A cable fault response method and system based on distributed monitoring
By deploying state-sensing nodes along the cable laying path using a distributed monitoring method, signal sampling and decoupling processing are performed, the heat accumulation and expansion path is deduced, and a collaborative response strategy is generated. This solves the problem of insufficient early fault detection in cables and enables precise prevention and control of cable fire risks.
Patent Information
- Application Number
- CN202611369869.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-04
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies struggle to continuously detect minute anomalies along cable routes, especially at joint locations. This results in small temperature rises and short durations during early localized overheating phases, making them easy to miss during periodic inspections and potentially leading to cable fires. Current monitoring methods are insufficient in early fault detection, precise location, and fire response.
Distributed state-aware nodes are deployed along the cable laying path. Through multi-source signal sampling, time alignment, and drift subtraction, a primary disturbance sequence is formed. Nonlinear decoupling processing is performed to extract the latent fault characterization vector. Combined with the cable laying structure and ventilation conditions, the heat accumulation propagation path is deduced, a risk propagation constraint set is generated, and it is mapped to the cooling, isolation, ventilation, and power-off execution units to construct a collaborative response strategy.
It enables early detection, precise location, and coordinated control of cable fire risks, improving the inherent safety level of cable operation and the accuracy of fire prevention and control, and avoiding the limitations of traditional post-fault response.
Smart Images

Figure CN123073769A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system monitoring and fault response, specifically to a cable fault response method and system based on distributed monitoring. Background Technology
[0002] With the widespread application of urban underground utility tunnels, power tunnels, and power distribution systems within large buildings, cables are typically laid at high density in enclosed or semi-enclosed spaces. Cable operation safety and fire prevention are becoming increasingly prominent issues, especially in underground utility tunnels where medium- and low-voltage cables have long distribution distances and numerous joints. Localized overheating or arcing can easily ignite the cable sheath and surrounding flammable materials, leading to cable fires and severe secondary disasters. Current technologies primarily rely on periodic manual inspections, infrared thermography, and the installation of temperature or current monitoring devices at key locations to monitor cable operation. However, these monitoring methods are typically single-point or discrete, making them difficult to implement effectively. Continuous sensing of minor anomalies along cable lines, especially at joint locations, is crucial, particularly in the early stages of localized overheating caused by factors such as increased contact resistance, installation deviations, or aging at cable joints. In these early stages, the temperature rise is small, the duration is short, and the impact is limited, often insufficient to trigger traditional protection devices and easily missed during periodic inspections. If this localized overheating is not detected and addressed promptly, it can progress to insulation deterioration, arcing, or even open flame combustion, potentially leading to cable fires that spread rapidly along densely laid lines, causing power outages and major safety accidents. Therefore, existing technologies still have significant shortcomings in early detection, precise location, and rapid response for fire prevention in cable faults. Summary of the Invention
[0003] To address the shortcomings of existing technologies, this invention provides a cable fault response method and system based on distributed monitoring, which has the advantage of improving the accuracy of fire prevention and control, and solves the problems mentioned in the background technology.
[0004] To achieve the aforementioned goal of improving the accuracy of fire prevention and control, this invention provides the following technical solution: a cable fault response method based on distributed monitoring, comprising the following steps: Multiple distributed state sensing nodes are deployed along the cable laying path to sample the conductor temperature rise slope, shield leakage current change, partial discharge pulse and sheath surface temperature field fluctuation. The sampled signals are then processed for time alignment and background drift subtraction to form a primary disturbance sequence that reflects the local energy abnormal accumulation state. The primary disturbance sequence is input into the time-series nonlinear decoupling processing unit to perform component separation and correlation reconstruction of thermal diffusion hysteresis, contact resistance variation trend and discharge randomness, and extract the corresponding latent fault characterization vector. Using the corresponding latent fault characterization vector as state input, under the constraints of cable laying topology, ventilation conditions and combustible load distribution, the heat accumulation expansion path and smoke spread trend caused by abnormal state are dynamically deduced, and a risk propagation constraint set oriented towards the spatial impact range is generated. The risk propagation constraint set is mapped to the corresponding cooling, isolation, ventilation and power outage execution units, the temporal dependency relationship between each response element is established, and multiple sets of response candidate strategies that meet the safety margin conditions are generated under the multi-unit collaborative constraint conditions. Based on the current operational redundancy, the importance level of power load, and the historical effectiveness assessment results of interventions, the set of candidate response strategies is optimized and ranked to determine the response path that works collaboratively across multiple execution units.
[0005] Preferably, the process of sampling the conductor temperature rise slope, shielding layer leakage current change, partial discharge pulse, and sheath surface temperature field fluctuation is as follows: Temperature sensing units, current leakage detection units, ultra-high frequency partial discharge detection units, and infrared temperature field sensing units are installed along the cable laying path at cable joint sections, bend sections, conduit sections, and high-load sections. Configure a unified clock source for each sensing unit and establish a time synchronization mechanism to align the sampling time of each node at the microsecond level. The conductor temperature rise rate, shielding leakage current fluctuation, partial discharge pulse intensity, and sheath surface temperature distribution sequences were obtained according to the preset sampling frequency.
