Disaster monitoring method and system based on multi-dimensional electromagnetic data recognition and storage medium

Abstract

The application relates to the technical field of safe operation and monitoring of power systems, and discloses a disaster monitoring method and system based on multi-dimensional electromagnetic data identification and a storage medium. The method collects voltage, current and magnetic field data from a first power node, identifies intermittent fluctuation characteristics from the voltage data, identifies transient fluctuation characteristics from the current data, and respectively calculates intermittent matching values and transient matching values to obtain electric data matching values. When the electric data matching values exceed a preset electric reference value, high and low frequency change characteristics in the magnetic field data are analyzed and a warning is given; corresponding values are calculated according to the high and low frequency characteristics and corresponding reference values, and the distance of a disaster is judged by comparing the corresponding values. The application utilizes multi-dimensional electromagnetic data to realize early and accurate warning of disasters of power equipment and fault distance judgment, effectively improves monitoring accuracy and operation and maintenance efficiency, and reduces disaster losses.

Description

Technical Field

[0001] The present application relates to the technical field of safe operation and monitoring of power systems, and in particular to a disaster monitoring method, system and storage medium based on multi-dimensional electromagnetic data recognition. Background Art

[0002] Monitoring power equipment for hazards, especially early identification of fire hazards, is crucial for the safe operation of power systems. Power equipment is subjected to complex operating conditions involving high voltages and high currents for extended periods. Issues such as aging, insulation failure, poor contact, and overload can lead to localized overheating, arcing, and even flames. For example, partial discharge from aging transformer winding insulation and abnormally high contact resistance caused by oxidation of cable joints can be precursors to fire.

[0003] Currently, power equipment disaster monitoring solutions based on infrared temperature detection are a mainstream technology in the industry. This solution uses infrared thermal imagers, infrared sensors, and other devices to collect real-time surface temperature data on equipment. It then identifies abnormally hot areas on the equipment through temperature threshold determination or thermal imaging analysis.

[0004] However, existing infrared temperature detection solutions have significant limitations in disaster early warning. Because infrared technology only reflects the surface temperature of equipment, it cannot detect early-stage hazards before significant temperature anomalies occur. By the time infrared equipment detects excessive temperatures, the fire may have already entered a significant stage of development or even reached a significant scale. At this point, taking action often makes it difficult to prevent further damage. Summary of the Invention

[0005] In order to be able to warn of fire earlier than infrared recognition methods, the present application provides a disaster monitoring method, system and storage medium based on multi-dimensional electromagnetic data recognition.

[0006] In a first aspect, the present application provides a disaster monitoring method based on multi-dimensional electromagnetic data recognition, which adopts the following technical solutions:

[0007] A disaster monitoring method based on multi-dimensional electromagnetic data recognition includes the following steps:

[0008] Acquire first voltage data, first current data, and first magnetic field data based on the first power node;

[0009] identifying intermittent fluctuation characteristics based on the first voltage data, and identifying transient fluctuation characteristics based on the first current data;

[0010] Calculating an intermittent matching value based on the identified intermittent fluctuation characteristics, and calculating a transient matching value based on the identified transient fluctuation characteristics;

[0011] Calculating an electrical data matching value based on the intermittent matching value and the transient matching value;

[0012] If the electrical data matching value is greater than a preset electrical reference value, a first frequency change feature and a second frequency change feature are identified from the first magnetic field data, and a disaster warning prompt is issued; wherein a first frequency corresponding to the first frequency change feature is lower than a second frequency corresponding to the second frequency change feature;

[0013] Calculating a first corresponding value based on the identified first frequency change feature and a preset first frequency reference value, and calculating a second corresponding value based on the identified second frequency change feature and a preset second frequency reference value;

[0014] Compare the first corresponding value with the second corresponding value. If the first corresponding value is greater than the second corresponding value, a long-distance disaster prompt is issued; otherwise, a short-distance disaster prompt is issued.

[0015] By adopting the above technical solution, a disaster monitoring method based on multidimensional electromagnetic data identification collects voltage, current, and magnetic field data from power nodes, identifies intermittent voltage fluctuation characteristics and transient current fluctuation characteristics, and calculates intermittent matching values, transient matching values, and electrical data matching values. When the electrical data matching value exceeds the preset electrical reference value, it further analyzes the high- and low-frequency variation characteristics in the magnetic field data to issue a disaster warning. At the same time, based on the corresponding values ​​calculated from the high- and low-frequency characteristics and the corresponding reference values, it determines the distance to the disaster. This method overcomes the limitation of traditional infrared monitoring that relies on the surface temperature of the equipment and uses electromagnetic signals to detect early-stage fault hazards in advance.

[0016] Optionally, the step of identifying intermittent fluctuation characteristics includes:

[0017] filtering the first voltage data to obtain temporary voltage data;

[0018] dividing the temporary voltage data into a plurality of continuous voltage data according to a preset time window;

[0019] Calculating the fluctuation amplitude and fluctuation duration of each of the continuous voltage data;

[0020] If the fluctuation amplitude is within a preset amplitude range and the fluctuation duration is within a preset duration range, then the voltage continuous data has the intermittent fluctuation feature, and the intermittent fluctuation feature includes the fluctuation amplitude and the fluctuation duration;

[0021] The step of calculating the intermittent matching value comprises:

[0022] Calculating a fluctuation similarity value based on the fluctuation amplitude and a preset amplitude template;

[0023] Calculating a duration similarity value based on the fluctuation duration and a preset duration template;

[0024] An interval matching value is calculated based on the fluctuation similarity value and the duration similarity value.

