponent indicates a chaotic state, with adjacent trajectories exponentially separating over time. If the exponent approaches zero, a transition to a stable state is indicated. Simultaneously, the attractor trajectory pattern is tracked in state space. The attractor is the final form of the system's long-term evolution, reflecting the stable state or trend of change.

[0076] Continuously monitor the attractor trajectory. When it is detected that it suddenly changes from a chaotic state (the trajectory presents an irregular, complex entangled shape) to a regular periodic state (the trajectory presents a simple shape that repeats periodically), record the moment of the state transition. In the time domain waveform of the standardized signal, focus on the signal before and after the transition moment and analyze the changes in amplitude and frequency. The amplitude is calculated using the root mean square (RMS) value, and the formula is: ,in is the number of sampling points, For the The signal value of each sampling point; the frequency analysis uses fast Fourier transform (FFT) to convert the time domain signal into the frequency domain to obtain the frequency components and corresponding amplitudes of the signal. According to the preset amplitude threshold and frequency change characteristics (such as the amplitude suddenly rises above the threshold, or the specific frequency component is enhanced), the signal segment with faults is identified. Wavelet packet decomposition is performed on the potential fault signal segment. Wavelet packet decomposition is a method of decomposing the signal into different frequency bands. It gradually subdivides the signal into narrower frequency bands through multi-layer decomposition. For example, a three-layer wavelet packet decomposition can decompose the signal into 8 different frequency bands. During the decomposition process, the convolution operation is performed on the wavelet function and the signal to obtain the coefficients of each frequency band, and then the energy distribution of different frequency bands is obtained. The energy is calculated as the sum of the squares of the frequency band coefficients. Based on the energy proportion of each frequency band, a frequency domain energy distribution vector is constructed. Calculate the proportion of the energy of each frequency band to the total energy. Assuming that there are frequency band, The energy of the frequency band is , total energy , then The energy proportion of each frequency band is , arrange these energy proportions in frequency band order to form a frequency domain energy distribution vector , which serves as the fault feature vector reflecting the operating status of the equipment.

[0077] Normalization eliminates dimensional differences, ensuring that multimodal signals are analyzed at the same scale, avoiding analytical bias caused by dimensional differences and laying the foundation for accurate fault feature extraction. The chaos detection structure is sensitive to weak signal changes. By combining dynamic stability indicators such as the Lyapunov exponent to analyze sudden changes in attractor trajectories, it can capture early equipment fault signals that are difficult to detect using traditional methods, improving the accuracy of fault feature extraction. Fault diagnosis is performed by combining sudden changes in attractor phase diagrams with amplitude and frequency changes in the time domain waveform. Signal analysis from both phase space and time domain dimensions avoids the limitations of single-dimensional analysis, improving the comprehensiveness and accuracy of fault diagnosis and effectively reducing the probability of false positives and missed detections. Wavelet packet decomposition adaptively decomposes signals into different frequency bands. Taking into account the complex frequency components of fault signals from power transmission and transformation equipment, it comprehensively captures the energy distribution of each frequency band and accurately characterizes the fault signal. The constructed frequency domain energy distribution vector contains rich signal feature information. Real-time calculation of dynamic stability indicators and tracking of attractor trajectories enables timely detection of fault signal changes and enables real-time fault monitoring. Through a rigorous calculation process and multi-step verification (such as amplitude threshold judgment and frequency analysis), the extracted fault feature vector is ensured to be true and reliable, providing a solid monitoring guarantee for the stable operation of power transmission and transformation equipment, and reducing the economic losses and safety risks caused by equipment failures.

[0078] In a preferred embodiment of the present invention, the fault feature vector is transmitted through a wired and wireless dual-channel hot standby transmission device. When the signal attenuation reaches a preset threshold, the transmission device automatically switches the transmission channel and triggers the relay node to perform gain compensation for the fault feature vector. This may include:

[0079] The main controller monitors the quality index of the currently activated main transmission channel. If the main channel is a wired channel, it monitors the bit error rate; if the main channel is a wireless channel, it monitors the received signal strength indicator value. When the quality index of the main channel is less than the corresponding preset threshold, it determines that the channel signal attenuation has reached the preset threshold and generates a channel switching instruction and a gain compensation trigger signal.

[0080] Based on the channel switching instruction and the gain compensation trigger signal, the transmission device performs a switching operation from the current active channel to the standby channel. At the same time, the gain compensation trigger signal activates the designated relay node on the transmission path to obtain the fault feature vector data packet of the transmission interruption caused by the disconnection of the active channel;

