Electrical fire early warning method and system

Abstract

The invention relates to the technical field of electrical safety early warning, and discloses an electrical fire early warning method and system. The method comprises the following steps: acquiring a multi-dimensional dynamic parameter set of an electrical circuit in real time, wherein the multi-dimensional dynamic parameter set comprises a current fluctuation sequence, temperature gradient distribution data and an insulating medium loss value; performing time-frequency domain joint decomposition processing on the dynamic parameter set, and extracting current harmonic distortion characteristics, a temperature field spatial evolution mode and a dielectric loss accumulation rate; generating a current anomaly index according to a harmonic distortion characteristic and a preset threshold deviation degree, calculating a thermal runaway risk level in combination with a temperature field spatial evolution mode, and determining an insulation degradation coefficient based on a dielectric loss accumulation rate; fusing the three to construct a comprehensive fire risk index; and when the index exceeds the dynamic warning threshold value, a grading early warning signal is triggered and a hidden danger positioning map is generated. According to the method, the electrical fire early warning effect is optimized through multi-dimensional parameter acquisition and fusion analysis.

Description

Technical Field

[0001] This invention relates to the field of electrical safety early warning technology, specifically to an electrical fire early warning method and system. Background Technology

[0002] In modern society, electrical circuits, as the core carriers of energy transmission, are directly related to the safety of people's lives and property. With the expansion of power systems and the diversification of electrical equipment, electrical circuits operate in complex environments, making them highly susceptible to fires caused by overload, poor contact, and insulation aging, resulting in serious losses to society. Currently, most existing electrical fire early warning technologies focus on single-parameter monitoring, such as collecting line current values ​​solely through current sensors or obtaining local temperature data using temperature sensors. These single-parameter monitoring methods have significant limitations. When a line experiences only slight current fluctuations that do not reach the overload threshold, single-parameter monitoring systems cannot identify potential risks; similarly, when a localized temperature anomaly occurs while the overall current remains normal, it is difficult to detect in a timely manner.

[0003] Current technologies for monitoring insulating media are mostly limited to static insulation resistance testing, failing to capture real-time trends in insulation loss during long-term operation. Insulation aging is a gradual process; static testing only reflects the insulation state at a specific moment and cannot predict the rate of degradation or potential faults. Furthermore, traditional early warning systems lack effective fusion and analysis of multi-dimensional monitoring data. Current, temperature, and insulation status data are processed independently, making it difficult to form a comprehensive assessment of the overall operating status of electrical circuits. This results in insufficient accuracy in early warnings, leading to false alarms or missed alarms. For example, when current harmonic distortion occurs in a line, relying solely on the effective current value may fail to identify the potential damage to the line insulation caused by harmonics. Similarly, when localized abnormal temperature distributions occur, without considering current changes, it is difficult to accurately determine whether the abnormality is due to poor line contact or external environmental influences. These problems hinder the full effectiveness of existing electrical fire early warning technologies in practical applications, failing to provide reliable protection for the safe operation of electrical circuits. Summary of the Invention

[0004] The purpose of this invention is to provide an electrical fire early warning method and system to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an electrical fire early warning method, the method comprising: Real-time acquisition of a multi-dimensional dynamic parameter set in electrical circuits, including current fluctuation sequence, temperature gradient distribution data and insulation dielectric loss value; The dynamic parameter set is subjected to joint time-frequency domain decomposition to extract current harmonic distortion characteristics, temperature field spatial evolution mode and dielectric loss accumulation rate. The current anomaly index is generated based on the deviation of the harmonic distortion characteristics from the preset threshold. The thermal runaway risk level is calculated by combining the temperature field spatial evolution model. The insulation degradation coefficient is determined based on the dielectric loss accumulation rate. A comprehensive fire risk index for electrical circuits is constructed by integrating the current anomaly index, thermal runaway risk level, and insulation degradation coefficient. When the comprehensive fire risk index exceeds the dynamic warning threshold, a graded early warning signal is triggered and a hazard location map is generated.

[0006] Preferably, the joint time-frequency domain decomposition process includes: The current fluctuation sequence is decomposed into sub-band energy distributions of different frequency bands through wavelet packet transform, and the ratio of each sub-band energy to the fundamental frequency energy is used as a harmonic distortion feature. A spatial interpolation algorithm is used to reconstruct the three-dimensional temperature field from the temperature gradient distribution data, and the maximum rate of change of the temperature field within adjacent sampling periods is calculated as the spatial evolution model. The slope of the time-domain integral curve of the dielectric loss value reflects the accumulation rate, and the slope is normalized by combining the environmental humidity correction coefficient.

[0007] Preferably, the method for generating the current anomaly index is as follows: The weighted difference between the proportion of odd harmonics and the proportion of even harmonics in the harmonic distortion characteristics is selected as the first distortion factor. The ratio of the peak-to-valley difference to the root mean square value of the current fluctuation sequence is calculated as the second distortion factor. The current anomaly index is generated by multiplying the geometric mean of the first distortion factor and the second distortion factor by the line load factor.

[0008] Preferably, the method for calculating the thermal runaway risk level is as follows: Regions in the three-dimensional temperature field whose gradient changes exceed the critical value are identified as hotspot regions. The product of the area ratio of the hotspot region and the rate of temperature rise is used as the first thermal risk factor. The consistency coefficient of the change direction of three consecutive sampling periods in the spatial evolution model of the temperature field is extracted, and a second thermal risk factor is generated by combining it with the thermal conductivity of the material. The first and second hot risk factors are linearly weighted, and the preset risk level range is matched based on the weighting result.