[0025] By adopting the above technical solution, the first voltage data is filtered and pre-processed to eliminate noise such as environmental electromagnetic interference and improve data quality. The data is then divided into continuous segments according to the preset time window to accurately capture the voltage fluctuation trend in a short period of time. By setting the fluctuation amplitude and duration range, a two-dimensional judgment is made on the continuous voltage data, effectively eliminating the misjudgment of a single parameter and realizing the quantitative identification of intermittent fluctuation characteristics. When calculating the intermittent matching value, the actual detected fluctuation amplitude and duration are compared with the preset template, and the intermittent matching value is comprehensively obtained after calculating the similarity value to achieve a comprehensive score of the voltage fluctuation characteristics and dynamically evaluate the degree of fault. This method avoids judgment deviations caused by individual differences in equipment or environmental changes, so that the monitoring results have a unified standard, effectively enhancing the ability to identify early fault hazards of power equipment and the reliability of early warning.

[0026] Optionally, the method further comprises the following steps:

[0027] Acquiring second voltage data, second current data, and second magnetic field data based on a second power node; wherein the second power node is located within a set distance range of the first power node;

[0028] In a recently set time period, the waveform similarity between the first voltage data and the second voltage data is calculated as a voltage similarity value; the waveform similarity between the first current data and the second current data is calculated as a current similarity value; and the waveform similarity between the first magnetic field data and the second magnetic field data is calculated as a magnetic field similarity value.

[0029] A comprehensive similarity value is obtained by weighted calculation based on the voltage similarity value, the current similarity value and the magnetic field similarity value;

[0030] If the comprehensive similarity value is greater than a preset reference value, all disaster situation prompts are shielded.

[0031] By adopting the above technical solution, the similarity of multiple-dimensional signals at the first power node and the second power node within a set distance range is compared. If similar changes occur in the data of the two nodes at the same time, it is more likely to be caused by external environmental interference or overall system fluctuations rather than equipment failure itself. At this time, the disaster prompt is shielded to effectively avoid false alarms caused by environmental factors.

[0032] Optionally, the step of identifying transient fluctuation characteristics includes:

[0033] filtering the first current data to obtain temporary current data;

[0034] Extracting a plurality of current peak data from the temporary current data according to a preset peak template;

[0035] Calculating a peak height and a peak width of the current peak data, and calculating a peak steepness value according to the peak height and the peak width;

[0036] If the peak steepness value is within the preset peak range, the current peak data has the transient fluctuation characteristics, and the transient fluctuation characteristics include the peak height, the peak width, and the peak steepness value;

[0037] The step of calculating the transient matching value comprises:

[0038] Calculating a height similarity value based on the peak height and a preset height template;

[0039] Calculating a width similarity value based on the peak width and a preset width template;

[0040] Calculating a temporary matching value based on the height similarity value and the width similarity value;

[0041] The temporary matching value is adjusted according to the positive correlation of the peak steepness value, and the adjusted temporary matching value is a transient matching value.

[0042] By employing this technical solution, filtering effectively eliminates environmental noise interference, ensuring the authenticity and reliability of current data. Pre-set spike templates are then used to quickly locate current spikes lasting from milliseconds to seconds. This closely matches the transient current surges caused by poor contact and sudden load changes before a line fire. Compared to traditional monitoring methods, this solution can capture transient current fluctuations at the earliest stages of a fault (e.g., when a conductor begins to overheat locally but before the temperature rises significantly, and fuse marks (e.g., spherical melt beads and dense pores) are present at the corresponding locations), significantly accelerating disaster warning times.

[0043] Optionally, the step of comparing the first corresponding value with the second corresponding value further includes the following steps:

[0044] Calculating a corresponding difference between the first corresponding value and the second corresponding value;

[0045] During the preset monitoring time period, if the absolute value of the change in the corresponding difference is greater than the preset change reference value, it indicates that the disaster situation is changing; if the corresponding difference changes from a positive number to a negative number, it indicates that the disaster is approaching; if the corresponding difference changes from a negative number to a positive number, it indicates that the disaster is spreading.

[0046] By adopting the above technical solution, when the absolute value of the corresponding difference change exceeds the preset change reference value, it immediately indicates that the disaster situation is changing. Compared with the traditional method of only monitoring fixed thresholds, this solution can promptly detect dynamic trends such as acceleration and deceleration during the development of the disaster. When the difference changes from a positive number to a negative number, it directly reflects that the disaster situation is developing in the direction of approaching the monitoring node, indicating that the local risk is increasing, which facilitates operation and maintenance personnel to prepare equipment protection and personnel evacuation in advance; when the difference changes from a negative number to a positive number, it indicates that the scope of the disaster is gradually expanding, which helps to promptly initiate measures such as isolating the fault area and deploying emergency repair resources. This trend-based early warning method greatly improves the foresight and effectiveness of emergency response in the power system.

[0047] Optionally, a plurality of contact resistance sensors are arranged in a set area around the plurality of power modules, and the contact resistance sensors are randomly arranged in the area; the method further comprises the following steps:

[0048] Acquiring contact resistance data based on the contact resistance sensor;

[0049] If a mutation term is identified from the contact resistance data, a periodic mutation feature is identified within a recent preset mutation time period;

[0050] If the periodic mutation feature is identified, a disaster warning prompt for the contact device will be issued.

