A switch cabinet partial discharge on-line intelligent detection system

Abstract

The present application relates to the technical field of switch cabinet partial discharge detection, and particularly relates to a switch cabinet partial discharge online intelligent detection system; the system comprises a preliminary screening module, an environment monitoring module, a first diagnosis module, a second diagnosis module and a feedback adjustment module. According to the present application, the diagnosis path is adaptively selected according to the real-time environmental noise level; in a low-noise environment, ultrasonic waves and very high frequency signals are synchronously fused, and through double consistency analysis of intensity distribution and phase spectrum, accurate judgment and positioning of the discharge type are realized; in a high-noise environment, the very high frequency detection with strong anti-interference ability is mainly used, combined with the synchronicity verification of ultrasonic wave signals and the concentration trend analysis of SF6 decomposition products, and the final quantitative confirmation and accurate positioning are carried out by using the pulse current method; the noise judgment threshold is dynamically optimized, the continuous self-optimization of the detection strategy is realized; the insulation abnormalities are accurately warned through online real-time monitoring, and the safety and reliability of the operation of the key electrical equipment are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of switch cabinet partial discharge detection, and particularly relates to a switch cabinet partial discharge online intelligent detection system. BACKGROUND

[0002] Insulation deterioration is one of the main threats to the safe operation of high-voltage switch cabinets. In the whole life cycle of the equipment, design defects (such as uneven electric field distribution), manufacturing process flaws (such as internal air gap of epoxy resin insulation), installation deviation (such as poor conductor contact) and other hidden dangers may lurk in the initial stage; during operation, factors such as damp condensation, dirt accumulation, mechanical vibration and long-term electric and thermal stress further aggravate the aging of the insulation material and induce partial discharge; as a typical representation of insulation defects, the initial energy of partial discharge is weak and highly concealed, but continuous discharge will cause irreversible damage such as carbonization of insulation medium and growth of electric tree branches, and eventually lead to insulation breakdown and even equipment explosion, which seriously threatens the safety of the power grid; at present, the commonly used effective means for on-line detection of partial discharge of switch cabinet equipment mainly include transient earth voltage method, ultrasonic method and ultra-high frequency method; and the single detection method has obvious limitations and is difficult to diagnose comprehensively and accurately; the ultrasonic method belongs to the category of mechanical vibration waves, although it is sensitive to surface discharge, it is easily disturbed by environmental noise (such as fans and vibration), has low signal-to-noise ratio in noisy field, and has weak penetration and detection ability for deep or internal discharge; the ultra-high frequency (UHF) method has strong anti-electric interference ability, but its signal is easily shielded by the metal shell of the switch cabinet, relies on the leakage of the gap, which leads to serious signal attenuation and difficult positioning, and it is difficult to distinguish the discharge type alone; the transient earth voltage (TEV) method is simple to operate and suitable for rapid screening, but it is easily disturbed by the complex electromagnetic environment on site, the detection result is greatly affected by the grounding condition and the cabinet structure, has high false positive rate and cannot be accurately positioned. The existing technologies independently or simply combine the above methods, lack of deep fusion and collaborative verification mechanism, which leads to insufficient diagnosis ability for complex discharge modes (such as the existence of internal discharge and surface discharge at the same time), and it is difficult to draw reliable conclusions from a single or a small number of signals, therefore, there is an urgent need for a switch cabinet partial discharge online intelligent detection system which can adapt to environmental changes, deeply fuse multi-modal information, intelligently make decisions and continuously optimize itself.

[0003] The partial discharge detection method of the switch cabinet and the high-voltage switch cabinet are disclosed in Chinese Patent Publication No. CN118444113A. The partial discharge detection method of the switch cabinet sets at least one group of ultrasonic sensors, voltage sensors, and coupling antennas outside the cabinet body. The ultrasonic sensors detect ultrasonic signals, the voltage sensors detect transient ground voltage signals, and the coupling antennas detect ultrahigh frequency electromagnetic wave signals. When the signal amplitude does not exceed the corresponding threshold, the state value of the switch cabinet is predicted by combining the first threshold with the first state parameter, the second threshold with the second state parameter, and the third threshold with the third amplitude. The state value is used to determine whether the switch cabinet has a partial discharge and issue a warning. As can be seen, in the existing online intelligent detection technology for partial discharge of switch cabinets, the optimal detection strategy cannot be dynamically adjusted according to the real-time changes in the field environment (such as background noise level, temperature and humidity), and there is a lack of deep feature extraction, correlation analysis and fusion decision-making capability for multi-source and heterogeneous signals (sound, electricity, and chemistry). The accuracy and reliability of the detection are low. SUMMARY

[0004] To this end, the present application provides a switch cabinet partial discharge online intelligent detection system to overcome the lack of a hierarchical progressive intelligent detection system from wide-area rapid screening to multi-dimensional accurate diagnosis to final quantitative confirmation in the prior art, resulting in the problem of low detection reliability.