[0009] Preferably, the method for determining the insulation degradation coefficient is as follows: The initial degradation value is generated by superimposing the attenuation coefficient corresponding to the aging years of the line on the basis of the cumulative rate of dielectric loss. The system detects the accelerating upward trend of the initial degradation value within a continuous time window. If the trend slope exceeds the historical average, a trend enhancement factor is introduced. The product of the initial degradation value and the trend enhancement factor is used as the insulation degradation coefficient.

[0010] Preferably, the method for constructing the comprehensive fire risk index is as follows: The current anomaly index is logarithmically transformed to compress the dynamic range, and the transformation result is multiplied by the first weighting coefficient. The thermal runaway risk level is mapped to the median of the numerical range and then multiplied by the second weighting coefficient. The insulation degradation coefficient is subjected to exponential smoothing and multiplied by a third weighting factor; The comprehensive fire risk index is generated by summing the three weighted results and dividing by the sum of the weight coefficients.

[0011] Preferably, the method for determining the dynamic warning threshold is as follows: The moving average and standard deviation of the comprehensive fire risk index in historical statistical data; Use the moving average plus three standard deviations as the base threshold; The base threshold is adjusted proportionally based on the deviation between the current ambient temperature and the standard operating temperature.

[0012] Preferably, the method for generating the hazard location map is as follows: Line sections where the current anomaly index exceeds the sub-threshold are marked as red highlighted areas. Overlay contour lines corresponding to the thermal runaway risk level onto a three-dimensional temperature field; Based on the magnitude of the insulation degradation coefficient, generate mesh shadows of different densities at corresponding locations; A multi-dimensional overlay of hazard location maps is generated by integrating red highlighted areas, contour lines, and grid shadows.

[0013] Preferably, the triggering method for the graded early warning signal is as follows: When the comprehensive fire risk index is in the first range, an audible and visual alarm is triggered and an early warning log is uploaded. When in the second zone, automatically disconnect non-critical load circuits and send remote alarm information; When in the third zone, the emergency power switching mechanism is activated and the fire control system is activated.

[0014] Preferably, the present invention also includes an electrical fire early warning system, the system including a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method described above.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By acquiring multi-dimensional dynamic parameters such as current fluctuation sequences, temperature gradient distribution data, and insulation dielectric loss values ​​in electrical lines in real time, this method overcomes the limitations of traditional single-parameter monitoring and can more comprehensively capture various status information during the operation of electrical lines. The acquisition of multi-dimensional parameters not only covers the dynamic changes in line current but also includes the spatial distribution characteristics of the temperature field and the real-time loss of the insulation dielectric, making the monitoring of line operation status more comprehensive and detailed, and avoiding the omission of potential risks due to blind spots caused by single-parameter monitoring.

[0016] Joint time-frequency domain decomposition of dynamic parameter sets enables precise extraction of current harmonic distortion features, spatial evolution patterns of the temperature field, and dielectric loss accumulation rates from complex parameter data. Compared to traditional time-domain or frequency-domain analysis methods, joint time-frequency domain decomposition technology can more effectively handle non-stationary signals, accurately identify different frequency components and their variation patterns in current harmonics, clearly grasp the spatial distribution trends of the temperature field, and understand the accumulation of insulation dielectric loss over time. This precise feature extraction provides a reliable foundation for subsequent risk assessment, making the judgment of potential problems in electrical circuits more accurate.

[0017] Based on the extracted features, a current anomaly index is generated, the thermal runaway risk level is calculated, and the insulation degradation coefficient is determined, enabling precise quantitative assessment of different risk types in electrical lines. The current anomaly index directly reflects the deviation of current harmonic distortion from a preset threshold, helping staff quickly understand abnormal current operation. The calculation of the thermal runaway risk level incorporates a temperature field spatial evolution model, accurately determining whether the line has potential thermal runaway hazards and their severity. The insulation degradation coefficient is determined based on the cumulative rate of dielectric loss, allowing real-time monitoring of the aging rate of the insulation medium. This multi-dimensional risk quantitative assessment method enables staff to clearly understand the specific status of the line in terms of current, temperature, insulation, and other aspects, allowing for targeted countermeasures.

[0018] By integrating multi-dimensional risk indicators to construct a comprehensive fire risk index, a holistic assessment of the overall operational status of electrical circuits is achieved. By effectively integrating current anomaly index, thermal runaway risk level, and insulation degradation coefficient, the limitations of single-indicator assessments are avoided, allowing for a comprehensive reflection of the overall safety status of the circuit. This comprehensive assessment method considers the interplay between different risk factors; for example, current anomalies may accelerate insulation aging, while insulation degradation may lead to current leakage and temperature increases, thus more accurately determining the overall risk of a fire on the circuit.

[0019] When the comprehensive fire risk index exceeds the dynamic warning threshold, a tiered early warning signal is triggered and a hazard location map is generated, improving the timeliness and practicality of the warning. The tiered early warning signal can issue corresponding warnings based on different risk levels, allowing staff to take different emergency response measures according to the warning level. This avoids the resource waste or untimely response problems of traditional single-warning-mode systems. The hazard location map can accurately indicate the specific location of hazards in the line, providing clear guidance for staff to quickly investigate and handle hazards, shortening hazard handling time, and reducing the probability of fire accidents.