[0006] Preferably, the process of forming a primary perturbation sequence reflecting the local energy anomalous accumulation state is as follows: Temporal interpolation and resampling are performed on the sampled sequences from different sensing nodes; Long-term trend modeling is performed on various signals, and background drift components caused by changes in ambient temperature, cyclical fluctuations in power grid load, and equipment aging are subtracted. The residual signal after drift subtraction is normalized and scaled, and then combined according to node position and time sequence to form a primary perturbation sequence.
[0007] Preferably, the process of component separation and correlation reconstruction of thermal diffusion hysteresis, contact resistance variation trend and discharge randomness is as follows: The primary perturbation sequence is sliced according to a time window and input into the decoupling processing unit; Based on the temporal correlation and physical coupling of signals, component extraction, trend analysis and dynamic weight adjustment are performed on temperature rise change, leakage current change and discharge pulse signals, respectively. Based on component extraction, the coupling relationship between each signal is reconstructed to generate multidimensional correlation data that reflects local energy transfer characteristics and potential abnormal evolution trends.
[0008] Preferably, the process of extracting the corresponding latent fault representation vector is as follows: The energy intensity, growth rate, duration, mutation frequency, inter-component correlation strength, and time delay characteristics are extracted from the various components and multidimensional correlation data after separation. Various feature parameters and coupling relationship information are combined into a unified multidimensional feature vector according to preset weights; The multidimensional feature vectors are scaled and stabilized to form a latent fault characterization vector that can characterize the latent abnormal state and local energy transfer characteristics of the cable.
[0009] Preferably, the process of generating a risk propagation constraint set oriented towards the spatial impact range is as follows: The latent fault characterization vector is mapped to the actual cable laying location and spatial layout; Based on the overall ventilation conditions and the distribution of the surrounding combustible load, we can deduce the heat accumulation and expansion path and the direction of smoke spread caused by abnormal conditions. Based on the simulation results, the spatial range, duration, and intensity of the possible expansion of the abnormal impact are determined, forming a risk propagation constraint set oriented towards the spatial impact range.
[0010] Preferably, the process of establishing the temporal dependencies between the response elements is as follows: The spatial scope, duration, and impact intensity of risk propagation constraints are matched with the coverage area, operational capabilities, and response time of cooling devices, isolation devices, ventilation devices, and power-off switch units. The triggering conditions, response timing, coordination relationships, and potential conflicts of each execution unit are analyzed to clarify the priority order and safety constraints of multi-unit joint operations. The mapped execution units and constraint relationships are structured and organized to form a response element dependency network that can be used for policy generation and dynamic scheduling.
[0011] Preferably, the process of generating multiple sets of response candidate strategies that satisfy the safety margin condition under multi-unit collaborative constraints is as follows: Under the constraint of the network of response elements, the action sequence, effect intensity and triggering time of the cooling, isolation, ventilation and power failure execution units are combined in multiple dimensions to generate a preliminary strategy set. Based on the simulation results of the preliminary strategy combination on heat accumulation suppression, flue gas spread control and equipment safety working boundary, strategies that do not meet the safety margin conditions are screened and adjusted. The optimized strategy combinations are summarized according to redundancy, energy efficiency, and multi-unit synergy effect, ultimately forming a set of multiple response candidate strategies that meet the safety margin conditions and can be used for scheduling reference.
[0012] Preferably, the process of determining the response path that coordinates across multiple execution units is as follows: Based on multiple sets of response candidate strategies, the degree of system redundancy occupation, the degree of impact on important loads, and the expected risk reduction effect of each response candidate strategy are calculated. The effectiveness of each strategy was adjusted based on historical intervention records; The strategies are ranked according to the principle of prioritizing risk mitigation and minimizing business impact, and the optimal strategy is selected as the final response path.
[0013] A cable fault response system based on distributed monitoring includes: Disturbance sensing module: Collects conductor temperature rise, leakage current change, partial discharge pulse and sheath temperature field signals along the cable path, and performs time alignment and background drift subtraction to form a primary disturbance sequence; Fault decoupling module: Performs nonlinear decoupling processing on the primary disturbance sequence, separates the components of thermal diffusion, contact resistance change and discharge randomness, and generates a latent fault characterization vector; Propagation simulation module: Based on the latent fault characterization vector, the module simulates the trend of heat accumulation and smoke spread under the constraints of laying topology, ventilation conditions and combustible load, and generates a risk propagation constraint set; Response building module: Maps the risk propagation constraint set to cooling, isolation, ventilation and power failure execution units and builds a set of multi-unit collaborative response strategies that meet safety margins; Path optimization module: Optimizes and sorts the set of response strategies and determines the response path that works collaboratively across multiple execution units.