[0051] By implementing this technical solution, we overcome the limitations of fixed-point monitoring, comprehensively cover potential fault points in complex power systems, eliminate blind spots, and increase the probability of detecting contact failure hazards. Real-time monitoring of contact resistance changes triggers an alarm at the early stages of a fault (when resistance increases significantly but temperature does not rise significantly), significantly exceeding the early warning time of traditional temperature monitoring methods. By identifying periodic mutation characteristics, we effectively filter out accidental mutations caused by normal equipment fluctuations, reduce false alarm rates, and ensure the reliability of early warnings.

[0052] Optionally, a plurality of insulation resistance sensors are arranged in a set area around the plurality of electrical devices, and positions of the insulation resistance sensors in the area are randomly arranged; the method further comprises the following steps:

[0053] Acquiring insulation resistance data based on the insulation resistance sensor;

[0054] Calculating an average value of the plurality of insulation resistance data; if a gradually decreasing trend of change is identified from the average value, identifying a continuous change feature within a recent preset mutation time period;

[0055] If the above-mentioned continuous change characteristics are identified, an early warning of power equipment disaster situation will be issued.

[0056] By adopting the above technical solutions, random point placement can eliminate monitoring blind spots and cover potential weak points; trend identification can provide early warning of insulation performance degradation, significantly earlier than traditional threshold judgment; continuous feature verification can eliminate interference from accidental factors and improve warning accuracy.

[0057] Optionally, the method further comprises the following steps:

[0058] If a persistent change feature is identified within the most recent preset mutation time period, analyzing the first current data;

[0059] If a residual current is identified from the first current data and the residual current is greater than a preset residual reference value, analyzing the first magnetic field data;

[0060] If a second frequency change feature is identified from the first magnetic field data, a leakage arc fire warning is issued.

[0061] By adopting the above technical solution, a progressive analysis logic of "insulation resistance trend determination - residual current detection - magnetic field frequency characteristic analysis" is employed. The system first identifies potential risks by detecting a continuous decrease in insulation resistance, then detects whether the residual current exceeds the standard, and finally analyzes the high-frequency characteristics of the magnetic field. This triple verification effectively filters out single-data misjudgments (such as occasional current fluctuations caused by environmental interference), avoids false alarms caused by a single abnormal indicator, and ensures the high reliability of leakage arc fire warnings.

[0062] In a second aspect, the present application provides a disaster monitoring system based on multi-dimensional electromagnetic data recognition, which adopts the following technical solutions:

[0063] A disaster monitoring system based on multi-dimensional electromagnetic data recognition includes a processor, wherein the processor executes the steps of any one of the above-mentioned disaster monitoring methods based on multi-dimensional electromagnetic data recognition.

[0064] In a third aspect, the present application provides a storage medium that adopts the following technical solution:

[0065] A storage medium stores a program, which, when executed by a processor, implements the steps of any one of the above-mentioned disaster monitoring methods based on multi-dimensional electromagnetic data recognition.

[0066] In summary, the present application includes at least one of the following beneficial technical effects: by collecting voltage, current, and magnetic field data of power nodes, combined with contact resistance and insulation resistance sensor data, the disaster characteristics of power equipment are captured in multiple dimensions. After filtering and segmentation, the voltage data identifies intermittent fluctuations, and the current data extracts peak characteristics. The two are combined to calculate the electrical data matching value, and the high and low frequency characteristics of the magnetic field are linked to achieve disaster warning and distance judgment; by comparing the similarity of multi-dimensional data of different nodes, false alarms caused by environmental interference can be shielded; by analyzing the contact resistance mutation cycle and the insulation resistance decline trend, the risk of poor contact and insulation degradation can be discovered in advance; using "insulation-current-magnetic field" progressive analysis, the hidden dangers of leakage arc fire can be accurately identified. This method breaks through the limitations of traditional monitoring, realizes early and accurate warning of power equipment disasters, dynamic trend tracking and fault type location, significantly improves monitoring accuracy and operation and maintenance efficiency, and reduces the risk of power disasters and economic losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 This is a step diagram of a disaster monitoring method based on multi-dimensional electromagnetic data recognition.

[0068] Figure 2 It is a step diagram for identifying intermittent fluctuation characteristics.

[0069] Figure 3 This is a diagram of the steps for calculating the intermittent matching value.

[0070] Figure 4 This is a step diagram of a method for disaster monitoring based on adding a second power node. DETAILED DESCRIPTION

[0071] Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings.

[0072] Throughout this specification, reference to the terms "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0073] The present application discloses a disaster monitoring method based on multi-dimensional electromagnetic data recognition, referring to Figure 1 , including the following steps:

[0074] First voltage data, first current data, and first magnetic field data are acquired based on a first power node. The first power node can be a cable branch point or transformer point in a factory. Real-time acquisition of first voltage data reflects insulation status and power supply stability; acquisition of first current data reveals load variations and contact performance; and acquisition of first magnetic field data characterizes electromagnetic energy distribution and discharge characteristics.

[0075] The intermittent fluctuation feature is identified based on the first voltage data, and the transient fluctuation feature is identified based on the first current data.