[0005] To achieve the above-mentioned purpose, the present application provides a switch cabinet partial discharge online intelligent detection system, comprising:

[0006] A preliminary screening module is used to scan each switch cabinet using a TEV detector to obtain corresponding real-time TEV detection values, calculate the real-time TEV value increase and the average TEV detection value according to the real-time TEV detection values, and determine the suspected abnormal cabinets and abnormal cabinets in each switch cabinet according to the real-time TEV value increase and the average TEV detection value;

[0007] An environmental monitoring module is connected to the preliminary screening module to monitor the environmental background noise level and determine whether the current environment is at a first noise level or a second noise level;

[0008] A first diagnosis module is connected to the preliminary screening module and the environmental monitoring module, respectively, and in response to the current environment being at a first noise level, simultaneously performs ultrasonic wave detection and ultra-high frequency detection on the abnormal cabinet, and outputs an abnormal diagnosis result according to the wave signal characteristics and atlas analysis;

[0009] A second diagnosis module is connected to the preliminary screening module and the environmental monitoring module, respectively, and in response to the current environment being at a second noise level, performs ultra-high frequency detection on the switch cabinet to be detected, and outputs an abnormal diagnosis result combining the UHF signal and the SF6 decomposition characteristics;

[0010] a feedback adjustment module connected with the environment monitoring module, the first diagnosis module and the second diagnosis module, configured to dynamically adjust a preset noise threshold in the environment monitoring module for distinguishing the first noise level from the second noise level according to actual verification results of historical diagnosis events.

[0011] Further, the preliminary screening module comprises:

[0012] a baseline establishment unit configured to establish dynamic baseline values and normal fluctuation ranges of each switch cabinet in different time periods according to historical TEV detection data;

[0013] a real-time calculation unit configured to compare the real-time TEV detection value with the dynamic baseline value of the corresponding time period to calculate the real-time TEV value increase amount;

[0014] a lateral comparison unit configured to calculate a relative difference between the real-time TEV detection value of each switch cabinet and the average TEV detection value of the adjacent switch cabinet at the same detection position;

[0015] a threshold determination unit configured to determine a first preset threshold based on the dynamic baseline value, the signal attenuation coefficient and the environmental parameter;

[0016] a comprehensive determination unit configured to determine that the corresponding switch cabinet is an abnormal cabinet when the real-time TEV value increase amount is greater than the first preset threshold, determine that the corresponding switch cabinet is an abnormal cabinet when the real-time TEV value increase amount is less than or equal to the first preset threshold and the relative difference exceeds a second preset threshold, and determine that the corresponding switch cabinet is a suspected abnormal cabinet otherwise.

[0017] Further, the threshold determination unit comprises:

[0018] a threshold calculation sub-unit configured to determine the first preset threshold based on a statistical standard deviation multiple of the dynamic baseline value;

[0019] a threshold correction sub-unit configured to adaptively correct the first preset threshold according to the signal attenuation coefficient and the environmental parameter of the detection point position.

[0020] Further, the environment monitoring module comprises:

[0021] an environment determination unit configured to determine a current environmental background noise level; wherein if the background noise is less than a preset noise threshold, it is determined that the current environment is at the first noise level, and if the background noise is greater than or equal to the preset noise threshold, it is determined that the current environment is at the second noise level.

[0022] Further, the feedback adjustment module is configured to:

[0023] record the diagnostic events and their final confirmation results completed under the first and second noise levels, and calculate the first and second misjudgment proportions;

[0024] if the first misjudgment proportion is greater than the first proportion threshold, the preset noise threshold is increased;

[0025] if the second misjudgment proportion exceeds the second proportion threshold, the preset noise threshold is decreased;

[0026] wherein, the first misjudgment proportion is the proportion of events finally confirmed as interference or false positives in the diagnoses completed under the first noise level; the second misjudgment proportion is the proportion of events finally confirmed as real discharge defects but initially diagnosed as interference or events to be observed in the diagnoses completed under the second noise level.

[0027] Further, the first diagnostic module comprises:

[0028] a joint detection unit configured to synchronously collect ultrasonic signals and UHF signals from the abnormal cabinet;

[0029] a feature extraction unit configured to extract first phase spectrum and acoustic signal intensity distribution features from the ultrasonic signals, and extract second phase spectrum and electromagnetic signal intensity distribution features from the UHF signals;

[0030] a fusion analysis unit configured to compare and analyze the first phase spectrum and the second phase spectrum, and determine the discharge type and locate the abnormal area by combining the intensity distribution consistency of the acoustic signal and the electromagnetic signal.

[0031] Further, the fusion analysis unit is specifically configured to:

[0032] if the acoustic signal intensity of the abnormal area located by the ultrasonic signal presents a gradient distribution, and both the first phase spectrum and the second phase spectrum present symmetrical double-peak features related to the power frequency, the discharge type is determined to be an insulation defect type discharge;

[0033] if the second phase spectrum presents a single-peak feature appearing only in the negative half cycle of the power frequency, the discharge type is determined to be a corona discharge.