[0020] This method uses dynamic warning thresholds, which can automatically adjust the warning standards according to different operating conditions and environmental conditions of electrical lines. Compared with traditional fixed thresholds, it is more in line with the changes in the actual operating conditions of the lines, reduces false alarms or missed alarms caused by the mismatch between fixed thresholds and actual operating conditions, and further improves the reliability and adaptability of the early warning system. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating the working principle of the electrical fire early warning method described in this invention. Figure 2 A flowchart for calculating the risk level of thermal runaway; Figure 3 A flowchart for constructing a comprehensive fire risk index. Detailed Implementation

[0022] 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.

[0023] Please see Figure 1This invention provides an electrical fire early warning method, comprising: acquiring current fluctuation sequences using a high-frequency current sensor at a millisecond-level sampling rate; measuring temperature gradient distribution data using a distributed temperature sensor network; and collecting insulation dielectric loss values ​​using a dielectric loss tangent detection device. The collected data is transmitted to a central processing unit for time-frequency domain joint decomposition processing. This processing simultaneously extracts feature parameters from both the time and frequency domains, including current harmonic distortion characteristics, temperature field spatial evolution patterns, and dielectric loss accumulation rates. Current harmonic distortion characteristics reflect the degree of current waveform distortion, the temperature field spatial evolution pattern characterizes the spatial variation law of temperature distribution, and the dielectric loss accumulation rate reflects the performance degradation rate of the insulation material. Based on these feature parameters, the system calculates three sub-indicators: a current anomaly index, a thermal runaway risk level, and an insulation degradation coefficient. The current anomaly index is generated by comparing harmonic distortion characteristics with a preset threshold; the thermal runaway risk level is calculated based on the temperature field spatial evolution pattern; and the insulation degradation coefficient is determined based on the dielectric loss accumulation rate. These three sub-indicators are fused to construct a comprehensive fire risk index, using a weighted fusion algorithm to integrate the influence of different physical quantities. When the comprehensive fire risk index exceeds the dynamic warning threshold, the system triggers a graded warning signal. The warning signal takes different response measures according to the risk level, and at the same time generates a hazard location map, which visually identifies high-risk areas in the line.

[0024] Example 1: The decomposition of the current fluctuation sequence begins with the application of wavelet packet transform. This transform uses the principle of multi-resolution analysis to decompose the non-stationary current signal into different frequency bands. The selected db4 wavelet basis function has appropriate tight support and regularity, which can effectively capture the transient distortion components in the current signal. The decomposition process decomposes the original current sequence layer by layer to form a complete binary tree structure, ultimately obtaining eight sub-band signals evenly distributed within the analysis frequency band. Each sub-band signal corresponds to different frequency components, such as the fundamental wave, various harmonics, and interharmonics. The energy value of each sub-band signal is calculated. The energy calculation is performed within a specified time window using the sum of squares formula. Then, the energy of each sub-band is compared with the energy value of the sub-band containing the power frequency fundamental wave, and the ratio is calculated and converted into a decibel value. This series of ratios constitutes the harmonic distortion feature characterizing the degree of current waveform distortion. This feature can distinguish different types of load characteristics and fault modes. The processing of temperature gradient distribution data relies on spatial interpolation algorithms to reconstruct the three-dimensional temperature field. The measurement points come from a distributed temperature sensor array pre-deployed at key nodes of the electrical line. These sensors are arranged in a certain spatial geometric relationship. Kriging interpolation is used to process these discrete spatial sampling points. This method estimates the temperature values ​​at unsampled locations based on variogram theory, providing optimal unbiased estimates while simultaneously providing the estimation error. The interpolation process generates a three-dimensional temperature field model composed of a regular voxel grid. Each voxel stores the estimated temperature value of that spatial unit, thus completely reproducing the temperature distribution throughout the entire line space. The extraction of spatial evolution patterns is achieved by comparing the changes in the three-dimensional temperature field within adjacent sampling periods. For each voxel, the difference in its temperature value between two consecutive periods is calculated, and then divided by the sampling time interval to obtain the rate of temperature change for that voxel. The entire three-dimensional temperature field is scanned to find the maximum value of the rate of change for all voxels. This maximum value reflects the intensity of the most drastic temperature change and is defined as the spatial evolution pattern of the temperature field. This pattern is particularly sensitive to the rapid formation of local hot spots.