[0081] For the fault feature vector data packet, the signal attenuation trajectory of the current transmission link is predicted based on the signal attenuation data, and the compensation gain value required for power amplification of the fault feature vector data packet is dynamically calculated based on the signal attenuation rate. Specifically, the following steps are performed: obtaining the channel attenuation data and environmental interference characteristics of the current transmission link, and predicting the trend of signal attenuation amplitude changing with transmission time and distance based on the correlation between the time characteristics of the channel attenuation data and the environmental interference characteristics; combining the signal attenuation trend with the real-time collected link impedance fluctuation parameters and the instantaneous value of the electromagnetic interference intensity to dynamically generate the real-time signal attenuation rate prediction value for each discrete location in the transmission path; based on the distribution of the real-time signal attenuation rate prediction value, the transmission path of the fault feature vector data packet is divided into continuous segments according to the attenuation rate mutation points, and the corresponding theoretical signal attenuation degree at each segment node is calculated; the theoretical signal attenuation degree of the segment node is compared with the actual attenuation degree monitored in real time by the sensor to obtain the attenuation degree deviation at each node; based on the attenuation degree deviation, combined with the proportion of the influence of the transmission distance of the corresponding segment on the signal attenuation, the compensation gain value required for power amplification of the fault feature vector data packet is dynamically generated.

[0082] In the embodiment of the present invention, the main controller monitors the fault feature vector data transmitted by the wired channel in real time. ), count the number of packets with errors (set as The formula for calculating the bit error rate is For example, if 1000 data packets are transmitted and 5 of them have errors, the bit error rate is =0.5%. When the calculated bit error rate is less than the preset wired channel bit error rate threshold (e.g., 1%), the wired channel signal attenuation is determined to have reached the preset threshold. For wireless channels, the main controller continuously reads the received signal strength indicator (RSSI). The RSSI value reflects the strength of the received signal and is typically measured in dBm. For example, the RSSI value read at a given moment is -70dBm. The real-time RSSI value is compared with the preset wireless channel RSSI threshold (e.g., -80dBm). When the RSSI value falls below this threshold, the wireless channel signal attenuation is determined to have reached the preset threshold. Once the quality indicator of the active channel meets the threshold, the main controller immediately generates a channel switching command and a gain compensation trigger signal.

[0083] After the transmission device receives the channel switching instruction, it first stops the data transmission operation on the main channel. Then, it performs the initialization settings for channel switching, including configuring the communication parameters of the backup channel (such as the frequency band of the wireless channel, the port protocol of the wired channel, etc.), and establishing a connection with the backup channel. After the connection is established, the transmission device switches the fault feature vector data packet to the backup channel for transmission. The gain compensation trigger signal is sent to the preset designated relay node on the transmission path. After the relay node receives the trigger signal, it starts the data receiving program and begins to obtain the fault feature vector data packet whose transmission is interrupted due to the disconnection of the main channel. The relay node will cache the received data packet and wait for gain compensation processing. The main controller collects the channel attenuation data of the current transmission link (such as the bit error rate or RSSI value change records in different time periods) and environmental interference characteristics (such as the distribution of surrounding electromagnetic equipment, weather conditions and other environmental factors that affect signal transmission). By analyzing the time series characteristics of the channel attenuation data, for example, using the moving average method, the past period of time (set as ) to predict how signal attenuation changes over transmission time. The prediction is then adjusted based on environmental interference characteristics. For example, if strong electromagnetic equipment is detected nearby, the predicted signal attenuation is appropriately increased. Furthermore, considering the impact of transmission distance on signal attenuation, the predicted attenuation value is adjusted based on the length of the transmission path, resulting in a comprehensive trend in signal attenuation over transmission time and distance.

[0084] The signal attenuation trend is combined with real-time collected link impedance fluctuation parameters (reflecting changes in the electrical characteristics of the transmission line) and the instantaneous electromagnetic interference intensity. For example, a sudden increase in link impedance may indicate poor contact in the transmission line, leading to increased signal attenuation. In this case, the predicted signal attenuation rate is increased accordingly based on the magnitude of the impedance change and empirical data. Similarly, the predicted signal attenuation rate is adjusted as the instantaneous electromagnetic interference intensity increases. This approach dynamically generates real-time signal attenuation rate predictions for each discrete point in the transmission path. Based on the distribution of the real-time signal attenuation rate predictions, the transmission path of the fault feature vector data packet is divided into continuous segments based on the attenuation rate mutation points. For each segment node, the theoretical signal attenuation level is calculated based on factors such as the segment length, the predicted attenuation rate, and the initial signal strength. Simultaneously, sensors monitor the actual attenuation level at each node in real time. The theoretical signal attenuation level at each segment node is compared with the actual attenuation level to determine the attenuation deviation at each node. The system dynamically generates the compensation gain required to amplify the power of the fault signature vector data packets based on the attenuation deviation and the influence of the transmission distance on signal attenuation in the corresponding section (for example, the longer the distance, the greater the weight of the distance factor on attenuation). For example, if the theoretical attenuation in a section is 10dB, the actual attenuation is 12dB, the deviation is 2dB, and the distance is long, with a distance factor weight of 60%, then the preset calculation rules will determine the compensation gain value to be 3dB to ensure sufficient signal strength during transmission.