[0034] Further, the second diagnostic module comprises:

[0035] a UHF dominant detection unit configured to perform UHF detection on the switch cabinet to be detected under the second noise level, and obtain the amplitude and phase spectrum of the UHF signal;

[0036] a verification analysis unit configured to determine whether the amplitude of the UHF signal continuously exceeds a third preset threshold and the phase spectrum has regularity;

[0037] If the judgment is yes, further acquire the ultrasonic signal feature and SF6 decomposition feature to determine whether the ultrasonic signal and the UHF signal are synchronous and have the same phase feature, and output the diagnosis result according to the SF6 decomposition concentration and the pulse current method.

[0038] Further, the second diagnosis module further comprises:

[0039] A result generation unit is configured to output a diagnosis result of surface or near-surface discharge in response to that the ultrasonic signal and the UHF signal are synchronous and have the same phase feature, and the SF6 decomposition concentration is less than a preset concentration threshold; and output a diagnosis result of high-energy discharge causing insulation chemical decomposition in response to that the SF6 decomposition concentration exceeds the preset concentration threshold and continuously rises.

[0040] Further, the second diagnosis module further comprises a depth confirmation unit configured to:

[0041] Trigger the pulse current method to perform confirmation when the UHF signal is strong but the ultrasonic signal is not synchronous and the SF6 decomposition concentration is normal;

[0042] The depth confirmation unit controls the installation of a differential Rogowski coil, applies an alternating test voltage, collects a pulse current signal, and analyzes the PRPD spectrum feature and discharge amount of the pulse current signal to diagnose and locate suspected internal air gap discharge or deep fault, and output a diagnosis result.

[0043] Compared with the prior art, the beneficial effects of the present application are that, firstly, wide-area rapid screening is carried out through transient voltage (TEV) detection, abnormal cabinets and suspected abnormal cabinets are identified by dynamic baseline analysis and transverse comparison; then, the diagnostic path is adaptively selected according to the real-time environmental noise level: in a low-noise environment, the first diagnostic module is started, ultrasonic waves and very high frequency signals are synchronously fused, and the discharge type is accurately judged and positioned through double consistency analysis of intensity distribution and phase spectrum; in a high-noise environment, the second diagnostic module is started, the very high frequency detection with strong anti-interference ability is dominated, and the reliable diagnosis is realized by combining with the ultrasonic signal synchronicity verification and the SF6 decomposition concentration trend analysis, and the final quantitative confirmation and accurate positioning are further carried out on difficult cases by using the pulse current method; the system is also provided with a feedback adjustment module, which dynamically optimizes the noise judgment threshold according to the verification situation of the historical diagnosis results, and realizes the continuous self-optimization of the detection strategy; the all-weather online intelligent monitoring and accurate early warning of partial discharge are realized, the accuracy and reliability of detection are improved through multi-technology fusion and layered decision-making, and the false alarm and missed alarm risks are effectively reduced; the environment-adaptive diagnostic path selection and closed-loop feedback optimization mechanism enhance the robustness of the system under different working conditions; finally, the system promotes the equipment operation and maintenance mode from passive response to fault to life cycle active defense, provides key technical support for guaranteeing the safe and stable operation of the power grid, reducing the operation and maintenance cost, and online real-time monitoring and accurate early warning of insulation abnormalities improve the safety and reliability of the operation of key electrical equipment. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 FIG. 1 is a structural schematic diagram of a switch cabinet partial discharge online intelligent detection system according to an embodiment of the present application;

[0045] Figure 2 FIG. 2 is a structural schematic diagram of a preliminary screening module according to an embodiment of the present application;

[0046] Figure 3 FIG. 3 is a structural schematic diagram of a first diagnostic module according to an embodiment of the present application;

[0047] Figure 4 FIG. 4 is a structural schematic diagram of a second diagnostic module according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to make the purpose and advantages of the present application clearer and more apparent, the present application will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present application, and do not limit the protection scope of the present application.

[0049] The preferred embodiments of the present application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application, and are not intended to limit the protection scope of the present application.

[0050] It should be noted that in the description of the present application, the terms indicating the direction or positional relationship of "upper", "lower", "left", "right", "inner", "outer" and the like are based on the direction or positional relationship shown in the drawings, which is only for the convenience of description, and does not indicate or imply that the device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0051] In addition, it should be noted that in the description of the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0052] Please refer to Figure 1 As shown in the structure schematic diagram of the switch cabinet partial discharge online intelligent detection system of the embodiment of the present application, the present application provides a switch cabinet partial discharge online intelligent detection system, comprising:

[0053] The preliminary screening module is used to scan each switch cabinet by using a TEV detector to obtain corresponding real-time TEV detection values, to calculate real-time TEV value increments and average TEV detection values according to the real-time TEV detection values, and to determine suspected abnormal cabinets and abnormal cabinets in each switch cabinet according to the real-time TEV value increments and the average TEV detection values;

[0054] The environmental monitoring module is connected with the preliminary screening module, and is used to monitor the environmental background noise level and determine whether the current environment is at a first noise level or a second noise level, and to determine the suspected abnormal cabinet as a to-be-detected switch cabinet or an abnormal cabinet according to the current environment;