[0014] Compared with the prior art, the present invention provides a cable fault response method and system based on distributed monitoring, which has the following advantages: This invention achieves continuous acquisition of multi-source abnormal signals such as conductor temperature rise, leakage current changes, partial discharge, and sheath temperature field by setting up distributed state sensing nodes along the entire cable path. Time alignment and drift subtraction are used to eliminate interference from environmental and operating condition disturbances on the judgment results, ensuring that early weak anomalies are no longer masked by background noise. Furthermore, by decoupling and reconstructing various disturbance components, latent fault states such as joint degradation, insulation fatigue, and micro-arc discharge can be identified in advance before triggering protection thresholds. Further, by combining cable laying structure, ventilation conditions, and combustible load distribution, the invention can detect the heat accumulation and smoke propagation paths that may be caused by abnormal states. By performing simulations, the traditional "post-fault response" is transformed into "pre-risk intervention." Simultaneously, by mapping the risk propagation constraints to multiple execution units such as cooling, isolation, ventilation, and power outage, and constructing a collaborative response strategy, the response is no longer limited to single-point isolation but can be tiered according to the scope of impact. Finally, by optimizing and prioritizing the response strategies, the spread of fire risk is suppressed while ensuring power supply continuity. Therefore, this invention can achieve early perception, precise location, and collaborative control of cable fire risks without significantly increasing hardware complexity, significantly improving the inherent safety level of cable operation and the accuracy of fire prevention and control. Attached Figure Description
[0015] Figure 1 This is a schematic diagram of the method of the present invention; Figure 2 This is a schematic diagram of the structure of the present invention. Detailed Implementation
[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0017] Example 1: Please refer to Figure 1 As shown in the embodiment of the present invention, a cable fault response method based on distributed monitoring includes the following steps: S1: Multiple distributed state sensing nodes are deployed along the cable laying path to sample the conductor temperature rise slope, shield leakage current change, partial discharge pulse and sheath surface temperature field fluctuation, and the sampled signals are processed by time alignment and background drift subtraction to form a primary disturbance sequence reflecting the local energy abnormal accumulation state.
[0018] The process of sampling conductor temperature rise slope, shielding layer leakage current change, partial discharge pulse, and sheath surface temperature field fluctuation in S1 is as follows: Temperature sensing units, current leakage detection units, ultra-high frequency partial discharge detection units, and infrared temperature field sensing units are installed along the cable laying path at cable joint sections, bend sections, conduit sections, and high-load sections. Along the cable laying path, temperature sensing units, current leakage detection units, ultra-high frequency partial discharge detection units, and infrared temperature field sensing units are deployed at key locations such as cable joint sections, bend sections, conduit sections, and high-load sections. Temperature sensing units can use PT100 thermistors, thermocouples, or fiber Bragg grating sensors, installed on the conductor surface or outside the sheath, fixed with insulating clamps, and protected by protective covers to prevent external mechanical impact and environmental interference. Current leakage detection units can use shunts or Hall effect sensors, installed on the shielding layer or grounding terminal, and connected to the acquisition module through an isolation amplification circuit. Ultra-high frequency partial discharge detection units use UHF antennas or discharge detectors with a bandwidth of 30~300MHz, deployed at joint sections, bend sections, and high-load sections to capture instantaneous partial discharge pulse signals. Infrared temperature field sensing units use infrared thermal imagers or infrared thermocouple arrays, with multiple scanning points deployed along the cable length to continuously collect temperature distribution data on the sheath surface. All sensors are connected via protective cables and wiring is made of high-temperature resistant insulating materials to ensure long-term stable operation.
[0019] Configure a unified clock source for each sensing unit and establish a time synchronization mechanism to align the sampling time of each node at the microsecond level. All sensing units are equipped with a unified high-precision clock source, which can be synchronized using a GPS module or IEEE1588PTP network to ensure microsecond-level alignment of sampling time. The timestamps of each node are transmitted through industrial Ethernet, fiber optic bus or wireless synchronization network and are time-corrected in the acquisition system. For long-distance cables, clock correction modules can be set at relay nodes every few hundred meters to eliminate time deviations caused by transmission delays, ensuring accurate correspondence of data from each node in the time dimension, and providing a reliable foundation for time-series analysis and anomaly detection.
[0020] The conductor temperature rise rate, shielding leakage current fluctuation, partial discharge pulse intensity and sheath surface temperature distribution sequences were obtained according to the preset sampling frequency. The temperature sensing unit collects conductor and sheath temperatures at a preset sampling frequency, such as 1kHz or higher, and performs differential operations on the continuous temperature sequence through edge computing nodes to calculate the rate of temperature change. The current leakage detection unit continuously monitors the leakage current of the shielding layer, using a high-speed ADC with a sampling frequency of 1-10kHz to sample microamps to milliamps of signal and record fluctuation amplitude and frequency characteristics. The ultra-high frequency partial discharge detection unit collects the amplitude, pulse width, and frequency characteristics of partial discharge pulse signals in real time at key nodes, records the pulse waveform using a high-speed oscilloscope or a dedicated acquisition board with a sampling rate ≥50MHz, and analyzes the location and development trend of partial discharge by combining time synchronization information. The infrared temperature field sensing unit continuously scans the sheath surface temperature along the cable length to obtain a two-dimensional or three-dimensional temperature distribution sequence. The collected data is constructed into a time-series matrix by the data processing unit, including the temperature rise slope, leakage current fluctuation, partial discharge pulse, and sheath temperature field. By combining the temperature rise slope and partial discharge data, local hot spots and potential insulation defects can be analyzed, providing basic data for latent anomaly detection and cable preventive maintenance.