[0085] The dual-channel hot standby transmission mechanism provides redundant transmission paths for fault feature vector data. When the quality of the primary channel degrades due to signal attenuation, equipment failure, or environmental interference, it automatically and rapidly switches to the backup channel to avoid data interruption. This ensures continuous and stable data transmission even in complex and volatile power transmission and transformation environments (such as strong electromagnetic interference, severe weather affecting wireless signals, or physical damage to wired lines). By monitoring the quality indicators of the primary channel in real time and making dynamic judgments and decisions based on preset thresholds, the system intelligently adjusts the transmission strategy based on actual transmission conditions. Whether it's an abnormal bit error rate in the wired channel or insufficient signal strength in the wireless channel, the corresponding response mechanism is triggered immediately. The relay node's gain compensation function dynamically adjusts the compensation gain based on the signal attenuation trajectory and real-time attenuation rate, precisely matching the signal enhancement requirements of different transmission sections. This effectively overcomes distance limitations and attenuation issues during signal transmission, ensuring high data quality over long distances and in complex environments, and improving transmission stability and efficiency. Automated channel switching and gain compensation eliminate the need for manual detection and switching of transmission channels or manual calculation and adjustment of signal gain. When problems arise in the transmission channel, the system can autonomously respond to and repair the fault, reducing the workload and maintenance costs for operations personnel. Furthermore, the rapid fault response mechanism shortens data transmission interruptions and mitigates the impact of data loss or transmission delays on power transmission and transformation equipment monitoring and fault diagnosis. In power transmission and transformation environments, electromagnetic interference and line aging can easily lead to signal attenuation and transmission failures. Dual-channel transmission and a dynamic gain compensation mechanism effectively address these interference and failures. The presence of backup channels provides fault tolerance, while gain compensation at relay nodes enhances resistance to signal attenuation. Even in the event of local transmission link problems, channel switching and gain adjustment can maintain data transmission, ensuring the normal operation of the monitoring system and improving the anti-interference and fault tolerance performance of the weak-signal multi-mode network adaptive monitoring system for power transmission and transformation.

[0086] In a preferred embodiment of the present invention, a transmission path is planned for fault feature vector data using a clustered multi-hop ad hoc network architecture. The cluster head node implements autonomous repair of wireless mesh network breakpoints based on a dynamic pathfinding strategy and transmits the fault feature vector data to a sink node. This may include:

[0087] Receive the fault feature vector data packet after relay node gain compensation, and establish the topological connection structure between nodes in the cluster in the clustered multi-hop ad hoc network based on the node geographic location information and real-time link quality data required for data packet transmission to form an initial transmission path;

[0088] Based on the topological connection structure, the cluster head node periodically detects the connectivity status of each neighboring node and updates the network routing table in real time based on the detection results. For each path recorded in the table, the performance is evaluated based on the comprehensive signal strength and node residual energy.

[0089] When connectivity status monitoring indicates that there is a breakpoint in the current transmission path, the cluster head node initiates a dynamic pathfinding strategy to perform autonomous repair;

[0090] After the path repair is completed, the cluster head node relays the fault feature vector data to the sink node through the repaired multi-hop path.

[0091] In this embodiment of the present invention, a cluster head node receives a gain-compensated fault feature vector data packet from a relay node. This data packet contains the data itself and accompanying metadata, such as the sending node's geographic coordinates (latitude and longitude) and the packet generation time. The cluster head node parses this information to extract the node's geographic location information required for packet transmission. The cluster head node sends a probe signal (such as a low-power broadcast signal) to each node in the cluster. After receiving the probe signal, each node calculates the link quality indicator between itself and the cluster head node based on its received signal strength and signal-to-noise ratio. Common methods for calculating link quality indicators include using the received signal strength indicator (RSSI), mapping the RSSI value to a quality score range of 0-100. For example, an RSSI value between -50dBm and -70dBm corresponds to a score of 80-60. Nodes then feed the calculated link quality score back to the cluster head node.

[0092] Based on the extracted node location information and link quality data, the cluster head node uses graph theory to construct a topological connectivity structure between nodes within the cluster. Each node is considered a vertex in the graph, and connections between nodes are considered edges. The edge weight is determined by the link quality score; a higher score indicates a lower weight (indicating better link quality). Using a shortest path algorithm, the shortest path from each node to the cluster head node is calculated (based on the link quality weights), thus forming the initial transmission path. For example, if there are three nodes in a cluster, H, M, and G, and the link quality scores between H and the cluster head node are 80, M is 70, and G is 60, the algorithm can determine the transmission paths from H to the cluster head node, MH to the cluster head node, and GMH to the cluster head node. The cluster head node periodically (e.g., every 10 seconds) sends connectivity probe packets (containing simple identification information) to neighboring nodes. Upon receiving the probe packets, the neighboring nodes immediately reply with acknowledgment packets. The cluster head node determines the connectivity status of the neighboring node based on whether or not it receives the acknowledgment packet. If no acknowledgment is received within a specified time (e.g., 2 seconds), the neighboring node is considered disconnected.