[0055] The first diagnosis module is connected with the preliminary screening module and the environmental monitoring module respectively, and in response to the current environment being at the first noise level, the abnormal cabinet is synchronously subjected to ultrasonic detection and ultra-high frequency detection, and an abnormal diagnosis result is output according to wave signal characteristics and atlas analysis;

[0056] The second diagnosis module is connected with the preliminary screening module and the environmental monitoring module respectively, and in response to the current environment being at the second noise level, the to-be-detected switch cabinet is subjected to ultra-high frequency detection, and an abnormal diagnosis result is output in combination with UHF signals and SF6 decomposition characteristics;

[0057] A feedback adjustment module connected with the environment monitoring module, the first diagnosis module and the second diagnosis module, for dynamically adjusting a preset noise threshold in the environment monitoring module for distinguishing the first noise level from the second noise level according to actual verification results of historical diagnosis events.

[0058] In the embodiment, first, wide-area rapid screening is performed through transient electric voltage (TEV) detection, abnormal cabinets and suspected abnormal cabinets are identified by dynamic baseline analysis and horizontal comparison; then, a diagnosis path is adaptively selected according to a real-time environmental noise level: in a low-noise environment, the first diagnosis module is started, ultrasonic waves and very high frequency signals are synchronously fused, and the discharge type is accurately judged and positioned through dual-consistency analysis of intensity distribution and phase spectrum; in a high-noise environment, the second diagnosis module is started, very high frequency detection with strong anti-interference capability is mainly used, and reliable diagnosis is realized by combining with ultrasonic wave signal synchronicity verification and SF6 decomposition product concentration trend analysis; the system is further provided with a feedback adjustment module, which dynamically optimizes a noise determination threshold according to verification of historical diagnosis results, and realizes continuous self-optimization of the detection strategy; all-weather online intelligent monitoring and accurate early warning of partial discharge are realized, the accuracy and reliability of detection are improved through multi-technology fusion and layered decision-making, and the false alarm and missed alarm risks are effectively reduced; the environment-adaptive diagnosis path selection and closed-loop feedback optimization mechanism enhance the robustness of the system under different working conditions; finally, the system promotes the equipment operation and maintenance mode to change from passive response to fault to active defense in the service life, provides key technical support for guaranteeing the safe and stable operation of the power grid, reducing operation and maintenance costs, and online real-time monitoring and accurate early warning of insulation abnormalities improve the safety and reliability of the operation of key electrical equipment.

[0059] Referring to the Figure 2 As shown in the figure, it is a structure schematic diagram of the preliminary screening module of the embodiment of the application.

[0060] Specifically, the preliminary screening module comprises:

[0061] A baseline establishment unit for establishing dynamic baseline values and normal fluctuation ranges of each switch cabinet in different time periods according to historical TEV detection data;

[0062] A real-time calculation unit for comparing the real-time TEV detection value with the dynamic baseline value of the corresponding time period, and calculating the real-time TEV value increase amount;

[0063] A horizontal comparison unit for calculating the relative difference between the real-time TEV detection value of each switch cabinet and the average TEV detection value of the adjacent switch cabinet at the same detection position;

[0064] a threshold determination unit configured to determine a first preset threshold based on the dynamic baseline value, the signal attenuation coefficient, and the environmental parameter;

[0065] a comprehensive determination unit configured to determine that the corresponding switch cabinet is an abnormal cabinet when the increase of the real-time TEV value is greater than the first preset threshold, or when the increase of the real-time TEV value is less than or equal to the first preset threshold and the relative difference exceeds the second preset threshold; otherwise, the corresponding switch cabinet is determined to be a suspected abnormal cabinet.

[0066] In this embodiment, TEV sensors are fixedly installed at key positions (front cabinet door, rear cabinet door, side plate) of each switch cabinet, a standard grid layout is adopted, each sensor is connected to a local acquisition unit, supporting 4-20mA or digital signal output, and one gateway is deployed for each row of switch cabinets, responsible for data aggregation, preprocessing and preliminary analysis; when new switch cabinet equipment is put into operation, the system will record the TEV value of each detection point every minute for 7 consecutive days, and each hour is a period, 24 hours a day is divided into 24 periods, the average value and normal fluctuation range of TEV value in each period are calculated, for example, for switch cabinet A, rear lower cabinet position, TEV value from 10:00 to 10:59 am every day: day 1: 22dB, day 2: 23dB,..., day 7: 21dB, then calculate the average value: 22.5dB, then calculate the fluctuation range: ±3dB, then 19.5-25.5dB is the normal range, the baseline value and normal fluctuation range are automatically updated every day, if 90% of today's data is within the normal range, the baseline is slightly adjusted with today's data, i.e. new baseline = 80% old baseline + 20% today's data, if today's data is abnormal, the baseline is not updated, but marked as "abnormal day"; if the historical data is insufficient, the data of similar equipment or simulation data is used as the initial baseline; by obtaining the real-time TEV detection value and comparing it with the corresponding baseline, if the increase of real-time TEV value is greater than the first preset threshold, the corresponding switch cabinet is determined to be an abnormal cabinet, if the increase of real-time TEV value is less than or equal to the first preset threshold, the corresponding switch cabinet is determined to be a normal cabinet; for example, it is now 10:30 am, the TEV value is detected to be 35dB, the baseline value of 10:00 am is found to be 22.5dB, the normal fluctuation range is ±3dB, i.e. 19.5-25.5dB, then the increase of real-time TEV value is calculated = current value - baseline average value = 35dB - 22.5dB = 12.5dB, the standardized increase is = (current value - baseline average value) ÷ normal fluctuation standard deviation of this period, if the normal fluctuation standard deviation of 10:00 am is 2dB, the standardized increase is = 12.5dB ÷ 2dB = 6.25 times, indicating that the current value is 6.25 times higher than the normal value, considering the cumulative effect, whether it is interference is determined, i.e. if it is only a momentary high value, it may be interference, then the cumulative increase of the past 4 hours is calculated: cumulative increase = the sum of all increases in the past 4 hours, for example, 24 data points are measured every 10 minutes in the past 4 hours, each is higher than the baseline, then the cumulative increase is large, it is determined that there is no interference, and the corresponding switch cabinet is determined to be an abnormal cabinet, otherwise, if there is interference, the relative difference of the average TEV detection value of the adjacent switch cabinet is obtained for further analysis.