[0021] The process of forming a primary perturbation sequence in S1 that reflects the local energy anomalous accumulation state is as follows: Temporal interpolation and resampling are performed on the sampled sequences from different sensing nodes; The original sampling sequences collected along the cable path at key nodes, including joint sections, bends, conduit sections, and high-load sections, including conductor temperature rise slope, shield leakage current changes, partial discharge pulses, and sheath surface temperature fluctuations, are preprocessed. For cases where sampling frequencies differ or sampling is missing at different nodes, linear interpolation, spline interpolation, or other numerical interpolation methods are used to interpolate the time series. Subsequently, based on the system analysis requirements, the sequences from each node are uniformly resampled to a preset time step, such as 1ms or 10ms, to form time-aligned sequences, ensuring the comparability and consistency of data across nodes in the time dimension during analysis.
[0022] Long-term trend modeling is performed on various signals, and background drift components caused by changes in ambient temperature, cyclical fluctuations in power grid load, and equipment aging are subtracted. For signals after various interpolations and resampling, a long-term trend model is established to characterize the slow drift components caused by changes in ambient temperature, cyclical fluctuations in power grid load, and aging of cable equipment. Trend modeling can be achieved using methods such as moving average filtering, low-pass filtering, or exponential smoothing. Alternatively, polynomial fitting or periodic decomposition based on Fourier / wavelet analysis can be used to subtract the long-term trend or background drift components from the original signal to obtain the residual signal after removing environmental influences, thus reflecting the true dynamic characteristics of local abnormal energy accumulation.
[0023] The residual signal after drift subtraction is normalized and scaled, and then combined according to node position and time order to form a primary perturbation sequence; The residual signal after drift subtraction is normalized to enable direct comparison of signals from different types of sensors, such as temperature rise slope, leakage current, partial discharge pulse amplitude, and infrared temperature field. Normalization methods can include Z-score normalization, maximum-minimum normalization, or quantile normalization, mapping the signal to a fixed interval and unifying the signal units and dimensions. This normalization ensures that in subsequent combined analysis, different nodes and signal types will not mask abnormal characteristics due to amplitude differences. The normalized residual signal is then combined according to the spatial location and time sequence of the sensor nodes, with each row representing the node location and each column representing the sampling time step, constructing a complete spatiotemporal perturbation matrix. This matrix is the primary perturbation sequence, which can intuitively reflect the accumulation state and spatiotemporal evolution characteristics of local energy anomalies. The primary perturbation sequence can serve as input for heat diffusion, partial discharge, and anomaly prediction models, providing an actionable data foundation for latent fault detection, local energy accumulation analysis, and cable preventative maintenance.
[0024] S2: Input the primary disturbance sequence into the time-series nonlinear decoupling processing unit to perform component separation and correlation reconstruction of thermal diffusion hysteresis, contact resistance variation trend and discharge randomness, and extract the corresponding latent fault characterization vector.
[0025] The process of component separation and correlation reconstruction of thermal diffusion hysteresis, contact resistance variation trend and discharge randomness in S2 is as follows: The primary perturbation sequence is sliced according to a time window and input into the decoupling processing unit; The primary disturbance sequence is sliced according to a preset time window. Each time window can be selected with an appropriate length based on the dynamic characteristics of the system, such as 100ms or 1s. Each window contains the normalized conductor temperature rise slope, shield leakage current fluctuation, partial discharge pulse, and sheath temperature field sequence for each node. The sliced data is input to the decoupling processing unit to provide local time-domain samples for component separation and signal reconstruction.
[0026] Based on the temporal correlation and physical coupling of signals, component extraction, trend analysis and dynamic weight adjustment are performed on temperature rise change, leakage current change and discharge pulse signals, respectively. In the decoupling processing unit, based on the temporal correlation and physical coupling relationship of each signal, the temperature rise change, leakage current change, and partial discharge pulse signals are extracted and trend analyzed respectively. Component extraction can be performed using time-series filtering, empirical mode decomposition, wavelet decomposition, or principal component analysis. The original signal is split into low-frequency trend components, mid-frequency dynamic components, and high-frequency random components. Among them, the temperature rise change signal is mainly used to analyze the thermal diffusion hysteresis characteristics, the leakage current signal is used to analyze the contact resistance change trend, and the partial discharge signal is used to analyze the random discharge characteristics. After the component extraction is completed, dynamic weight adjustment is performed on each component. The initial weight is calculated based on the energy intensity, rate of change, and abrupt change amplitude of the component within the current time window. Combined with the spatial location of the node and the signal status of the neighboring nodes, the weight of key abnormal components with sustained high values or spatial accumulation is appropriately increased, while the weight of low-amplitude or noise components is reduced. Finally, the dynamic weight is applied to each component to generate a weighted signal for signal coupling reconstruction and latent fault characterization, so as to enhance the system's sensitivity and identification ability to local abnormal energy accumulation.