[0093] Based on connectivity detection results, the cluster head node updates the network routing table in real time. The routing table records information about each reachable node, including its node ID, geographic location, link quality score with the current node, and next-hop node information. When a change in a node's connectivity is detected, the corresponding node's information is updated. If a new reachable node is discovered, a new record is added; if a node is disconnected, the node record is deleted or marked. For each path recorded in the routing table, the cluster head node evaluates the performance of both signal strength and node remaining energy. Signal strength is represented by a link quality score, while node remaining energy is measured by the battery charge percentage regularly reported by the node. Weighting can be adjusted to prioritize signal strength or node remaining energy based on actual needs. Paths with higher scores are associated with better performance. During data transmission, if the cluster head node does not receive an acknowledgment from a next-hop node within a specified time after sending a data packet to it, and if there is still no response after multiple retransmissions (e.g., three times), the current transmission path is considered to have a breakpoint. Upon detecting a breakpoint, the cluster head node initiates a dynamic routing strategy. First, it selects candidate nodes from the network routing table that are adjacent to the breakpoint node and are connected. Then, with the cluster head node as the starting point, candidate nodes as intermediate nodes, and the sink node as the end point, the shortest path algorithm (based on updated link quality weights and node residual energy weights) is used again to calculate a new path from the cluster head node to the sink node. During the calculation process, path segments with confirmed faults are excluded, and paths with higher performance evaluation metrics are prioritized. After the path is repaired, the cluster head node relays the fault feature vector data along the repaired multi-hop path. The cluster head node sends the data packet to the next hop node on the new path. After receiving the data packet, the node forwards it to the next hop node based on its routing table information, and the relay continues until the data packet reaches the sink node. When forwarding a data packet, each node checks the integrity of the data packet (for example, by calculating a checksum). If the data packet is found to be corrupted, it is discarded and the previous hop node is notified to retransmit it.

[0094] The clustered multi-hop ad hoc network architecture effectively addresses the complex and ever-changing network conditions in power transmission and transformation environments by establishing an intra-cluster topological connection structure, combined with a dynamic routing strategy and path repair mechanism. When a transmission path breaks, it can rapidly and autonomously repair the path, preventing data transmission interruptions caused by single-point failures. Even in harsh environments (such as strong electromagnetic interference causing signal loss at some nodes), network connectivity is maintained, ensuring stable transmission of fault feature vector data to the aggregation node, enhancing network stability and reliability. Path performance is evaluated based on a comprehensive analysis of signal strength and node residual energy. When planning transmission paths, paths with good signal quality and sufficient node energy are selected to avoid overuse of low-energy nodes, effectively balancing the load across the network. Cluster head nodes periodically detect the connectivity status of neighboring nodes and update the routing table in real time, enabling the network to rapidly detect environmental changes (such as equipment relocation, new node additions, or existing node exits). The dynamic routing strategy ensures that the network can promptly adjust transmission paths based on environmental changes, adapting to the complex and ever-changing power transmission and transformation field environments without manual intervention. This enhances the network's adaptability and flexibility, meeting the monitoring requirements of widely distributed equipment and complex environments in power transmission and transformation construction. The clustered multi-hop transmission method breaks down long-distance data transmission into multiple shorter relay transmissions. Combined with dynamic pathfinding to select the final path, it effectively reduces data transmission latency. Furthermore, the rapid response mechanism of path repair enables data to be quickly switched to a new path in the event of a fault, avoiding long wait times and data accumulation. This ensures the real-time nature of monitoring data and provides strong support for rapid fault diagnosis and timely resolution of power transmission and transformation equipment. The clustered multi-hop ad hoc network architecture facilitates network expansion and upgrades.

[0095] In a preferred embodiment of the present invention, a multi-source fault feature vector is temporally and spatially correlated and fused according to the sink node, and the diagnosis results are matched based on the pre-trained fault pattern library. The sampling rate, transmission power, and filtering parameters are dynamically adjusted according to the diagnosis results, and an alarm event is generated. The following may be included:

[0096] The aggregation node extracts the timestamp and device location tag of each fault feature vector and aligns the signals from different devices using a sliding time window based on the timestamp and location tags. After alignment, the weighted evidence theory is used to fuse the fault features associated in the temporal and spatial dimensions to generate a fused feature vector.

[0097] The fused feature vector is input into the pre-trained fault pattern library for similarity matching, and the equipment fault type and probability value are obtained based on the matching results. When the probability value is greater than the preset diagnostic threshold, a deterministic diagnostic result is generated, specifically including: for each fault mode feature vector in the pre-trained fault pattern library, the Euclidean distance between the fused feature vector and the feature vector is calculated in the feature space, a distance measurement value set including the distance values ​​corresponding to all fault modes is generated, and the distance measurement value set is normalized to obtain a normalized distance value set; the normalized distance value set is input into the preset similarity converter, and the matching probability value of the fused feature vector corresponding to each fault mode is dynamically generated through the built-in distance-probability inverse relationship of the converter;

[0098] The matching probability values ​​are sorted in descending numerical order and compared with the preset diagnostic threshold one by one to determine the probability values ​​greater than the threshold and the associated fault modes to form a set of candidate fault modes. The fault mode ranked first in the candidate fault mode set is extracted, and the equipment fault type code and affected component identification corresponding to the mode are analyzed. Based on the fault type code and affected component identification, combined with the preset confidence grading rules, a deterministic diagnostic result is generated, including the specific fault location identification, fault type description, and confidence rating.