[0067] The second preset threshold in the embodiment is based on a large amount of normal historical data statistics. Through field measurement statistics, in the long-term test of multiple substations, the adjacent switch cabinets in the same batch and under the same operating state have a difference of no more than 6 dB in the background value of the TEV at the same detection point. In order to leave a certain safety margin to distinguish the real abnormality, the threshold is floated to 8-10 dB. When the partial discharge signal in the switch cabinet propagates to the external sensor through the cabinet body, there is usually a 6-15 dB attenuation, and a difference of more than 8 dB may mean that the signal source is not a common mode interference, but a partial discharge source in the cabinet. The second preset threshold can be configured in the range of 6-12 dB, which can be optimized according to the cabinet structure, sensor arrangement density and on-site electromagnetic environment. The system initialization stage can automatically set the initial value based on one week of learning data, and fine-tune in operation. The adjacent switch cabinets are the switch cabinets that are physically adjacent to the target cabinet (such as adjacent cabinets in the same row), have similar electrical circuits (such as the same section of busbar power supply), have the same equipment model and normal current operating state. Usually, 1-2 cabinets that meet the conditions are selected on the left and right of the target cabinet to form a “adjacent reference cabinet set”. For example, for the target cabinet 925, the set can be: {924 cabinet, 926 cabinet}. The process of calculating the relative difference value is as follows: the TEV detection values of the target cabinet and each adjacent reference cabinet at the same time or in a very short time at the same detection position (such as the front lower cabinet and the rear middle cabinet) are obtained. The arithmetic mean value of the TEV values of the adjacent reference cabinet set at the corresponding detection point is calculated as the adjacent cabinet average value at that position. The difference between the target cabinet detection value and the adjacent cabinet average value is calculated as the real-time difference. The relative difference value is calculated using the logarithmic form (dB) as follows: relative difference value = 20*log10(real-time difference); if the relative difference value is greater than the second preset threshold (such as 10 dB), it is considered that the target cabinet is significantly abnormal relative to its surrounding environment, and is determined as an abnormal cabinet; if the relative difference value is less than or equal to the second preset threshold, it is considered that the target cabinet and the adjacent cabinet have the same state, and the abnormality is low, and is determined as a normal cabinet. 10 (real-time difference); if the relative difference value is greater than the second preset threshold (such as 10 dB), it is considered that the target cabinet is significantly abnormal relative to its surrounding environment, and is determined as an abnormal cabinet; if the relative difference value is less than or equal to the second preset threshold, it is considered that the target cabinet and the adjacent cabinet have the same state, and the abnormality is low, and is determined as a normal cabinet.

[0068] Specifically, the threshold determination unit comprises:

[0069] a threshold calculation sub-unit configured to determine the first preset threshold based on a statistical standard deviation multiple of the dynamic baseline value;

[0070] a threshold correction sub-unit configured to adaptively correct the first preset threshold according to a signal attenuation coefficient and an environmental parameter of a detection point position.