[0076] Reference Figure 2 , wherein the step of identifying intermittent fluctuation characteristics includes: filtering the first voltage data to obtain temporary voltage data; dividing the temporary voltage data into multiple voltage continuous data according to a preset time window, and the time intervals corresponding to the multiple voltage continuous data are different. A low-pass filter can be used to eliminate high-frequency noise, and the baseline drift can be smoothed by a moving average algorithm (window length 100ms) to obtain pure temporary voltage data. Calculate the fluctuation amplitude and fluctuation duration of each voltage continuous data; if the fluctuation amplitude is within the preset amplitude range and the fluctuation duration is within the preset duration range, then the voltage continuous data has intermittent fluctuation characteristics, and the intermittent fluctuation characteristics include fluctuation amplitude and fluctuation duration. Divide the voltage sequence into variable time windows, and dynamically adjust the window length to 10ms-1s to adapt to the fluctuation characteristics of different periods; calculate for each window data:

[0077] Fluctuation amplitude: (peak value - valley value) / rated voltage × 100%;

[0078] Fluctuation duration: the duration from the onset of fluctuation to the return to steady state;

[0079] When the fluctuation amplitude falls within the range of [1%, 5%] and the duration is between [5ms, 50ms], it is considered a valid intermittent fluctuation characteristic, preventing transient overvoltages (such as lightning pulses) from being mistakenly included in the analysis. Under normal operating conditions, the voltage waveform presents a stable sinusoidal waveform. However, when the insulation dielectric ages, it triggers periodic partial discharges, resulting in intermittent, slight oscillations in the voltage waveform. The typical amplitude is 1%-5% of the rated voltage and the duration is 5-50ms. These intermittent, slight oscillations reflect the degree of insulation degradation.

[0080] The steps for identifying transient fluctuation characteristics include: filtering the first current data to obtain temporary current data; extracting multiple current spike data from the temporary current data according to a preset spike template; calculating the peak height and peak width of the current spike data, and calculating the peak steepness value based on the peak height and peak width; if the peak steepness value is within the preset peak range, the current spike data has transient fluctuation characteristics, and the transient fluctuation characteristics include peak height, peak width, and peak steepness value. Poor contact or sudden load changes can cause high-frequency current spikes, such as a rising edge of less than 1ms and a peak value of 1.5-3 times the rated current. The steepness (peak steepness = peak height / peak width) is positively correlated with the contact resistance, which can be used to quantify the severity of the contact fault.

[0081] Three typical spike templates are preset (corresponding to mild, moderate, and severe contact faults). The template parameters include:

[0082] Peak height threshold: 1.2In, 1.5In, 2.0In (In is the rated current);

[0083] Peak width range: 1-5ms, 0.5-3ms, 0.1-1ms;

[0084] The cross-correlation algorithm is used to quickly match the template with the real-time current data and extract the current spikes that meet the conditions.

[0085] Steepness calculation:

[0086] The formula for defining the spike steepness is: S = H / W, where H is the spike height (relative to the steady-state current amplitude) and W is the spike width (ms). When S ≥ 0.5 A / ms, it is considered to have transient fluctuation characteristics. This threshold corresponds to a fault scenario with a contact resistance greater than 50 mΩ.

[0087] For mild fault scenarios, the preset peak steepness value range is 0.5-1.5A / ms;

[0088] The corresponding range for moderate faults is 1.5-3.0A / ms;

[0089] The severe fault range is set to be greater than 3.0A / ms.

[0090] An intermittent matching value is calculated based on the identified intermittent fluctuation feature, and a transient matching value is calculated based on the identified transient fluctuation feature.

[0091] Reference Figure 3 , wherein the step of calculating the intermittent matching value includes: calculating the fluctuation similarity value based on the fluctuation amplitude and the preset amplitude template; calculating the duration similarity value based on the fluctuation duration and the preset duration template; and calculating the intermittent matching value by a weighted method based on the fluctuation similarity value and the duration similarity value.

[0092] Fluctuation similarity value: Sv=1-|A-A0| / A0; where A is the measured fluctuation amplitude and A0 is the template amplitude, which is 3%.

[0093] Duration similarity value: St = 1-|T-T0| / T0, where T is the measured duration and T0 is the template duration, which is 20ms.

[0094] Intermittent matching value: Mi=0.5×Sv+0.5×St. The weight coefficient is set according to the sensitivity of voltage fluctuation to insulation fault.

[0095] Among them, the step of calculating the transient matching value includes: calculating the height similarity value based on the peak height and the preset height template; calculating the width similarity value based on the peak width and the preset width template; and calculating the temporary matching value in a weighted manner based on the height similarity value and the width similarity value.

[0096] Height similarity value: Sh = 1-|H-H0| / H0, where Sh is the height similarity value, H is the actual detected current spike height, and H0 is the preset height template value corresponding to the fault level.

[0097] Width similarity value: Sw=1-|W-W0| / W0, where Sw is the width similarity value, W is the actual detected current spike width, and W0 is the preset width template value corresponding to the fault level.

[0098] Temporary matching value: Mtemporary = 0.5 × Sh + 0.5 × Sw. Based on a large number of experiments and data analysis, combined with the operating characteristics of different power equipment, appropriate weight coefficients are assigned to the height similarity value and width similarity value.

[0099] The temporary matching value is adjusted according to the positive correlation of the peak steepness value, and the adjusted temporary matching value is the transient matching value.