[0025] Temperature gradient distribution data processing relies on spatial interpolation algorithms, particularly Kriging interpolation. This method aims to transform discrete temperature measurement points into a continuous spatial temperature field model. A distributed temperature sensor network is deployed along electrical lines. These sensors are spatially discrete, each recording the temperature value at its location, but the temperature conditions in other areas of the line without sensors are unknown. The core task of spatial interpolation algorithms is to predict the temperature value at any point within the entire region of interest using mathematical methods based on these limited spatial sampling points, thereby constructing a complete three-dimensional temperature field. This forms the basis for subsequent analysis of temperature spatial evolution patterns. Kriging interpolation, as a geostatistical method, begins with the construction and fitting of a variogram. The variogram is a tool for describing the spatial correlation of regionalized variables; in this application scenario, the regionalized variable is the temperature value. Half the variance of the temperature values ​​between all sensor point pairs is calculated, and a functional relationship is established between this variance and the spatial distance between corresponding point pairs. By plotting a scatter plot of the variance half-value as a function of distance and fitting a curve using an appropriate mathematical model (such as a spherical model, exponential model, or Gaussian model), a variogram model that quantifies the spatial correlation of temperature is obtained. This model clearly expresses the law of temperature similarity decaying with distance; that is, the closer the sensors are, the stronger the correlation of their temperature values. This step provides a theoretical basis for subsequent spatial prediction. After completing the variogram modeling, the unknown points whose temperatures need to be predicted are estimated. For each voxel center point in the 3D temperature field grid, the Kriging method first searches for known temperature sensor points within its neighborhood. The estimation process is not a simple weighted average, but rather assigns an optimal weight coefficient to each known point based on the spatial correlation structure calculated by the variogram model. The determination of the weight coefficient follows two basic principles: first, unbiasedness, that is, the sum of all weights is 1, ensuring that the estimated value will not be systematically overestimated or underestimated; second, optimality, that is, the variance of the difference between the estimated value and the true value is minimized, which means that the Kriging estimation is statistically optimal linear unbiased estimation. This process ultimately generates a unique weight combination for each point to be estimated, thereby calculating the predicted temperature value for that point. A significant feature of the Kriging method is that it can provide a measure of estimation uncertainty, namely the Kriging variance. For the temperature value of each predicted point, the algorithm simultaneously calculates a corresponding estimation variance. This variance value intuitively reflects the reliability of the prediction results. In the final generated three-dimensional temperature field model, the kriging variance of the predicted values ​​will be larger in areas far from all sensors and with weak data support, indicating higher uncertainty in the temperature estimation at those locations. This uncertainty information can be used as a credibility reference in subsequent risk assessments; for example, lower confidence weights can be assigned to the temperature evolution analysis results in high-uncertainty areas. After applying the above process to all grid points, the reconstruction from discrete sensor data to a continuous three-dimensional temperature field model is completed.This model is a volumetric data set containing spatial coordinates and temperature attributes, clearly demonstrating the temperature distribution along the route and in the surrounding space. For example, it can identify localized high-temperature regions (hotspots) and areas with large temperature gradients. This accurate spatial temperature distribution model is an indispensable input for subsequent calculations of the spatial evolution of the temperature field, providing a solid spatial data foundation for the analysis of thermal runaway risk. The advantage of the Kriging method lies in that it not only provides predicted values ​​but also reveals the uncertainties in the predictions, and rigorously considers the spatial correlation structure of variables, making the reconstructed temperature field more statistically and physically reasonable.

[0026] The processing of dielectric loss values ​​focuses on their accumulation rate over time. Continuous monitoring data of the dielectric loss tangent constitute a time series. Numerical integration of this series yields a curve showing the cumulative dielectric loss over time. The accumulation rate is calculated by fitting a linear relationship between the accumulation curve and a recent time window using the least squares method. The slope of the fitted line represents the accumulation rate of dielectric loss; a larger slope indicates faster degradation of the insulation material. Since ambient humidity significantly affects dielectric loss measurements, a correction coefficient based on real-time relative humidity measurement is introduced. This coefficient, obtained from a pre-established humidity-dielectric loss relationship curve, normalizes the calculated slope, eliminating interference from ambient humidity fluctuations and ensuring the accumulation rate more accurately reflects the state of the insulation material. The generation of the current anomaly index requires integrating harmonic distortion characteristics and information from the current fluctuation sequence. First, the proportions of odd and even harmonic energy are separated from the harmonic distortion characteristics. Odd harmonics are typically associated with power electronic equipment and certain faults, while even harmonics may indicate magnetic saturation or asymmetrical operation. The weighted difference between the proportions of odd and even harmonics is calculated. This weighting considers that odd harmonics typically pose a greater potential hazard to equipment heating and line safety, thus assigning them a higher weight. This weighted difference serves as the first distortion factor, reflecting the type and characteristics of current waveform distortion. Simultaneously, time-domain analysis is performed on the original current fluctuation sequence to identify the peaks and troughs within a complete cycle and calculate their differences. This peak-to-trough difference reflects the amplitude of current fluctuations. Then, the root mean square (RMS) value of the current sequence is calculated to characterize its effective power. The ratio of the peak-to-trough difference to the RMS value serves as the second distortion factor, reflecting the degree of current volatility. The geometric mean of the first distortion factor and the second distortion factor is calculated. This mean combines the distortion information in the frequency domain and the time domain. Finally, this geometric mean is multiplied by a line load factor, which is the ratio of the real-time operating current to the line's rated current carrying capacity. This factor is used to correct the baseline level of the anomaly index under different load conditions. The final product is the current anomaly index, which is a dimensionless value and its magnitude directly indicates the degree of anomaly risk in the current loop.

[0027] The implementation process relies on a high-speed data acquisition system and powerful computing capabilities. The acquisition of current fluctuation sequences requires a sampling rate at the kilohertz level to accurately capture harmonic components. The temperature sensor network needs synchronous acquisition capabilities to ensure the accuracy of spatial temperature field reconstruction. The dielectric loss measurement equipment must possess high precision and anti-interference capabilities. Data processing algorithms are implemented in embedded systems or industrial computers. Wavelet packet transform and spatial interpolation calculations require algorithm optimization to reduce computational latency and ensure real-time early warning. The calculation process for the current anomaly index is designed as a fixed data flow processing pipeline to ensure stability and repeatability from raw data to the index output. All weight parameters and thresholds during index generation can be calibrated and adjusted according to specific line characteristics and operational experience to adapt to different application scenarios.