[0071] In this embodiment, the first preset threshold is the threshold corresponding to the increase of the TEV value, that is, the standardized increase. Under the assumption of normal distribution, the probability of the TEV value fluctuating within 3 times of the standard deviation around the baseline is 99.7%, and the events exceeding this range are considered as statistical outliers. In industrial monitoring and fault diagnosis, the 3σ principle is widely used for initial screening of outliers, and is also recommended in ISO 13379 "Mechanical Equipment Condition Monitoring and Diagnosis". Therefore, the first preset threshold = 3 x standard deviation. The average value at 10:00 am is 22.5 dB, and the standard deviation is 2 dB. Therefore, the normal range of the TEV value is 22.5 ± (3 x 2) = 16.5-28.5 dB. In actual application, the multiple can be adaptively corrected according to the false alarm statistics of the feedback adjustment module. For example, when false alarms occur continuously, it can be appropriately increased to 3.5 times, and when the number of missed alarms increases, it can be reduced to 2.8 times. The setting of the first preset threshold can also consider the difference between different time periods, that is, the threshold is different at different times. For example, at 2:00 am, the load is low and the background is quiet, so the first preset threshold can be set to 2.5 times the standard deviation. At 10:00 am, the load is high and the interference is large, so the first preset threshold can be set to 3.5 times the standard deviation. Since the signal attenuation of each detection point on the switch cabinet is different, the first preset threshold corresponding to each detection point needs to be corrected. That is, the signal attenuation coefficient corresponding to the detection point is calculated according to the position of the detection point. The signal attenuation coefficient = 10 x log 10 (reference point signal intensity / detection point signal intensity). The process of calculating the corresponding signal attenuation coefficient is as follows. The position where the signal is most easily transmitted is selected as the reference point, and the signal attenuation relative to the reference point is measured at other positions. The position where the signal is most easily transmitted is near the observation window. For example, the reference point is the signal intensity of the observation window, which is 100%. The signal intensity of the detection point located in the lower corner of the rear cabinet is only 25% of the reference point. Therefore, the signal attenuation coefficient = 10 x log 10 (100 / 25) = 6 dB. Therefore, the corrected first preset threshold = first preset threshold x (1 + attenuation coefficient / 20). For example, the first preset threshold is 15 dB, and the attenuation coefficient is 6 dB. Therefore, the corrected first preset threshold = 15 x (1 + 6 / 20) = 15 x 1.3 = 19.5 dB. The threshold is relaxed for the position with large signal attenuation to avoid missed alarms. At the same time, the attenuation coefficient can be adaptively corrected according to environmental parameters, including temperature, humidity and background noise. The threshold increases by 5% for every 10% increase in humidity, the threshold changes by 3% for every 10°C change in temperature, and the threshold increases by 8% for every 10 dB increase in noise. For example, switch cabinet A, detection point located in the lower corner of the rear cabinet, detected at 10:00 am. The specific calculation process of the actual TEV detection value threshold corresponding to the detection point is as follows:

[0072] Step 1: Obtain basic data;

[0073] Average value 22.5dB, standard deviation 2dB, first preset threshold = 3x2dB = 6dB (first preset threshold)

[0074] Step 2: Attenuation correction: the position signal attenuation coefficient is 6dB, the first preset threshold after attenuation correction = 6x(1+6 / 20) = 7.8dB;

[0075] Step 3: Environmental correction;

[0076] Current humidity 70% (20% higher than normal), humidity correction = 20% / 10%x5% = 10%

[0077] Current temperature 30℃ (5℃ higher than normal), temperature correction: 5℃ / 10℃x3% = 1.5%

[0078] Background noise 5dB higher than normal, noise correction = 5dB / 10dBx8% = 4%

[0079] Then the total environmental correction coefficient = 1+10%+1.5%+4% = 1.155

[0080] Step 4: Calculate the final threshold;

[0081] Final threshold = 7.8dBx1.155 = 9.0dB (final first preset threshold)

[0082] Step 5: Convert to actual TEV detection value threshold;

[0083] Actual TEV detection value threshold = baseline average value + final first preset threshold = 22.5dB + 9.0dB = 31.5dB.

[0084] Specifically, the environmental monitoring module determines the current environmental background noise level by an independent noise sensor or analyzing the background noise spectrum in the TEV detection signal; wherein the first noise level is a low-noise environment where the background noise is lower than a preset noise threshold, and the second noise level is a high-noise environment where the background noise is higher than or equal to the preset noise threshold.

[0085] Specifically, the feedback adjustment module is configured to:

[0086] Record the diagnostic events completed at the first noise level and the second noise level and their final confirmation results;

[0087] If the proportion of final confirmation as interference or false alarm in the diagnosis completed at the first noise level exceeds a first proportion threshold, it is determined that the current preset noise threshold is too low, and the preset noise threshold is adjusted higher;

[0088] If in the diagnosis completed at the second noise level, there are events that are finally confirmed as real discharge defects but the initial diagnosis conclusion is interference or observation, and the proportion exceeds the second proportion threshold, it is determined that the current preset noise threshold is too high, and the preset noise threshold is adjusted lower.

[0089] The feedback adjustment module further comprises an adaptive adjustment unit, configured to:

[0090] According to the diagnosis results output by the first diagnosis module and the second diagnosis module and the subsequent verification, the noise threshold used to distinguish the first noise level and the second noise level is feedback optimized to improve the accuracy of the detection path selection.