[0027] Based on component extraction, the coupling relationship between each signal is reconstructed to generate multidimensional correlation data that reflects local energy transfer characteristics and potential abnormal evolution trends. Based on the extraction and dynamic weight adjustment of each signal component, the coupling relationship between different types of signals is reconstructed by combining the spatial location and physical connection relationship of the nodes. By calculating the correlation coefficient, time delay characteristics and energy transfer path between components, the separated temperature rise, leakage current and partial discharge components are associated and combined according to the physical coupling rules to generate multidimensional correlation data. This multidimensional correlation data can reflect the local energy transfer characteristics and potential abnormal evolution trend, and provide feasible data input for latent fault characterization, risk analysis and response strategy generation.
[0028] The process of extracting the corresponding latent fault representation vector in S2 is as follows: The energy intensity, growth rate, duration, mutation frequency, inter-component correlation strength, and time delay characteristics are extracted from the various components and multidimensional correlation data after separation. Along each sampling node of the cable, signal components such as temperature rise change, shield leakage current, partial discharge, and sheath temperature field are input into the data processing module. The signal energy intensity is calculated using a sliding window, which can be obtained by sum of squares or integral summation. The growth rate is calculated by linear fitting slope or first-order difference within the window. The duration of continuous signal mutations exceeding the threshold is counted. The number of signal mutations exceeds the preset threshold. For the coupling relationship between different nodes or different signals, cross-correlation analysis or time delay correlation calculation is used to obtain the correlation strength and propagation delay information, thereby forming a comprehensive feature set.
[0029] Various feature parameters and coupling relationship information are combined into a unified multidimensional feature vector according to preset weights; Based on the importance of cable nodes, signal sensitivity, and historical fault experience, weights are assigned to different types of features, such as a weight of 0.4 for temperature rise changes, 0.35 for partial discharge, and 0.25 for leakage current. The extracted feature vectors are then combined into a unified multidimensional vector according to the weights. The combination method can be linear weighted summation or principal component weighted mapping to ensure that various signal features are reasonably represented in the multidimensional vector. After the combination is completed, each multidimensional vector corresponds to a sampling node and a time slice, preserving complete space-time information.
[0030] The multidimensional feature vectors are scaled and stabilized to form a latent fault characterization vector that can characterize the latent abnormal state and local energy transfer characteristics of the cable. The combined multidimensional feature vectors are normalized using Z-score or min-max normalization to map each feature to a uniform numerical range, eliminating amplitude differences. The normalized vectors are then smoothed or subjected to exponential weighted averaging to reduce instantaneous noise interference and enhance time series consistency. After processing, the resulting latent fault characterization vector contains both local energy anomaly features and retains inter-node coupling information, which can be directly used for cable anomaly detection, fault location, and response strategy generation.
[0031] S3: Using the corresponding latent fault characterization vector as the state input, under the constraints of cable laying topology, ventilation conditions and combustible load distribution, dynamically deduce the heat accumulation expansion path and smoke spread trend caused by abnormal state, and generate a risk propagation constraint set oriented towards the spatial impact range.
[0032] The process of generating a risk propagation constraint set oriented towards spatial impact range in S3 is as follows: The latent fault characterization vector is mapped to the actual cable laying location and spatial layout; Along the cable laying path, the three-dimensional spatial coordinates X, Y, Z, cable type, cross-section, and laying environment information of each sampling node are recorded. The latent fault characterization vector corresponding to each node is bound to the node coordinates to form a node-feature mapping table. The mapping information can be stored in a database or in-memory data structure to facilitate spatial analysis and risk propagation calculation, ensuring that each anomaly characterization vector has a clear spatial location, providing a basis for predicting the anomaly propagation path.
[0033] Based on the overall ventilation conditions and the distribution of the surrounding combustible load, we can deduce the heat accumulation and expansion path and the direction of smoke spread caused by abnormal conditions. Based on the node spatial mapping information and the latent fault characterization vector, heat conduction simulation is performed on local hot spots or potential partial discharge areas of the cable. The diffusion rate and direction of heat in the surrounding medium are calculated. Combined with the layout of ventilation devices, air velocity and flow direction information, the spread trajectory of smoke or high-temperature air is simulated. The calculation is performed using the finite difference method, finite element method or CFD computational fluid dynamics method. At the same time, the heat capacity and heat release characteristics of the combustible load around the cable are considered to predict the spatial expansion path of the impact of abnormal events. Through simulation, the spatial boundary and development trend of potential hazardous areas can be obtained, providing data support for the determination of risk constraints.
[0034] Based on the simulation results, the potential spatial range, duration, and intensity of the impact of the anomaly are determined, forming a risk propagation constraint set oriented towards the spatial impact range. The heat accumulation and flue gas spread range are transformed into specific constraint information, including: spatial range: defining the area that the anomaly may affect in the form of a three-dimensional grid or bounding box; duration: calculating the duration window of the anomaly's impact based on the heat diffusion rate, ventilation conditions, and heat load; and intensity of impact: assessing the potential threat level to the environment and equipment based on heat, flue gas concentration, or local cable load status. The above information is stored in tables or data structures to form a risk propagation constraint set that can be used for risk analysis and dynamic scheduling. This constraint set can be directly input into the response strategy generation module to ensure that spatial impact and safety margins are considered when formulating multi-unit intervention measures.