[0099] Dynamically adjust parameters based on the diagnosis results. If the fault probability in the diagnosis results is greater than the preset probability threshold, the signal sampling rate of the relevant equipment is increased; if the current link quality index is less than the preset quality threshold, the node transmission power is increased; if a frequency band interference signal is detected in the fusion feature, the suppression parameter of the filter is adaptively enhanced;

[0100] The voltage signal in the fusion feature is monitored in real time. When the amplitude of the transient voltage mutation is detected to be greater than the safety limit, the hierarchical alarm event generation mechanism is immediately triggered.

[0101] In an embodiment of the present invention, a convergence node receives fault feature vectors from different devices. Each vector records the signal acquisition time (accurate to milliseconds) and the device installation location (including the device number and geographic coordinates). The convergence node analyzes the data format and extracts this time and location information separately to prepare for subsequent unified processing. First, the length of the time window (for example, 500 milliseconds) and the interval between each slide (for example, 100 milliseconds) are set. Based on the signal acquisition time, the signals collected by different devices in the same time window are treated as data at the same moment. If the acquisition time of a device signal is inconsistent with the window start time, the signal is matched to the appropriate time point in the window through reasonable estimation to achieve time alignment of the signals of different devices. At the same time, based on the device location information, the signals of devices with similar locations are associated to ensure consistency in the time and space dimensions.

[0102] Each device's fault signature is assigned a weight based on its importance (for example, devices on critical transmission lines are more important) and the accuracy of signal acquisition (high-precision sensors provide more reliable signals). The sum of all device weights equals 1. Then, the values ​​for each device in each feature dimension (such as vibration frequency and partial discharge intensity) are summed according to the weights to obtain a fused feature value. These fused feature values ​​are combined to form a fused feature vector. The fused feature vector is then compared with each fault mode feature vector in the pre-trained fault mode library to determine their similarity. This comparison is performed by calculating the distance between the two. A closer distance indicates a higher similarity. The calculated distances are first adjusted so that all distances are within the range of 0 to 1. Then, based on the correspondence between distance and probability, fault modes with closer distances are assigned a higher matching probability.

[0103] The calculated matching probability values ​​are sorted from highest to lowest and compared against a set diagnostic threshold (e.g., 0.7). Fault modes with probability values ​​exceeding the threshold are screened out to form a candidate list. From this list, the fault mode with the highest probability is selected and its corresponding device fault type and affected components are analyzed. Combined with a pre-set confidence level (e.g., probabilities between 0.7 and 0.8 are considered low confidence, 0.8 to 0.9 are considered medium confidence, and 0.9 and above are considered high confidence), a diagnostic result is generated that includes the specific fault location, a description of the fault type, and a confidence rating. The fault probability in the diagnostic result is compared against the set probability threshold (e.g., 0.6). If the fault probability exceeds the threshold, the risk of device failure is high, and more detailed signal information is required. In this case, the signal sampling frequency of the relevant device is increased by a certain percentage, and the new sampling frequency setting is sent to the device. Upon receipt, the device automatically adjusts its sampling frequency. The current signal transmission quality is monitored in real time (using comprehensive evaluation metrics such as bit error rate and signal strength) and compared against the set quality threshold (e.g., 80 points). If the transmission quality is lower than the standard, it means that the signal transmission effect is not good. According to the degree of quality degradation, the node's transmission power is increased proportionally, and an adjustment instruction is sent to the node. The node will increase the signal transmission power according to the instruction.

[0104] The fused feature vector is analyzed to check for frequency interference signals (the interference frequency range is set and the signal strength within this range is determined to be abnormal). If interference is detected, the filter's suppression capability for the interfering frequency band is proportionally increased based on the severity of the interference. The new parameter settings are then sent to the filter, which then filters out more interference signals. The voltage signal portion of the fused feature vector is extracted in real time, and voltage values ​​at adjacent moments are compared to calculate the magnitude of the voltage signal change. The calculated voltage change is compared with a set safety standard (such as 20% of the rated voltage). If the change exceeds the standard, an alarm of varying severity is triggered based on the degree of the excess.