[0091] In the embodiment, the initial preset noise threshold is set to 40 dB, the first proportion threshold is 20%, and the second proportion threshold is 10%. By obtaining the historical records, in one month, the system records 100 diagnosis events, 60 of which are determined to be the first noise level and processed by the first diagnosis module. Among the 60 events, 15 have the initial diagnosis conclusion of insulation defect discharge, but subsequent manual inspection or long-term tracking confirms that there is no defect, that is, false positives. In addition, 40 of them are determined to be the second noise level and processed by the second diagnosis module. Among the 40 events, 6 have the initial conclusion of "interference, continuous observation", but subsequent device failure or power outage maintenance confirms the discharge defect, that is, a false negative. The first misjudgment proportion, that is, the false positive rate under the first noise level = 15 / 60 = 25%, and the second misjudgment proportion, that is, the false negative rate under the second noise level = 6 / 40 = 15%. This situation shows that the current threshold is set near an awkward critical point, resulting in both types of errors (false positives and false negatives) being serious. At this time, because the false negative may lead to device failure not being handled in time, the harm is greater than the false positive, so the threshold is adjusted to reduce the false negative rate, that is, the threshold is adjusted from 40 dB to 38 dB, so that more scenarios are classified into the first noise level and a more refined diagnosis path is used to improve the detection rate of real defects.

[0092] Referring to Figure 3 As shown in the figure, it is a structural schematic diagram of the first diagnosis module of the embodiment of the application.

[0093] Specifically, the first diagnosis module comprises:

[0094] The joint detection unit is configured to synchronously collect ultrasonic signals and very high frequency signals from the abnormal cabinet.

[0095] The feature extraction unit is configured to extract a first phase spectrum and a sound signal intensity distribution feature from the ultrasonic signals, and extract a second phase spectrum and an electromagnetic signal intensity distribution feature from the very high frequency signals.

[0096] A fusion analysis unit is configured to compare and analyze the first phase pattern and the second phase pattern, and determine the discharge type and locate the abnormal area according to the consistency of the intensity distribution of the acoustic signal and the electromagnetic signal.

[0097] The ultrasonic sensor in the embodiment adopts a contact or non-contact ultrasonic sensor with a frequency response of 20 kHz-100 kHz, and is equipped with a preamplifier. The UHF sensor adopts an ultra-wideband antenna sensor with a frequency band of 300 MHz-3 GHz, and is arranged at a gap or a special detection interface of the switch cabinet. The two sensors are strictly synchronized in data acquisition through a hardware trigger channel of the same acquisition device or a high-precision (nanosecond level) time synchronization module based on GPS / Beidou, so as to ensure the time alignment in subsequent analysis. In the process of synchronously collecting the ultrasonic signal and the UHF signal, the ultrasonic probe and the UHF sensor are simultaneously attached to the surface to be detected of the target switch cabinet, and original waveform data of at least 10 power frequency periods (0.2 seconds) are synchronously recorded at a sampling rate of not less than 10 MS / s. The feature extraction unit receives the original waveform data collected by the joint detection unit, and performs the following processing: ultrasonic signal processing and UHF signal processing. The process of the ultrasonic signal processing includes acoustic signal intensity distribution feature extraction and first phase pattern extraction. The process of the acoustic signal intensity distribution feature extraction is as follows. The original ultrasonic signal is subjected to band-pass filtering (for example, 20-100 kHz), the signal effective value (RMS) or peak value of each sampling point or short-time window is calculated, the acoustic pressure intensity contour map or three-dimensional sound intensity distribution cloud map is drawn according to the moving track or fixed point array layout of the sensor on the surface of the cabinet body, the abnormal area is located by finding the local maximum value point in the sound intensity distribution map and analyzing the signal attenuation gradient (for example, calculating the area radius when the signal intensity decreases to half of the peak value) around the local maximum value point, and the steeper the gradient, the closer the sound source to the measurement point. The process of the first phase pattern (PRPD) extraction is as follows. The filtered ultrasonic signal is subjected to pulse detection, the amplitude and phase (phase angle relative to the power frequency voltage) of the pulse event exceeding the set threshold are extracted, the phase (0°-360°) is taken as the abscissa, the pulse amplitude is taken as the ordinate, the phase resolution pulse sequence pattern is drawn, the number or average amplitude of pulses in each phase interval is counted, and the phase resolution partial discharge pattern is formed. The UHF signal processing includes electromagnetic signal intensity distribution feature extraction and second phase pattern (PRPD) extraction. The process of the electromagnetic signal intensity distribution feature extraction is as follows. The original UHF signal is subjected to digital down-conversion and filtering, the signal envelope is extracted, the UHF sensor is moved to different detection points or a sensor array is used, and the signal envelope amplitude (usually represented by dBm) of each position is recorded. The electromagnetic signal leakage intensity distribution map is drawn, and the strongest leakage point of the signal is identified. The process of the second phase pattern (PRPD) extraction is similar to the ultrasonic signal processing. The UHF signal is subjected to pulse detection and phase correlation, and the PRPD pattern of the UHF signal is drawn.

[0098] Specifically, the fusion analysis unit is specifically configured to:

[0099] if the acoustic signal intensity of the abnormal region positioned by the ultrasonic signal is gradient distributed, and the first phase spectrum and the second phase spectrum both present symmetrical double-peak characteristics related to the power frequency, it is determined that the discharge type is an insulation defect type discharge;

[0100] if the second phase spectrum presents a single-peak characteristic appearing only in the negative half cycle of the power frequency, it is determined that the discharge type is a corona discharge.