[0035] S4: Map the risk propagation constraint set to the corresponding cooling, isolation, ventilation and power failure execution units, establish the temporal dependency relationship between each response element, and generate multiple sets of response candidate strategies that meet the safety margin conditions under the multi-unit collaborative constraint conditions.
[0036] The process of establishing temporal dependencies between response elements in S4 is as follows: The spatial scope, duration, and impact intensity of risk propagation constraints are matched with the coverage area, operational capabilities, and response time of cooling devices, isolation devices, ventilation devices, and power-off switch units. Based on the three-dimensional spatial information of the risk propagation constraint set, each potential abnormal area is spatially overlaid with the execution units deployed on site to determine the risk area that each execution unit can cover. Combining the operational capabilities of each unit, such as maximum cooling power, ventilation flow rate, power outage capacity, and response time, as well as the time required from the trigger signal to the effect taking effect, the scope of action and availability of each execution unit under different abnormal events are calculated. The matching results are recorded as a node-constraint correspondence table for strategy generation, ensuring that each risk point can find a schedulable execution unit and guaranteeing the feasibility of intervention measures.
[0037] The triggering conditions, response timing, coordination relationships, and potential conflicts of each execution unit are analyzed to clarify the priority order and safety constraints of multi-unit joint operations. The triggering conditions of each execution unit, such as temperature threshold, flue gas concentration, leakage current intensity, response timing, action delay, duration, and cooperative constraints, as well as the possible conflicts or enhancement effects of simultaneous operation of multiple units, are organized into tables or data structures. During the analysis process, logical judgments or rule engines can be used to identify conflicting conditions. For example, if the simultaneous operation of cooling devices may cause local airflow reversal, sequence constraints or time interval limits can be set. The analysis results are used to determine the priority order and safe operating range of multi-unit joint operation and generate a preliminary set of constraint rules to provide input for dynamic scheduling and strategy generation.
[0038] The mapped execution units and constraint relationships are structured and organized to form a response element dependency network that can be used for policy generation and dynamic scheduling; Each execution unit is treated as a network node, recording its operation type, scope of action, response time, triggering conditions, and security constraints. Directed edges are established based on the triggering order, collaborative constraints, and potential conflicts, with minimum interval time, priority, or collaborative conditions marked on the edges. Nodes and edges are organized in multiple layers according to spatial location and operational logic, forming a basic operation layer, a collaborative operation layer, and a redundant security layer structure. After completing the above steps, a structured response element dependency network is obtained. After cyclic dependency checks, conflict corrections, and coverage gap optimization, the network can finally be directly used for strategy generation and dynamic scheduling, achieving effective control of risks to cables and surrounding spaces.
[0039] In S4, the process of generating multiple sets of response candidate strategies that satisfy the safety margin condition under multi-unit cooperative constraints is as follows: Under the constraint of the network of response elements, the action sequence, effect intensity and triggering time of the cooling, isolation, ventilation and power failure execution units are combined in multiple dimensions to generate a preliminary strategy set. Under the network constraints of response element dependence, the operational range and allowed action window of each execution unit's cooling, isolation, ventilation, and power-off are determined according to its triggering conditions, response timing, and cooperative constraints. For each abnormal scenario, the action sequence, effect intensity, and triggering time of the execution unit are used as discrete decision variables. All feasible operation combinations are generated through exhaustive search, branch and bound, or heuristic search algorithms. Each set of operation combinations records the specific unit action sequence, intensity setting, and triggering time, forming a preliminary strategy set that can be directly used for simulation and screening. The strategies generated in this way can be implemented and are guaranteed to be operable within the constraints of the execution unit.
[0040] Based on the simulation results of the preliminary strategy combination on heat accumulation suppression, flue gas spread control and equipment safety working boundary, strategies that do not meet the safety margin conditions are screened and adjusted. For each strategy in the initial strategy set, simulation verification is performed on the thermal accumulation suppression, flue gas spread control, and equipment safety operating boundary. The simulation methods may include finite element heat conduction analysis, flue gas diffusion simulation, and equipment operation boundary constraint checks to evaluate the safety margin of the strategy under abnormal conditions. For strategies that do not meet the safety margin requirements, iterative optimization is performed by adjusting the action sequence of the execution unit, the trigger time, or the intensity of the action until its thermal state, flue gas concentration, and equipment operating state are all within the safe range.
[0041] The optimized strategy combinations are summarized according to redundancy, energy efficiency and multi-unit synergy effect, and finally a set of multiple response candidate strategies that meet the safety margin conditions and can be used for scheduling reference. The optimized strategy set is summarized and sorted according to redundancy, energy efficiency, and multi-unit collaboration effect. Strategy combinations with high redundancy, low energy consumption, and good collaboration effect are retained first. Applicable conditions and operational constraints are recorded for each strategy set, forming a set of multiple response candidate strategies that can be dynamically referenced by the scheduling system. The final strategy set is archived and verified and can be directly used as a scheduling reference, providing a practical data foundation for the system to quickly select the optimal response path in abnormal situations.