[0100] The electrical data matching value is calculated based on the weighted sum of the intermittent matching value and the transient matching value.

[0101] The electrical data matching value = 0.5 × intermittent matching value + 0.5 × transient matching value; the value is selected based on a large amount of experimental data, historical fault case analysis, and the operating principle of power equipment.

[0102] If the electrical data match value is greater than the preset electrical reference value, the first and second frequency change characteristics are identified from the first magnetic field data, and a disaster warning is issued. The first frequency corresponding to the first frequency change characteristic is lower than the second frequency corresponding to the second frequency change characteristic. The electromagnetic radiation generated by partial discharge contains a variety of frequency components: the first frequency corresponds to a low-frequency component, such as 50 Hz, which primarily originates from power-frequency electromagnetic coupling, and its intensity variation reflects the spatial distance of the discharge area. The second frequency corresponds to a high-frequency component, such as 100-200 Hz, which is related to the discharge pulse frequency, and sudden changes in intensity indicate intensified discharge activity.

[0103] Based on the identified first frequency change characteristics and the preset first frequency reference value, a first corresponding value is calculated, and the first corresponding value = the rate of change of the first frequency relative to the historical baseline value / the first frequency reference value; based on the identified second frequency change characteristics and the preset second frequency reference value, a second corresponding value is calculated, and the second corresponding value = the rate of change of the second frequency relative to the historical baseline value / the second frequency reference value. In addition, the first frequency reference value is adjusted in an anti-correlation manner according to the inductive distribution of the load, and the second frequency reference value is adjusted in a positive correlation manner. In actual power systems, the inductive distribution of the load, such as the proportion of inductive devices such as motors and transformers, will significantly affect the frequency characteristics of the magnetic field signal. In order to improve the accuracy of judgment, the system introduces a load adaptive adjustment algorithm:

[0104] The first frequency reference value = the basic reference value 1 × (1-k × L), where k is the adjustment coefficient and L is the proportion of inductive load in the total load. Because inductive loads have a shielding effect on the power frequency magnetic field, anti-correlation adjustment is adopted: the higher the proportion of inductive loads, the lower the first frequency reference value, to avoid misinterpreting normal power frequency fluctuations as fault signals.

[0105] The second frequency reference value = the basic reference value 2 × (1 + k × L); k is the adjustment coefficient, and L is the proportion of inductive load in the total load. High-frequency signals are less affected by eddy current losses in the inductive load, and the high-frequency components generated by the fault decay more slowly when propagating over short distances. Therefore, positive correlation adjustment is adopted: the higher the proportion of inductive load, the higher the second frequency reference value, which enhances sensitivity to close-range high-frequency fault signals.

[0106] Compare the first corresponding value with the second corresponding value. If the first corresponding value is greater than the second corresponding value, a long-distance disaster situation prompt is issued; otherwise, a short-distance disaster situation prompt is issued.

[0107] Long-distance disaster alert: When the first corresponding value is greater than the second corresponding value, it indicates that the power frequency component is dominant and the high-frequency component is relatively weak. Because power frequency signals propagate over long distances and attenuate slowly, while high-frequency signals attenuate rapidly with distance, this situation usually indicates that the fault point is far from the monitoring equipment (>10 meters). In this case, the system triggers a yellow alert and prompts maintenance personnel to expand the inspection range.

[0108] Close-range disaster alert: When the first corresponding value is less than or equal to the second corresponding value, the high-frequency component has significantly changed, exceeding or approaching the power frequency component. This indicates that the fault point is close to the monitoring equipment (≤10 meters), and the high-frequency electromagnetic radiation generated by partial discharge is strong. The system immediately triggers a red alert and accurately locates the fault area, providing critical information for emergency response.

[0109] This method deploys voltage, current, and magnetic field sensors at power nodes to collect multi-dimensional electromagnetic data in real time. Voltage data is used to capture intermittent fluctuations caused by changes in insulation state, while current data focuses on transient current fluctuations caused by contact faults. The system uses algorithms such as filtering and noise reduction and template matching to deeply process the collected data. It calculates intermittent matching values, which reflect the characteristics of voltage fluctuations, and transient matching values, which characterize the degree of current fluctuations. Through weighted fusion, it derives a comprehensive evaluation metric: the electrical data matching value.

[0110] When the electrical data match value exceeds a preset threshold, the system automatically triggers the second phase of analysis, performing spectral analysis on the low-frequency (e.g., 50Hz) and high-frequency (100-200Hz) components in the magnetic field data. By comparing the high- and low-frequency variation characteristics with dynamically adjusted reference values, the corresponding frequency characteristic ratio is calculated to determine the fault distance. If the low-frequency characteristic ratio is higher than the high-frequency characteristic ratio, the fault is determined to be far away; otherwise, the fault point is determined to be in the close range.

[0111] Compared to traditional infrared monitoring technology, which relies on passive monitoring of equipment surface temperature changes, this method leverages the advanced response characteristics of electromagnetic signals to detect changes in the electromagnetic signatures of early-stage faults, such as insulation degradation and poor contact, before localized overheating occurs, significantly accelerating disaster warning times. Furthermore, differentiated analysis of high- and low-frequency signals allows for precise location of fault distances, providing operations and maintenance personnel with more targeted guidance, effectively reducing the risk of power equipment failures and ensuring the safe and stable operation of the power system.