[0028] Example 2: See Figure 2 The calculation of thermal runaway risk level is based on a detailed analysis of the three-dimensional temperature field and an in-depth identification of spatial evolution patterns. The acquisition of the three-dimensional temperature field relies on a continuous temperature distribution model reconstructed using a spatial interpolation algorithm. This model stores temperature data in the form of a voxel grid, where each voxel contains its spatial coordinates and corresponding temperature value. Gradient changes are calculated using the Sobel operator, which detects the rate of change of the temperature field in three spatial directions through convolution operations. For each voxel, its gradient magnitude is obtained by taking the square root of the sum of the squares of the partial derivatives in the three directions. Preset critical values ​​are determined based on the heat resistance level of the circuit insulation material. For example, for common cross-linked polyethylene insulation, the critical gradient might be set to 10 degrees Celsius per centimeter. The identification process involves traversing the entire temperature field and marking the set of voxels whose gradient magnitudes exceed this critical value.

[0029] The labeled voxels often form continuous regions in three-dimensional space. A region growing algorithm is used to aggregate these spatially adjacent high-gradient voxels, forming several independent hotspot regions. The geometric characteristics of each hotspot region are analyzed, and the ratio of its occupied surface area to the total surface area of ​​the entire line's temperature field is calculated to obtain the area proportion parameter. Simultaneously, for each hotspot region, temperature time-series data from the most recent sampling periods are extracted, and a linear regression method is used to fit a trend line of temperature change over time. The slope of this trend line represents the temperature rise rate of the hotspot region. The product of the area proportion and the temperature rise rate constitutes the first thermal risk factor, which comprehensively reflects the spatial scale of the overheated area and the rate of temperature deterioration. The analysis of the spatial evolution pattern focuses on the changes in the temperature field over time, specifically the overall direction of temperature change over three consecutive sampling periods. For each sampling period, the temperature change field relative to the previous period is calculated. This change field is treated as a vector field, and principal component analysis is used to extract its principal direction vector. Three principal direction vectors are obtained over three consecutive cycles. The cosine of the angle between any two adjacent vectors is calculated, and the average of these three cosine values ​​is taken to obtain the consistency coefficient of the changing direction. The closer this coefficient is to 1, the more consistent the changing direction of the temperature field, and the more stable the thermal runaway trend. The thermal conductivity of the circuit material is obtained from a material library. This coefficient is multiplied by the consistency coefficient to generate a second thermal risk factor. The thermal conductivity here acts as a physical scaling factor, because high thermal conductivity materials slow down the accumulation of local heat.

[0030] The linear weighted fusion of the primary and secondary overheating risk factors is a crucial step in risk level determination. The weighting coefficients are determined based on historical failure data statistical analysis and expert experience, typically assigning a higher weight to the primary overheating risk factor as it directly reflects the severity of the overheating phenomenon. The weighted calculation process uses the following mathematical expression:

[0031] in: This represents the quantified risk value after fusion. This indicates the first heat risk factor. This indicates the second heat risk factor. and These are the corresponding weight coefficients, and they satisfy... Weighting coefficients The value is usually significantly greater than For example, it might be set to 0.7, reflecting the dominant role of the direct overheating indicator in risk assessment. It can then be set to 0.3 to incorporate the corrective effect on the stability of the trend.

[0032] Calculated risk quantification value Ultimately, the risk levels are mapped to predefined risk level intervals. These intervals are predefined, for example, they might be divided into low-risk intervals (0 ≤ R < 0.3), medium-risk intervals (0.3 ≤ R < 0.7), and high-risk intervals (0.7 ≤ R ≤ 1.0). The mapping process uses an interval matching algorithm to map continuous risk levels. The values ​​are categorized into corresponding discrete risk levels. The entire calculation process starts from multi-dimensional physical quantities, employs rigorous mathematical operations, and ultimately outputs an intuitive thermal runaway risk level, providing a quantitative and graded assessment basis for subsequent comprehensive early warning decisions regarding the thermal state of the line. The implementation of this method relies on a high-precision temperature sensing network and efficient three-dimensional data processing capabilities, ensuring the accuracy and timeliness of risk perception.

[0033] Example 3: The determination process of the insulation degradation coefficient takes the dielectric loss accumulation rate as the core starting point and comprehensively considers the accelerating trend of line aging and performance degradation. The dielectric loss accumulation rate is obtained from the results of the previous time-frequency domain joint decomposition processing. It is a key parameter characterizing the rate at which the dielectric loss tangent of the insulating material accumulates over time. The calculation of the initial degradation value first involves time integration of the dielectric loss accumulation rate. The integration interval is set as a fixed time window, for example, selecting all dielectric loss accumulation rate data points collected in the most recent hour. The numerical integration adopts the trapezoidal rule, multiplying the rate values ​​between consecutive sampling points by the time interval and then summing them to obtain the total accumulation within the time window. This total accumulation reflects the overall loss of the insulating material in the near future. The attenuation coefficient corresponding to the aging years of the line needs to be evaluated based on the actual operating time and environmental conditions of the electrical line. The evaluation model refers to the basic principles of the Arrhenius equation, which describes the thermal aging law of materials. The coefficient calculation considers the cumulative duration of the line's operation and uses the average ambient temperature in the line's operating history as an accelerating aging factor. For lines that have exceeded their design life, the attenuation coefficient will increase significantly. The initial degradation value is generated by multiplying the total accumulated amount obtained from the above integration with the calculated attenuation coefficient. The attenuation coefficient, as a multiplier greater than 1, amplifies the degree of degradation caused by long-term aging, so that the initial degradation value not only contains current loss information, but also contains the accumulation of historical aging effects.