[0042] S5: Based on the current operational redundancy, power load importance level, and historical intervention effectiveness assessment results, the set of response candidate strategies is optimized and ranked to determine the response path that works collaboratively across multiple execution units.
[0043] The process of determining the response path across multiple execution units in S5 is as follows: Based on multiple sets of response candidate strategies, the degree of system redundancy occupation, the degree of impact on important loads, and the expected risk reduction effect of each response candidate strategy are calculated. Using a simulation platform or digital twin model, the action sequence, trigger time, and intensity of each execution unit in the candidate strategy are input into the model to simulate the system response under abnormal conditions. The simulation results quantify the impact of each strategy on critical loads, such as cable loads and temperature-sensitive equipment, as well as the occupancy of redundant resources. At the same time, the risk reduction effect of the strategy in terms of heat accumulation suppression, flue gas spread control, and equipment safety boundary maintenance is calculated. All indicators are recorded in numerical form to provide an objective basis for strategy selection.
[0044] The effectiveness of each strategy was adjusted based on historical intervention records; Establish a historical intervention database to record the triggering sequence, triggering time, intensity of action, and synergistic effect of each execution unit in past events. Compare the simulation results of candidate strategies with historical data to identify deviations or potential conflicts, and optimize and correct the triggering sequence, action sequence, or intensity of action to ensure the feasibility, safety, and synergistic efficiency of the strategy under the current system conditions.
[0045] The strategies are ranked according to the principle of prioritizing risk mitigation and minimizing business impact, and the strategy with the best ranking is selected as the final response path. Each strategy group is comprehensively scored using a weighted scoring or multi-objective optimization method. Indicators such as risk reduction effect, system redundancy, critical load impact, and multi-unit collaboration effect are integrated to calculate a comprehensive score. The strategies are ranked according to their comprehensive scores, and the strategies that perform best in ensuring system safety, critical load stability, and rapid response are selected as the final path. The action sequence, triggering time, and intensity of each execution unit are output, which can be directly used by the scheduling system to achieve multi-unit collaborative response.
[0046] Example 2: Please refer to Figure 2 As shown, a cable fault response system based on distributed monitoring includes: Disturbance sensing module: Collects conductor temperature rise, leakage current change, partial discharge pulse and sheath temperature field signals along the cable path, and performs time alignment and background drift subtraction to form a primary disturbance sequence; Fault decoupling module: Performs nonlinear decoupling processing on the primary disturbance sequence, separates the components of thermal diffusion, contact resistance change and discharge randomness, and generates a latent fault characterization vector; Propagation simulation module: Based on the latent fault characterization vector, the module simulates the trend of heat accumulation and smoke spread under the constraints of laying topology, ventilation conditions and combustible load, and generates a risk propagation constraint set; Response building module: Maps the risk propagation constraint set to cooling, isolation, ventilation and power failure execution units and builds a set of multi-unit collaborative response strategies that meet safety margins; Path optimization module: Optimizes and sorts the set of response strategies and determines the response path that works collaboratively across multiple execution units.
[0047] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0048] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A cable fault response method based on distributed monitoring, characterized in that, Includes the following steps: Multiple distributed state sensing nodes are deployed along the cable laying path to sample the conductor temperature rise slope, shield leakage current change, partial discharge pulse and sheath surface temperature field fluctuation. The sampled signals are then processed for time alignment and background drift subtraction to form a primary disturbance sequence that reflects the local energy abnormal accumulation state. The primary disturbance sequence is input into the time-series nonlinear decoupling processing unit to perform component separation and correlation reconstruction of thermal diffusion hysteresis, contact resistance variation trend and discharge randomness, and extract the corresponding latent fault characterization vector. Using the corresponding latent fault characterization vector as state input, under the constraints of cable laying topology, ventilation conditions and combustible load distribution, the heat accumulation expansion path and smoke spread trend caused by abnormal state are dynamically deduced, and a risk propagation constraint set oriented towards the spatial impact range is generated. The risk propagation constraint set is mapped to the corresponding cooling, isolation, ventilation and power outage execution units, the temporal dependency relationship between each response element is established, and multiple sets of response candidate strategies that meet the safety margin conditions are generated under the multi-unit collaborative constraint conditions. Based on the current operational redundancy, the importance level of power load, and the historical effectiveness assessment results of interventions, the set of candidate response strategies is optimized and ranked to determine the response path that works collaboratively across multiple execution units.
2. The cable fault response method based on distributed monitoring according to claim 1, characterized in that, The process of sampling the conductor temperature rise slope, shielding leakage current change, partial discharge pulse, and sheath surface temperature field fluctuation is as follows: Temperature sensing units, current leakage detection units, ultra-high frequency partial discharge detection units, and infrared temperature field sensing units are installed along the cable laying path at cable joint sections, bend sections, conduit sections, and high-load sections. Configure a unified clock source for each sensing unit and establish a time synchronization mechanism to align the sampling time of each node at the microsecond level. The conductor temperature rise rate, shielding leakage current fluctuation, partial discharge pulse intensity, and sheath surface temperature distribution sequences were obtained according to the preset sampling frequency.