[0105] By simultaneously considering the temporal and spatial information of multiple device signals, the system avoids the limitations of relying solely on single device signals for diagnosis, improving the accuracy and comprehensiveness of fault diagnosis. Fusion methods assign weights based on device importance and signal quality, making diagnostic results more reliable. Automatically adjusting the sampling rate, transmission power, and filtering parameters based on diagnostic results ensures optimal resource allocation. Automatically increasing the signal acquisition frequency when the risk of device failure is high, increasing the transmission power when signal transmission quality is poor, and optimizing the filtering effect when interference occurs, these adjustments are tailored to actual operating conditions, improving monitoring efficiency and data quality while avoiding resource waste. Real-time monitoring of voltage signals within the fused features quickly detects abnormal voltage changes and issues alerts of varying levels based on the magnitude of the change. This hierarchical alerting approach allows operations and maintenance personnel to quickly determine the severity of faults, prioritize high-risk faults, shorten troubleshooting time, reduce the risk of equipment damage or grid accidents caused by voltage anomalies, and ensure the safe and stable operation of the power transmission and transformation system. The combination of multi-source data fusion and dynamic parameter adjustment enables adaptation to complex and changing power transmission and transformation operating environments. Regardless of equipment failure, network fluctuation or environmental interference, it can maintain stable operation through its own diagnosis and adjustment mechanism, reduce the impact of external factors on monitoring and diagnosis results, enhance anti-interference ability and stability, and improve the reliability and availability of the entire monitoring system.

[0106] In a preferred embodiment of the present invention, based on fault diagnosis results and network status feedback, the multi-mode network topology is automatically reconstructed, the working mode of each sensor unit is dynamically adjusted, and adaptive maintenance and adjustment of multi-mode network monitoring are achieved, which may include:

[0107] Based on the fault location identification and confidence rating in the deterministic diagnosis results, combined with real-time network status feedback data, including node connectivity status, link quality index, and node remaining energy, multi-mode network topology reconstruction instructions are generated;

[0108] According to the topology reconstruction instructions, cluster areas are re-divided and cluster head nodes are determined in the clustered multi-hop ad hoc network. Fault areas with confidence ratings exceeding a preset threshold are isolated, and redundant transmission paths are constructed to bypass the fault areas.

[0109] Based on the reconstructed topology and the impact range of the fault type in the deterministic diagnosis results, the working mode of the sensor units in the associated area is dynamically adjusted;

[0110] According to the link quality change trend in the network status feedback, the effectiveness of the reconstructed topology is periodically verified, and the adaptive operation status of the monitoring network is realized based on the verification results.

[0111] In this embodiment of the present invention, the deterministic diagnostic results output by the fault diagnosis module are obtained, and the fault location identifier (such as device number or geographic coordinates) and confidence rating (indicating the reliability of the diagnostic result, with a value range of 0-1) are extracted. Simultaneously, network status feedback data is collected in real time, including the connectivity status of each node (determining whether a node is online by sending a probe signal and receiving a reply), the link quality index (calculated based on parameters such as bit error rate, signal strength, and signal-to-noise ratio, with higher values ​​indicating better link quality), and the node's remaining energy (determined by the node's periodic reporting of battery charge percentage or remaining power capacity).

[0112] If the fault confidence rating in a certain area exceeds a preset threshold (e.g., 0.8), the connectivity of nodes in the area deteriorates, the link quality index drops to a certain level (e.g., below 60), and the remaining energy of nodes is low (e.g., less than 30%), then the area is determined to require topology reconstruction. Based on these conditions, a multi-module network topology reconstruction instruction is generated, including information such as the fault area scope and reconstruction objectives (e.g., isolating the faulty area and optimizing transmission paths).

[0113] Based on a clustered multi-hop ad hoc network architecture, clusters are replanned with the fault area as the center, taking into account the geographic distribution of network nodes. A distance metric is used to calculate the distance from each node to the fault area boundary. Nodes that are farther from the fault area and have good interconnectedness are grouped into new clusters. For example, a distance threshold (e.g., 50 meters) is set, and nodes whose distance from the fault area exceeds this threshold and whose inter-node link quality scores exceed a certain threshold (e.g., 70 points) are grouped into the same cluster. For each newly divided cluster, a cluster head node is elected. The election criteria comprehensively consider factors such as node remaining energy (higher energy means higher priority), processing power (e.g., CPU performance and memory capacity), and link quality (average link quality score compared to other nodes in the cluster). The cluster head node is determined through a voting mechanism or a weighted scoring method, for example, with a node remaining energy score of 40%, processing power of 30%, and link quality of 30%. A comprehensive score is calculated for each node, and the node with the highest score is elected as the cluster head. For fault areas whose confidence ratings exceed a preset threshold, the links connecting them to other healthy areas are severed to achieve physical isolation. At the same time, based on the remaining healthy nodes, a graph theory path search algorithm (such as a variant of Dijkstra's algorithm) is used to construct multiple redundant transmission paths from the source node (a healthy node near the fault area) to the destination node (a sink node or other key node), avoiding the fault area. During the construction process, paths with good link quality and high node residual energy are prioritized.