[0112] Reference Figure 4 In order to improve the accuracy of early warning, the method further includes the following steps:

[0113] The second voltage data, second current data, and second magnetic field data are acquired based on the second power node. The second power node is located within a set distance range of the first power node, such as a radius of 50 meters. It can also be a cable branching point or transformer point in the factory, and the first power node and the second power node do not belong to the same circuit network. The two nodes adopt the principle of non-homologous deployment to ensure that they belong to different circuit networks and avoid data homology interference caused by faults in the same circuit. For example, in a large industrial park, the first node can be set at the outlet of the main transformer, and the second node can be deployed at the distribution box of an independently powered workshop. The two nodes achieve synchronous data collection through a fiber optic network.

[0114] Within the most recently set time period, the system calculates the waveform similarity between the first voltage data and the second voltage data as the voltage similarity value; the waveform similarity between the first current data and the second current data as the current similarity value; and the waveform similarity between the first magnetic field data and the second magnetic field data as the magnetic field similarity value. For example, the system selects the most recent 10 minutes of monitoring data as the analysis window and uses the Dynamic Time Warping (DTW) algorithm to calculate data similarity between nodes.

[0115] Based on the voltage similarity value, current similarity value and magnetic field similarity value, a weighted calculation is performed to obtain a comprehensive similarity value; the comprehensive similarity value = weight coefficient 1 × voltage similarity value + weight coefficient 2 × current similarity value + weight coefficient 3 × magnetic field similarity value. The weight coefficient is determined through a large number of simulation verifications and is initially set to 1 / 3.

[0116] If the comprehensive similarity value is greater than the preset reference value, all disaster prompts will be blocked.

[0117] Compare the similarity of multiple-dimensional signals at the first power node and the second power node within a set distance range. If similar changes occur in the data of the two nodes at the same time, it is more likely to be caused by external environmental interference or overall system fluctuations rather than equipment failure. At this time, the disaster situation prompt is blocked to effectively avoid false alarms caused by environmental factors.

[0118] In order to further optimize the early warning method, the step of comparing the first corresponding value with the second corresponding value further includes the following steps:

[0119] The corresponding difference between the first corresponding value and the second corresponding value is calculated, and the corresponding difference reflects the relative intensity difference between the high-frequency and low-frequency features.

[0120] During the preset monitoring period, if the absolute value of the change in the corresponding difference is greater than the preset reference value for change, indicating that the disaster situation is in a rapidly changing stage, it indicates that the disaster situation is changing; if the corresponding difference changes from a positive number to a negative number, it means that the intensity of the high-frequency feature exceeds that of the low-frequency feature, indicating that the fault point is moving toward the monitoring node, indicating that the disaster is approaching. If the corresponding difference changes from a negative number to a positive number, it means that the low-frequency feature has regained its dominance and the scope of influence has expanded, indicating that the disaster situation is spreading. High-frequency signals attenuate quickly, have strong directionality, and have significant intensity when propagating over short distances, corresponding to localized or short-range faults. Low-frequency signals attenuate slowly, have strong penetration, and have superior intensity when propagating over long distances, reflecting large-scale or long-range faults.

[0121] Compared with the traditional fixed threshold warning method, this solution calculates the difference and change between the first corresponding value and the second corresponding value in real time. When the absolute value of the change exceeds the preset threshold, it can capture dynamic trends such as acceleration and deceleration of the disaster. At the same time, the direction of the disaster is judged based on the positive and negative changes of the difference. A change from positive to negative indicates that the disaster is approaching and the risk is increasing, while a change from negative to positive indicates that the scope of impact is expanding. In this way, the transition from static judgment to dynamic tracking is achieved, which greatly improves the foresight and effectiveness of emergency response of the power system.

[0122] A plurality of contact resistance sensors are arranged in a set area around the plurality of power modules, and positions of the contact resistance sensors in the area are randomly arranged; the method further comprises the following steps:

[0123] Contact resistance sensors are used to obtain multiple contact resistance data points. Within a monitoring area with a radius of 10-30 meters around power modules, such as substations and distribution rooms, sensor deployment locations are determined using a grid-based random sampling algorithm. Within each monitoring area, 8-12 contact resistance sensors are randomly deployed, forming an irregularly distributed monitoring network.

[0124] If a mutation term is identified from the contact resistance data, the periodic mutation feature is identified within the most recent preset mutation time period (15-30 minutes). First, the sensor collects contact resistance data in real time and uses a Kalman filter algorithm to eliminate noise generated by environmental electromagnetic interference and equipment vibration to ensure data accuracy. Method for identifying mutation terms: When the change in contact resistance value per unit time (such as 1 second) exceeds the set threshold, it is determined to be a mutation term, and each contact resistance sensor is identified. The method for identifying periodic mutation features is as follows: When the number of mutations of all contact resistance sensors within a unit period (such as 10 seconds) exceeds the set threshold (such as 3 times), and this condition is met for 3 consecutive periods, it is confirmed as a periodic mutation feature; the periodic mutation feature represents the frequency of mutations within the period.

[0125] If periodic mutation characteristics are identified, a disaster warning prompt for the contact device will be issued.