[0034] The detection of accelerating upward trends targets the changes in the initial degradation value at a finer time granularity, employing a sliding time window method to capture its dynamic behavior. A short time window, such as ten minutes, is set, and linear regression analysis is performed on the initial degradation value within this window to fit a trend line. The slope of this trend line is defined as the current trend slope. Calculating the historical mean requires backtracking through a longer period of historical data, such as taking the arithmetic mean of all such short-term trend slopes over the past 24 hours, which serves as a benchmark for judging whether the current change is abnormal. The introduction of the trend reinforcement factor is conditional upon the current trend slope being strictly greater than the historical mean. The magnitude of this factor is determined by calculating the ratio of the current slope to the historical mean. When the current growth momentum is significantly stronger than historical norms, this factor will be much greater than 1, thus significantly improving the final degradation coefficient assessment.

[0035] The final determination of the insulation degradation factor involves multiplying the initial degradation value (after aging correction) by a trend enhancement factor. This multiplication operation comprehensively incorporates not only the absolute cumulative amount of dielectric loss and the influence of historical aging, but also crucial information about whether the recent degradation rate has accelerated. Its mathematical expression is as follows:

[0036] Where: symbol This represents the final calculated insulation degradation coefficient, which is a dimensionless quantitative indicator. (Symbol) This represents the initial degradation value obtained after correction for the attenuation coefficient corresponding to the aging years of the line. This value reflects the combined effect of accumulated loss and long-term aging. (Symbol) The trend reinforcement factor is a coefficient of not less than 1, and its value depends on the degree to which the recent rate of deterioration deviates from the historical average.

[0037] The entire determination process constitutes a multi-level, multi-timescale comprehensive evaluation framework, from current monitoring data to historical operating status, and from steady-state assessment to dynamic trend analysis. This method avoids potential misjudgments that may arise from relying solely on instantaneous measurements. By introducing cumulative and trend analysis over time, it makes the assessment results of the insulation degradation coefficient more robust and predictive. The calculation process relies on a continuous and complete historical database to ensure the accuracy of trend analysis. All parameters, such as the time window length and the historical mean calculation period, can be configured and adjusted according to the monitoring requirements and data availability of specific application scenarios. The output of this coefficient provides a quantitative input on insulation condition deterioration for the subsequent construction of a comprehensive fire risk index.

[0038] Example 4: See Figure 3The construction method of the comprehensive fire risk index and the determination mechanism of the dynamic warning threshold are achieved by weighted fusion of multiple sub-indicators and establishing an adaptive alarm threshold. Assuming the early warning system is deployed on a 10 kV distribution trunk line, the system collects and generates specific values ​​for three basic sub-indicators at an assessment time. The current anomaly index, after monitoring and calculation by a high-frequency sensor, yields an initial value of 8.75. The thermal runaway risk level is determined to be "medium risk" through three-dimensional temperature field analysis, and the insulation degradation coefficient is calculated to be 0.62 based on the dielectric loss trend. The construction of the comprehensive fire risk index begins with the preprocessing and standardization of these three heterogeneous indicators to eliminate dimensional differences and balance the contribution of each indicator to the total risk. Since the initial value of the current anomaly index may have a large range, it is first logarithmically transformed to compress its dynamic range. The transformation function uses the natural logarithm plus one; substituting the initial value of 8.75 into the calculation yields a transformed result of approximately 2.27. This transformed result needs to be multiplied by a preset first weighting coefficient, which is set to 0.40 based on the statistical probability of current-induced faults in historical fire accidents. The thermal runaway risk level is a discrete level, which needs to be mapped to a continuous numerical range for arithmetic operations. The mapping rule is defined as follows: low risk corresponds to 0.15, medium risk to 0.50, and high risk to 0.85. The current medium risk level is mapped to the value 0.50, which is multiplied by a second weighting coefficient, set to 0.35, mainly considering the severity of the consequences of thermal runaway. The insulation degradation coefficient is already a standardized value. To smooth out its possible short-term fluctuations, exponential smoothing is applied with a smoothing factor of 0.3, resulting in a smoothed value of approximately 0.59. This value is then multiplied by a third weighting coefficient of 0.25. After completing the weighted calculations for each item, the three weighted results are added together: (2.27 * 0.40) + (0.50 * 0.35) + (0.59 * 0.25), yielding a total of approximately 1.218. Since the sum of the three weighting coefficients is 1.0, this sum is the final comprehensive fire risk index. This index is a dimensionless value between 0 and 1, and the higher the value, the greater the fire risk.

[0039] The determination of the dynamic warning threshold is an adaptive process based on historical data statistical analysis. Its purpose is to allow the alarm threshold to fluctuate with changes in the system's normal operating status, reducing false alarms. The system continuously records the comprehensive fire risk index values ​​calculated over a past period (e.g., 30 days), forming a time series data set. Refer to Table 1, which presents a simplified set of historical data to illustrate the basis for calculating the moving average and standard deviation: Table 1: Historical Data of Comprehensive Fire Risk Indicators for a Certain Railway Line