[0114] Based on the fault type in the deterministic diagnosis results, determine its impact scope. For example, if the fault type is a partial discharge fault, analyze the potentially affected equipment and surrounding areas. If it is a network link fault, determine the affected transmission paths and related nodes. For sensor units within the associated area, adjust their operating mode based on the fault's impact and network status. If the fault's impact is large, increase the sensor unit's sampling frequency (for example, from once per minute to five times per minute) to collect data more frequently. If the network link quality is poor, reduce the sensor unit's data transmission frequency (for example, from once per second to once every five seconds) to reduce data transmission pressure. For sensor units near the fault area but not directly affected, activate early warning mode to enhance detection sensitivity for abnormal signals. Based on link quality trends in network status feedback, periodically (for example, every 10 minutes) collect data such as bit error rate, signal strength, and transmission delay for each transmission path. Calculate a comprehensive performance index for each path, weighted by 30% for bit error rate, 40% for signal strength, and 30% for transmission delay. The calculated comprehensive performance indicators are compared with pre-set performance standards (e.g., a comprehensive score of 75 or above). If the comprehensive performance indicators of a particular path or the entire reconstructed topology do not meet the standards, this indicates a problem with the reconstructed topology. In this case, the network status and fault conditions are reassessed and the topology reconstruction process is repeated, adjusting the cluster area division, cluster head node selection, and redundant path construction strategies until the reconstructed topology meets the performance requirements and achieves adaptive operation of the monitoring network.

[0115] By automatically reconfiguring the multi-module network topology, fault areas are promptly isolated and redundant transmission paths are established, effectively preventing fault propagation and network paralysis. Even in the event of emergencies such as equipment failures or link interruptions in the power transmission and transformation environment, the network structure can be rapidly adjusted to maintain data transmission continuity and ensure stable collection and transmission of monitoring data. The operating mode of sensor units is dynamically adjusted to rationally allocate resources based on the fault impact scope and network status. The sampling frequency is increased in high-fault areas to ensure detailed fault information, while the transmission frequency is reduced during periods of network congestion or poor link quality to reduce network burden. This intelligent resource allocation approach not only improves the validity of monitoring data, but also extends the service life of sensor units and network nodes, reducing overall energy consumption and operation and maintenance costs. Based on real-time network status feedback and fault diagnosis results, the system can rapidly detect and respond to changes in the power transmission and transformation environment (such as equipment failures or environmental interference leading to decreased link quality). Periodically verifying the effectiveness of the reconfigured topology ensures that the monitoring network always maintains its final operating state, enhancing its adaptability to complex and changing environments and meeting the requirements for long-term stable operation of power transmission and transformation projects. The isolation mechanism for high-confidence fault areas effectively prevents the spread of faults and reduces the impact of faults on the entire network. Furthermore, the establishment of redundant transmission paths ensures that data can still be transmitted through other paths even if some links fail, shortening network recovery time. This adaptive maintenance and adjustment mechanism automates and intelligentizes network management, reducing the workload of manual intervention and configuration. Operations and maintenance personnel no longer need to monitor network status in real time or manually adjust topology structures. Instead, they can perform targeted actions based on generated alarms and diagnostic results. This improves the efficiency and accuracy of operations and management, and promotes the intelligent development of power transmission and transformation construction and operations.

[0116] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the above-described method. All implementations in the above-described method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0117] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the above-described method. All implementations in the above-described method embodiment are applicable to this embodiment and can achieve the same technical effects.

[0118] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction, characterized in that: include: Multimodal module, used to deploy multimodal sensing units at key monitoring points of power transmission and transformation equipment, collect vibration, partial discharge, leakage current and environmental parameter signals, and process the signals through electromagnetic shielding enclosure and built-in harmonic filtering circuit to obtain original monitoring signal data; The detection module is used to input the original monitoring signal data into the chaos-enhanced weak signal detection device, and extract the fault feature vector by capturing the attractor phase diagram mutation and the time domain waveform threshold; The transmission module is used to transmit the fault feature vector through a wired and wireless dual-channel hot standby transmission device. When the signal attenuation reaches a preset threshold, the transmission device automatically switches the transmission channel and triggers the relay node to perform gain compensation for the fault feature vector. Specifically, it includes: obtaining the channel attenuation data and environmental interference characteristics of the current transmission link, and predicting the trend of signal attenuation amplitude changing with transmission time and distance based on the correlation between the time characteristics of the channel attenuation data and the environmental interference characteristics; combining the signal attenuation change trend with the real-time collected link impedance fluctuation parameters and the instantaneous value of the electromagnetic interference intensity to dynamically generate a real-time signal attenuation rate prediction value for each discrete position point in the transmission path; based on the distribution of the real-time signal attenuation rate prediction value, the transmission path of the fault feature vector data packet is divided into continuous segments according to the attenuation rate mutation point, and the corresponding theoretical signal attenuation degree at each segment node is calculated; comparing the theoretical signal attenuation degree of the segment node with the actual attenuation degree monitored in real time by the sensor, and obtaining the attenuation degree deviation at each node; based on the attenuation degree deviation, combined with the proportion of the influence of the transmission distance of the corresponding section on the signal attenuation, dynamically generating the compensation gain value required for power amplification of the fault feature vector data packet; The planning module is used to plan the transmission path of the fault feature vector data using a clustered multi-hop self-organizing network architecture. The cluster head node implements autonomous repair of wireless mesh network breakpoints based on a dynamic pathfinding strategy and transmits the fault feature vector data to the sink node. The diagnostic module is used to perform spatiotemporal correlation fusion of multi-source fault feature vectors based on the convergence node, match the diagnostic results based on the pre-trained fault mode library, dynamically adjust the sampling rate, transmission power and filtering parameters according to the diagnostic results, and generate alarm events at the same time, including: calculating the Euclidean distance between each fault mode feature vector and the fused feature vector in the pre-trained fault mode library, generating a distance measurement value set and normalizing it to obtain a standardized distance value set; inputting it into a preset similarity converter, and dynamically generating matching probability values ​​corresponding to each fault mode of the fused feature vector based on the built-in distance-probability inverse relationship; arranging the matching probability values ​​in descending order of value, comparing them with the preset diagnostic threshold item by item, determining the probability values ​​greater than the threshold and the associated fault modes, and forming a candidate fault mode set; extracting the fault mode ranked first in the set, parsing the equipment fault type code and the affected component identification; combining with the preset confidence grading rules, generating a deterministic diagnostic result including the specific fault location identification, fault type description and confidence rating; The control module is used to automatically reconstruct the multi-module network topology based on fault diagnosis results and network status feedback, dynamically adjust the working mode of each sensor unit, and realize adaptive maintenance and adjustment of multi-module network monitoring.