3. The cable fault response method based on distributed monitoring according to claim 2, characterized in that, The process of forming a primary perturbation sequence reflecting the local energy anomalous accumulation state is as follows: Temporal interpolation and resampling are performed on the sampled sequences from different sensing nodes; Long-term trend modeling is performed on various signals, and background drift components caused by changes in ambient temperature, cyclical fluctuations in power grid load, and equipment aging are subtracted. The residual signal after drift subtraction is normalized and scaled, and then combined according to node position and time sequence to form a primary perturbation sequence.
4. The cable fault response method based on distributed monitoring according to claim 3, characterized in that, The process of component separation and correlation reconstruction of thermal diffusion hysteresis, contact resistance variation trend and discharge randomness is as follows: The primary perturbation sequence is sliced according to a time window and input into the decoupling processing unit; Based on the temporal correlation and physical coupling of signals, component extraction, trend analysis and dynamic weight adjustment are performed on temperature rise change, leakage current change and discharge pulse signals, respectively. Based on component extraction, the coupling relationship between each signal is reconstructed to generate multidimensional correlation data that reflects local energy transfer characteristics and potential abnormal evolution trends.
5. The cable fault response method based on distributed monitoring according to claim 4, characterized in that, The process of extracting the corresponding latent fault representation vector is as follows: The energy intensity, growth rate, duration, mutation frequency, inter-component correlation strength, and time delay characteristics are extracted from the various components and multidimensional correlation data after separation. Various feature parameters and coupling relationship information are combined into a unified multidimensional feature vector according to preset weights; The multidimensional feature vectors are scaled and stabilized to form a latent fault characterization vector that can characterize the latent abnormal state and local energy transfer characteristics of the cable.
6. The cable fault response method based on distributed monitoring according to claim 5, characterized in that, The process of generating a risk propagation constraint set oriented towards the spatial scope of influence is as follows: The latent fault characterization vector is mapped to the actual cable laying location and spatial layout; Based on the overall ventilation conditions and the distribution of the surrounding combustible load, we can deduce the heat accumulation and expansion path and the direction of smoke spread caused by abnormal conditions. Based on the simulation results, the spatial range, duration, and intensity of the possible expansion of the abnormal impact are determined, forming a risk propagation constraint set oriented towards the spatial impact range.
7. A cable fault response method based on distributed monitoring according to claim 6, characterized in that, The process of establishing temporal dependencies between response elements is as follows: The spatial scope, duration, and impact intensity of risk propagation constraints are matched with the coverage area, operational capabilities, and response time of cooling devices, isolation devices, ventilation devices, and power-off switch units. The triggering conditions, response timing, coordination relationships, and potential conflicts of each execution unit are analyzed to clarify the priority order and safety constraints of multi-unit joint operations. The mapped execution units and constraint relationships are structured and organized to form a response element dependency network that can be used for policy generation and dynamic scheduling.
8. A cable fault response method based on distributed monitoring according to claim 7, characterized in that, The process of generating multiple sets of response candidate strategies that satisfy the safety margin condition under multi-unit cooperative constraints is as follows: Under the constraint of the network of response elements, the action sequence, effect intensity and triggering time of the cooling, isolation, ventilation and power failure execution units are combined in multiple dimensions to generate a preliminary strategy set. Based on the simulation results of the preliminary strategy combination on heat accumulation suppression, flue gas spread control and equipment safety working boundary, strategies that do not meet the safety margin conditions are screened and adjusted. The optimized strategy combinations are summarized according to redundancy, energy efficiency, and multi-unit synergy effect, ultimately forming a set of multiple response candidate strategies that meet the safety margin conditions and can be used for scheduling reference.
9. A cable fault response method based on distributed monitoring according to claim 8, characterized in that, The process of determining the response path that involves collaboration across multiple execution units is as follows: Based on multiple sets of response candidate strategies, the degree of system redundancy occupation, the degree of impact on important loads, and the expected risk reduction effect of each response candidate strategy are calculated. The effectiveness of each strategy was adjusted based on historical intervention records; The strategies are ranked according to the principle of prioritizing risk mitigation and minimizing business impact, and the optimal strategy is selected as the final response path.
10. A cable fault response system based on distributed monitoring, applied to the method described in any one of claims 1-9, characterized in that, include: Disturbance sensing module: Collects conductor temperature rise, leakage current change, partial discharge pulse and sheath temperature field signals along the cable path, and performs time alignment and background drift subtraction to form a primary disturbance sequence; Fault decoupling module: Performs nonlinear decoupling processing on the primary disturbance sequence, separates the components of thermal diffusion, contact resistance change and discharge randomness, and generates a latent fault characterization vector; Propagation simulation module: Based on the latent fault characterization vector, the module simulates the trend of heat accumulation and smoke spread under the constraints of laying topology, ventilation conditions and combustible load, and generates a risk propagation constraint set; Response building module: Maps the risk propagation constraint set to cooling, isolation, ventilation and power failure execution units and builds a set of multi-unit collaborative response strategies that meet safety margins; Path optimization module: Optimizes and sorts the set of response strategies and determines the response path that works collaboratively across multiple execution units.