[0126] By monitoring changes in contact resistance in real time and comparing them against a set threshold, sudden changes in resistance can be quickly identified. Because contact resistance typically increases significantly before a fault or fire, this method can trigger an alarm at the earliest signs of overheating in the contact device, significantly improving warning times compared to traditional monitoring methods that rely on rising temperatures. After detecting a sudden change, further analysis is performed to determine if it exhibits periodic mutation characteristics. This effectively distinguishes occasional changes caused by fluctuations in normal equipment operation from persistent anomalies caused by poor contact.

[0127] In the power system safety protection system, insulation performance degradation is one of the core causes of equipment failure and fire hazards. To achieve early detection of insulation degradation hazards, multiple insulation resistance sensors are set in a set area around multiple power equipment. The insulation resistance sensors are randomly positioned in the area. The method also includes the following steps:

[0128] Insulation resistance sensors are used to obtain multiple insulation resistance data points. Under normal circumstances, the insulation resistance data has a large value ratio; before a power equipment failure or fire occurs, this value gradually decreases. A monitoring area with a radius of 5-20 meters is defined around power equipment clusters, such as substations and distribution cabinets. Using an existing random sampling algorithm, 6-10 insulation resistance sensors are randomly deployed within the area. This deployment method overcomes the limitations of traditional grid-like fixed-point monitoring. By simulating the random distribution characteristics of insulation weaknesses in power equipment, it ensures that the monitoring network can cover hidden insulation defects caused by equipment aging and environmental corrosion.

[0129] Calculate the average of multiple insulation resistance data points. If a gradually decreasing trend (i.e., a negative acceleration) is detected within the average, identify a persistent change within the most recent preset mutation time period. If a persistent change is detected, issue a warning indicating a power equipment disaster.

[0130] The method for identifying persistent change characteristics uses a sliding average method. At one-minute intervals, the sliding average of the insulation resistance is calculated for five consecutive minutes, resulting in the sequence S1, S2, S3, etc. If six consecutive sliding averages show a continuous downward trend within the preset 30-minute mutation period (i.e., S1>S2>S3>S4>S5>S6), a persistent change characteristic can be determined, and an early warning can be issued.

[0131] To improve the reliability and accuracy of leakage arc fire warnings, this monitoring method constructs a three-level progressive analysis method of "insulation resistance trend judgment - residual current detection - magnetic field frequency characteristic analysis", which includes the following steps:

[0132] If a persistent change characteristic is detected within the most recent preset mutation time period, indicating a decrease in line insulation resistance, indicating possible high-temperature aging, the first current data is analyzed. If a residual current is detected in the first current data and is greater than a preset residual reference value, the first magnetic field data is analyzed. A high-precision residual current transformer is used to collect line current data in real time, and the harmonic components are separated through Fourier transform. Characteristic harmonics such as the third and fifth harmonics are extracted to obtain the residual current value. If the residual current value is greater than the threshold, it indicates a possible leakage current, and the arc generated by the leakage current heats the insulation layer, thus initiating the magnetic field data detection process.

[0133] If a second frequency variation characteristic is identified from the first magnetic field data, a leakage arc fire warning is issued. For lines with excessive residual current, the system performs spectral analysis on the first magnetic field data. A magnetic field sensor with a bandwidth of 1MHz is used to collect data, focusing on analyzing the energy distribution in the 100-200kHz frequency band. When the power spectral density (PSD) in this frequency band exceeds three times the historical baseline value and persists for more than five seconds, the presence of a second frequency variation characteristic is determined, confirming the occurrence of a leakage arc. The system immediately issues a leakage arc fire warning (audio-visual alarm) and simultaneously pushes a detailed report containing the fault location, current / magnetic field waveforms, and risk level to the operation and maintenance terminal. Upon arriving at the scene, the operation and maintenance personnel discovered obvious signs of carbonization at the cable joints and partial melting of the underlying PVC trunking, thus preventing a fire accident in a timely manner.

[0134] The values ​​of the parameters such as the threshold in this embodiment are based on a large amount of experimental data and are obtained in combination with industry standards. The specific values ​​in the case are used as examples for reference to facilitate the explanation of the principles of the technical solution.

[0135] An embodiment of the present application further discloses a disaster monitoring system based on multi-dimensional electromagnetic data recognition, comprising a processor, wherein the processor executes the steps of any one of the above-described disaster monitoring methods based on multi-dimensional electromagnetic data recognition.

[0136] An embodiment of the present application further discloses a storage medium, in which a program is stored. When the program is executed by a processor, the steps of any one of the above-mentioned disaster monitoring methods based on multi-dimensional electromagnetic data recognition are implemented.

[0137] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A disaster monitoring method based on multi-dimensional electromagnetic data recognition, characterized in that: The steps include: Acquire first voltage data, first current data, and first magnetic field data based on the first power node; identifying intermittent fluctuation characteristics based on the first voltage data, and identifying transient fluctuation characteristics based on the first current data; Calculating an intermittent matching value based on the identified intermittent fluctuation characteristics, and calculating a transient matching value based on the identified transient fluctuation characteristics; Calculating an electrical data matching value based on the intermittent matching value and the transient matching value; If the electrical data matching value is greater than a preset electrical reference value, a first frequency change feature and a second frequency change feature are identified from the first magnetic field data, and a disaster warning prompt is issued; wherein a first frequency corresponding to the first frequency change feature is lower than a second frequency corresponding to the second frequency change feature; Calculating a first corresponding value based on the identified first frequency change feature and a preset first frequency reference value, and calculating a second corresponding value based on the identified second frequency change feature and a preset second frequency reference value; Compare the first corresponding value with the second corresponding value. If the first corresponding value is greater than the second corresponding value, a long-distance disaster prompt is issued; otherwise, a short-distance disaster prompt is issued.