[0040] Based on this historical data, the system calculates the moving average of the most recent seven data points (e.g., from T-6 to the current day T), which is the arithmetic mean of these seven values. The calculated moving average is approximately 0.47. Simultaneously, the moving standard deviation of these seven data points is calculated to measure the dispersion of the historical data; the calculated moving standard deviation is approximately 0.08. The base threshold is set as the moving average plus three times the moving standard deviation, i.e., 0.47 + 3 * 0.08 = 0.71. This base threshold reflects the upper limit of the indicator value within the normal fluctuation range. The final dynamic warning threshold also needs to consider the impact of current environmental conditions, especially the deviation of the ambient temperature from the standard reference temperature (usually set at 25 degrees Celsius). If the current ambient temperature is 35 degrees Celsius, the deviation is +10 degrees Celsius. According to preset rules, for every 10 degrees Celsius the ambient temperature deviates from the standard temperature, the base threshold needs to be adjusted proportionally by 2%. The current temperature is higher than the standard value, with a positive deviation. Therefore, the base threshold is adjusted upwards by 2%, resulting in a dynamic warning threshold of 0.71*(1+0.02)=0.7242, approximately equal to 0.72. Comparing the currently calculated comprehensive fire risk index value of 0.61 with the dynamic warning threshold of 0.72, since 0.61 < 0.72, the system will not trigger a high-level alarm. This example illustrates the entire process of index construction and threshold determination, demonstrating how multi-source information is integrated into a comprehensive index and how a dynamic and reasonable alarm threshold is established based on historical operating status and the current environment.

[0041] Example 5: The generation of the hazard location map and the triggering of graded early warning signals transform abstract risk data into intuitive visual information and specific emergency operation instructions. The generation of the hazard location map begins with the construction of a line spatial model. This model is typically based on the actual route of the power distribution line and the electrical connection diagram, digitally modeled to form a two-dimensional or simplified three-dimensional map containing coordinate information and electrical node relationships. Spatial labeling of the current anomaly index is performed first. The system reads the current anomaly index of all monitoring points and compares it with a preset sub-threshold, which is usually set to the higher percentile of historical data (e.g., the 95th percentile). For all line sections where the index exceeds the sub-threshold, a highlight mark is made at the corresponding spatial map location. The mark is filled with a striking red and accompanied by a flashing effect to attract the operator's special attention. The range and shape of the mark are consistent with the geometric layout of the actual line, enabling operators to quickly locate the specific branch or switchgear with the current anomaly.

[0042] The overlay of temperature field information is more complex, requiring the fusion and display of the thermal runaway risk level calculated in the aforementioned embodiments with the three-dimensional temperature field model. Based on the different risk levels, the system draws contour lines with different attributes on the temperature field isosurface map. For example, dense red contour lines are drawn for high-risk areas, with a line spacing possibly set to five degrees Celsius; slightly wider yellow contour lines are drawn for medium-risk areas; and sparse green contour lines represent low-risk areas. The contour line generation algorithm traverses the temperature field data points, finding points with equal temperature values ​​and connecting them into smooth curves. These curves are layered on top of the line model, clearly showing the spatial distribution gradient of temperature and the boundaries of different temperature rise areas. The visualization of insulation degradation information is achieved through semi-transparent grid shadows. The insulation degradation coefficient value at each monitoring point is mapped to a shadow pattern of a specific density. The mapping rule is usually linear: the lower the degradation coefficient value, the sparser and lighter the corresponding grid shadow, and vice versa. These shadows, with their transparency properties, are precisely overlaid on the corresponding geographical locations of the line sections. When operators observe the map, they can see the underlying line layout through the shadows and clearly perceive the differences in insulation conditions between different sections. Three visualization elements—red highlighted areas, temperature contour lines, and grid shadows—are ultimately superimposed on the same base map using graphic compositing technology. The compositing process requires handling the occlusion relationships and transparency blending between layers to ensure all information is clearly discernible and does not obscure each other, forming a hazard location map that integrates multi-dimensional risk information on current, temperature, and insulation.

[0043] The triggering mechanism of the tiered early warning signal is strictly linked to the numerical range of the comprehensive fire risk index, which is pre-defined according to the principle of increasing risk. Assume the system's set ranges are: first range 0.5 to 0.7, second range 0.7 to 0.9, and third range 0.9 to 1.0. When the comprehensive fire risk index calculated by the system falls into the first range, for example, a value of 0.6, a preset primary response procedure will be triggered. The primary response typically includes activating the audible and visual alarm device at the local monitoring center. The alarm will emit a specific frequency beep and a yellow warning light will illuminate. Simultaneously, key information about the warning event, such as the time of occurrence, risk index value, and main risk factors, will be automatically recorded in the system's early warning log database, generating a pending event record.

[0044] When the risk index rises further and enters the second range, for example, reaching 0.78, the system's response measures automatically escalate. In addition to maintaining local audible and visual alarms, the system will send detailed alarm information to preset remote monitoring terminals or management personnel's mobile terminals. The information includes the risk location, risk type, and recommended actions. More importantly, the system will automatically execute control logic to cut off the power supply to the circuits marked as non-critical loads on the line. The list of non-critical loads needs to be predefined and may include unnecessary lighting circuits, auxiliary equipment power supplies, etc. This action aims to reduce the total load current, alleviate line pressure, and buy time for subsequent handling.

[0045] If the risk indicators continue to deteriorate and exceed the lower limit of the third interval, for example, rising to 0.93, the system will activate the highest level of emergency response. At this time, in addition to all the alarm and control measures mentioned above, the emergency power switching mechanism will be activated immediately. If the line has backup power, the system will issue an instruction to switch to backup power to isolate potential sources of failure. Simultaneously, the system sends a linkage signal to the building's fire control system via hardwiring or standard communication protocols. This signal may trigger the closing of fire doors, the activation of emergency ventilation systems, or provide early warning for gas extinguishing systems, achieving deep linkage between electrical fire early warning and overall fire protection facilities. The entire graded triggering process emphasizes automation and timeliness, minimizing delays in human intervention. However, a brief confirmation prompt is usually given on the human-machine interface before important operation instructions are issued to prevent accidental activation. The generation of the graph and the triggering of the early warning complement each other. The graph provides decision support for operators, while the automated early warning response ensures the rapid execution of measures in emergency situations.