2. The weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction according to claim 1 is characterized in that: The fault feature vector is transmitted through a wired and wireless dual-channel hot standby transmission device. When the signal attenuation reaches a preset threshold, the transmission device automatically switches the transmission channel and triggers the relay node to perform gain compensation for the fault feature vector, including: The main controller monitors the quality index of the currently activated main transmission channel. If the main channel is a wired channel, it monitors the bit error rate; if the main channel is a wireless channel, it monitors the received signal strength indicator value. When the quality index of the main channel is less than the corresponding preset threshold, it determines that the channel signal attenuation has reached the preset threshold and generates a channel switching instruction and a gain compensation trigger signal. Based on the channel switching instruction and the gain compensation trigger signal, the transmission device performs a switching operation from the current active channel to the standby channel. At the same time, the gain compensation trigger signal activates the designated relay node on the transmission path to obtain the fault feature vector data packet of the transmission interruption caused by the disconnection of the active channel; For the fault feature vector data packet, the signal attenuation trajectory of the current transmission link is predicted based on the signal attenuation data, and the compensation gain value required for power amplification of the fault feature vector data packet is dynamically calculated according to the signal attenuation rate.

3. The weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction according to claim 2 is characterized in that: For fault feature vector data, a clustered multi-hop ad hoc network architecture is used to plan the transmission path. The cluster head node implements autonomous repair of wireless mesh network breakpoints based on a dynamic pathfinding strategy and transmits the fault feature vector data to the sink node, including: Receive the fault feature vector data packet after relay node gain compensation, and establish the topological connection structure between nodes in the cluster in the clustered multi-hop ad hoc network based on the node geographic location information and real-time link quality data required for data packet transmission to form an initial transmission path; Based on the topological connection structure, the cluster head node periodically detects the connectivity status of each neighboring node and updates the network routing table in real time based on the detection results. For each path recorded in the table, the performance is evaluated based on the comprehensive signal strength and node residual energy. When connectivity status monitoring indicates that there is a breakpoint in the current transmission path, the cluster head node initiates a dynamic pathfinding strategy to perform autonomous repair; After the path repair is completed, the cluster head node relays the fault feature vector data to the sink node through the repaired multi-hop path.

4. The weak signal multi-mode networking adaptive monitoring system for power transmission and transformation construction according to claim 3 is characterized in that: Based on the aggregation node, the multi-source fault feature vectors are temporally and spatially correlated and fused, and the diagnosis results are matched based on the pre-trained fault pattern library. The sampling rate, transmission power, and filtering parameters are dynamically adjusted based on the diagnosis results, and alarm events are generated at the same time, including: The aggregation node extracts the timestamp and device location tag of each fault feature vector and aligns the signals from different devices using a sliding time window based on the timestamp and location tags. After alignment, the weighted evidence theory is used to fuse the fault features associated in the temporal and spatial dimensions to generate a fused feature vector. The fused feature vector is input into the pre-trained fault pattern library for similarity matching. Based on the matching results, the equipment fault type and probability value are obtained. When the probability value is greater than the preset diagnostic threshold, a deterministic diagnostic result is generated. Dynamically adjust parameters based on the diagnosis results. If the fault probability in the diagnosis results is greater than the preset probability threshold, the signal sampling rate of the relevant equipment is increased; if the current link quality index is less than the preset quality threshold, the node transmission power is increased; if a frequency band interference signal is detected in the fusion feature, the suppression parameter of the filter is adaptively enhanced; The voltage signal in the fusion feature is monitored in real time. When the amplitude of the transient voltage mutation is detected to be greater than the safety limit, the hierarchical alarm event generation mechanism is immediately triggered.

5. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the system according to any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program, which, when executed by a processor, implements the system according to any one of claims 1 to 4.

Images