2. The disaster monitoring method based on multi-dimensional electromagnetic data recognition according to claim 1 is characterized in that: The step of identifying intermittent fluctuation characteristics comprises: filtering the first voltage data to obtain temporary voltage data; dividing the temporary voltage data into a plurality of continuous voltage data according to a preset time window; Calculating the fluctuation amplitude and fluctuation duration of each of the continuous voltage data; If the fluctuation amplitude is within a preset amplitude range, and the fluctuation duration is within a preset duration range, then the voltage continuous data has the intermittent fluctuation feature, and the intermittent fluctuation feature includes the fluctuation amplitude and the fluctuation duration; The step of calculating the intermittent matching value comprises: Calculating a fluctuation similarity value based on the fluctuation amplitude and a preset amplitude template; Calculating a duration similarity value based on the fluctuation duration and a preset duration template; An interval matching value is calculated based on the fluctuation similarity value and the duration similarity value.

3. The disaster monitoring method based on multi-dimensional electromagnetic data recognition according to claim 1, characterized in that: The method further comprises the steps of: Acquiring second voltage data, second current data, and second magnetic field data based on a second power node; wherein the second power node is located within a set distance range of the first power node; In a recently set time period, the waveform similarity between the first voltage data and the second voltage data is calculated as a voltage similarity value; the waveform similarity between the first current data and the second current data is calculated as a current similarity value; and the waveform similarity between the first magnetic field data and the second magnetic field data is calculated as a magnetic field similarity value. A comprehensive similarity value is obtained by weighted calculation based on the voltage similarity value, the current similarity value and the magnetic field similarity value; If the comprehensive similarity value is greater than a preset reference value, all disaster situation prompts are shielded.

4. The disaster monitoring method based on multi-dimensional electromagnetic data recognition according to claim 1, characterized in that: The step of identifying transient fluctuation characteristics comprises: filtering the first current data to obtain temporary current data; Extracting a plurality of current peak data from the temporary current data according to a preset peak template; Calculating a peak height and a peak width of the current peak data, and calculating a peak steepness value according to the peak height and the peak width; If the peak steepness value is within the preset peak range, the current peak data has the transient fluctuation characteristics, and the transient fluctuation characteristics include the peak height, the peak width, and the peak steepness value; The step of calculating the transient matching value comprises: Calculating a height similarity value based on the peak height and a preset height template; Calculating a width similarity value based on the peak width and a preset width template; Calculating a temporary matching value based on the height similarity value and the width similarity value; The temporary matching value is adjusted according to the positive correlation of the peak steepness value, and the adjusted temporary matching value is a transient matching value.

5. The disaster monitoring method based on multi-dimensional electromagnetic data recognition according to claim 1, characterized in that: The step of comparing the first corresponding value with the second corresponding value further includes the following steps: Calculating a corresponding difference between the first corresponding value and the second corresponding value; During the preset monitoring time period, if the absolute value of the change in the corresponding difference is greater than the preset change reference value, it indicates that the disaster situation is changing; if the corresponding difference changes from a positive number to a negative number, it indicates that the disaster is approaching; if the corresponding difference changes from a negative number to a positive number, it indicates that the disaster is spreading.

6. The disaster monitoring method based on multi-dimensional electromagnetic data recognition according to claim 1, characterized in that: A plurality of contact resistance sensors are arranged in a set area around the plurality of power modules, and the contact resistance sensors are randomly arranged in the area; the method further comprises the following steps: Acquiring contact resistance data based on the contact resistance sensor; If a mutation term is identified from the contact resistance data, a periodic mutation feature is identified within a recent preset mutation time period; If the periodic mutation feature is identified, a disaster warning prompt for the contact device will be issued.

7. The disaster monitoring method based on multi-dimensional electromagnetic data recognition according to claim 6, characterized in that: A plurality of insulation resistance sensors are arranged in a set area around a plurality of electrical devices, wherein the insulation resistance sensors are randomly arranged in the area; the method further comprises the following steps: Acquiring insulation resistance data based on the insulation resistance sensor; Calculating an average value of the plurality of insulation resistance data; if a gradually decreasing trend of change is identified from the average value, identifying a continuous change feature within a recent preset mutation time period; If the above-mentioned continuous change characteristics are identified, an early warning of power equipment disaster situation will be issued.

8. The disaster monitoring method based on multi-dimensional electromagnetic data recognition according to claim 7, characterized in that: The method further comprises the steps of: If a persistent change feature is identified within the most recent preset mutation time period, analyzing the first current data; If a residual current is identified from the first current data and the residual current is greater than a preset residual reference value, analyzing the first magnetic field data; If a second frequency change feature is identified from the first magnetic field data, a leakage arc fire warning is issued.

9. A disaster monitoring system based on multi-dimensional electromagnetic data recognition, characterized in that: The method comprises a processor, wherein the processor executes the steps of the disaster monitoring method based on multi-dimensional electromagnetic data recognition as described in any one of claims 1 to 8.

10. A storage medium, characterized in that: The medium stores a program, and when the program is executed by the processor, the steps of the disaster monitoring method based on multi-dimensional electromagnetic data recognition described in any one of claims 1 to 8 are implemented.

Images