[0046] 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.

[0047] 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. An electrical fire early warning method, characterized in that, include: Real-time acquisition of a multi-dimensional dynamic parameter set in electrical circuits, including current fluctuation sequence, temperature gradient distribution data and insulation dielectric loss value; The dynamic parameter set is subjected to joint time-frequency domain decomposition to extract current harmonic distortion characteristics, temperature field spatial evolution mode and dielectric loss accumulation rate. The current anomaly index is generated based on the deviation of the harmonic distortion characteristics from the preset threshold. The thermal runaway risk level is calculated by combining the temperature field spatial evolution model. The insulation degradation coefficient is determined based on the dielectric loss accumulation rate. A comprehensive fire risk index for electrical circuits is constructed by integrating the current anomaly index, thermal runaway risk level, and insulation degradation coefficient. When the comprehensive fire risk index exceeds the dynamic warning threshold, a graded early warning signal is triggered and a hazard location map is generated.

2. The electrical fire early warning method according to claim 1, characterized in that, The joint time-frequency domain decomposition process includes: The current fluctuation sequence is decomposed into sub-band energy distributions of different frequency bands through wavelet packet transform, and the ratio of each sub-band energy to the fundamental frequency energy is used as a harmonic distortion feature. A spatial interpolation algorithm is used to reconstruct the three-dimensional temperature field from the temperature gradient distribution data, and the maximum rate of change of the temperature field within adjacent sampling periods is calculated as the spatial evolution model. The slope of the time-domain integral curve of the dielectric loss value reflects the accumulation rate, and the slope is normalized by combining the environmental humidity correction coefficient.

3. The electrical fire early warning method according to claim 2, characterized in that, The method for generating the current anomaly index is as follows: The weighted difference between the proportion of odd harmonics and the proportion of even harmonics in the harmonic distortion characteristics is selected as the first distortion factor. The ratio of the peak-to-valley difference to the root mean square value of the current fluctuation sequence is calculated as the second distortion factor. The current anomaly index is generated by multiplying the geometric mean of the first distortion factor and the second distortion factor by the line load factor.

4. The electrical fire early warning method according to claim 3, characterized in that, The method for calculating the thermal runaway risk level is as follows: Regions in the three-dimensional temperature field whose gradient changes exceed the critical value are identified as hotspot regions. The product of the area ratio of the hotspot region and the rate of temperature rise is used as the first thermal risk factor. The consistency coefficient of the change direction of three consecutive sampling periods in the spatial evolution model of the temperature field is extracted, and a second thermal risk factor is generated by combining it with the thermal conductivity of the material. The first and second hot risk factors are linearly weighted, and the preset risk level range is matched based on the weighting result.

5. The electrical fire early warning method according to claim 4, characterized in that, The method for determining the insulation degradation coefficient is as follows: The initial degradation value is generated by superimposing the attenuation coefficient corresponding to the aging years of the line on the basis of the cumulative rate of dielectric loss. The system detects the accelerating upward trend of the initial degradation value within a continuous time window. If the trend slope exceeds the historical average, a trend enhancement factor is introduced. The product of the initial degradation value and the trend enhancement factor is used as the insulation degradation coefficient.

6. The electrical fire early warning method according to claim 5, characterized in that, The method for constructing the comprehensive fire risk index is as follows: The current anomaly index is logarithmically transformed to compress the dynamic range, and the transformation result is multiplied by the first weighting coefficient. The thermal runaway risk level is mapped to the median of the numerical range and then multiplied by the second weighting coefficient. The insulation degradation coefficient is subjected to exponential smoothing and multiplied by a third weighting factor; The comprehensive fire risk index is generated by summing the three weighted results and dividing by the sum of the weight coefficients.

7. The electrical fire early warning method according to claim 6, characterized in that, The method for determining the dynamic warning threshold is as follows: The moving average and standard deviation of the comprehensive fire risk index in historical statistical data; Use the moving average plus three standard deviations as the base threshold; The base threshold is adjusted proportionally based on the deviation between the current ambient temperature and the standard operating temperature.

8. The electrical fire early warning method according to claim 7, characterized in that, The method for generating the hazard location map is as follows: Line sections where the current anomaly index exceeds the sub-threshold are marked as red highlighted areas. Overlay contour lines corresponding to the thermal runaway risk level onto a three-dimensional temperature field; Based on the magnitude of the insulation degradation coefficient, generate mesh shadows of different densities at corresponding locations; A multi-dimensional overlay of hazard location maps is generated by integrating red highlighted areas, contour lines, and grid shadows.

9. The electrical fire early warning method according to claim 8, characterized in that, The triggering method for the graded early warning signal is as follows: When the comprehensive fire risk index is in the first range, an audible and visual alarm is triggered and an early warning log is uploaded. When in the second zone, automatically disconnect non-critical load circuits and send remote alarm information; When in the third zone, the emergency power switching mechanism is activated and the fire control system is activated.

10. An electrical fire early warning system, characterized in that, It includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the steps of the method according to any one of claims 1 to 9.

Images