[0101] The negative half cycle of the power frequency in the embodiment is around 270°; the working process of the fusion analysis unit is as follows: the fusion analysis unit receives two types of features from the feature extraction unit, and performs fusion judgment according to the intensity distribution consistency analysis logic and the phase spectrum consistency analysis logic, wherein the process of intensity distribution consistency analysis (used for positioning) includes spatial coordinate alignment, consistency calculation and positioning output, the process of spatial coordinate alignment is to map the sound pressure intensity distribution graph and the electromagnetic signal leakage intensity distribution graph onto the same switch cabinet three-dimensional structure model or two-dimensional development graph, the process of consistency calculation is to calculate the spatial correlation coefficient of the intensity distribution of the two signals, for example, the cabinet surface is meshed, the normalized cross-correlation value of the intensity of the two signals at each grid point is calculated, the region where the intensity of the two signals both exceeds the respective background threshold and overlaps in space is identified as a strong consistency abnormal region; the determination process of positioning output is: if there is a strong consistency abnormal region, it is determined that the discharge source is located in the cabinet internal space corresponding to the region, and the positioning accuracy can reach decimeter level, only one signal appears strong region, and the other signal is weak or irregular, it is prompted that it may be external interference or a specific type of discharge (such as pure corona discharge only UHF signal is strong), which needs to be further analyzed in combination with the spectrum; the phase spectrum consistency analysis is used for type identification, and the specific process includes spectrum comparison and feature parameter calculation and matching, the spectrum comparison is to superimpose and compare the ultrasonic PRPD spectrum (the first phase spectrum) and the very high frequency PRPD spectrum (the second phase spectrum), and the process of feature parameter calculation and matching is as follows: the double-peak index of the two spectra is calculated: whether the spectrum appears one main pulse aggregation peak in the positive half cycle (0°-180°) and the negative half cycle (180°-360°) of the power frequency cycle is checked; the symmetry index is calculated: the cumulative amplitude or quantity ratio of the positive and negative half cycle pulses is compared, and the ratio close to 1:1 is high symmetry; the phase aggregation degree index is calculated: the distribution width of the pulse on the phase axis is measured, the narrower the width, the stronger the aggregation.

[0102] For example, for the switch cabinet 925, the specific analysis process of the fusion analysis unit is as follows: the input sound intensity distribution map shows that the sound intensity at P3 point (rear lower cabinet observation window) is the highest (32 dBμV), and the attenuation gradient is presented upward, leftward and rightward; the input electromagnetic intensity distribution map shows that the UHF signal at the gap of the rear lower cabinet observation window is the strongest (58 dBm), and the signals at other positions are weaker; the input ultrasonic PRPD spectrum (P3 point) is a clear double peak, and the positive and negative half cycles are symmetrical; the input UHF PRPD spectrum (observation window point) is a clear double peak, and the positive and negative half cycles are symmetrical; through spatial consistency analysis, the two intensity maps are superimposed, it is found that the maximum sound intensity point and the maximum electromagnetic leakage point almost coincide in projection (both at the rear lower cabinet observation window), and the spatial correlation coefficient is as high as 0.85 (> threshold value 0.6), it is determined that the consistency is strong in the abnormal area, through spectrum consistency analysis: the double-peak, symmetry and aggregation degree indexes of the two spectrum are calculated, all meet the conditions for determining the discharge of the insulation defect type, and the sound signal gradient is extremely steep, the UHF signal is highly concentrated, which is more in line with the characteristics of surface discharge; then the output discharge type is the discharge of the insulation defect type, and is surface discharge; the positioning result is the inside of the rear of the rear lower cabinet observation window, coordinates (X, Y, Z), and the severity level is determined to be medium according to the amplitude size and historical cases comparison.

[0103] Referring to Figure 4 Fig. 2 is a structural schematic diagram of a second diagnostic module according to an embodiment of the present application;

[0104] Specifically, the second diagnostic module comprises:

[0105] a UHF dominant detection unit, configured to perform UHF detection on the switch cabinet to be detected under the second noise level, and acquire an amplitude and phase spectrum of a UHF signal;

[0106] a verification analysis unit, configured to determine whether the amplitude of the UHF signal continuously exceeds a third preset threshold value and the phase spectrum has regularity;

[0107] If the determination is yes, further acquire ultrasonic signal characteristics and SF6 decomposition characteristics to determine whether the ultrasonic signal and the UHF signal are synchronous and have the same phase characteristics, and output a diagnostic result according to the SF6 decomposition concentration and the pulse current method.

[0108] In the embodiment, the UHF dominant detection unit is a UHF sensor. The sensor has good directivity and high shielding performance, is preferentially deployed at a known signal leakage point (such as an observation window or a cable chamber cover plate gap) of the switch cabinet, is set to a high sampling rate (such as ≥1GS / s) to capture nanosecond-level pulses, and is enabled with hardware or digital filtering functions of the device to focus on analyzing a 300MHz-1.5GHz frequency band to avoid common communication frequency band interference. The verification and analysis unit performs envelope detection and pulse detection on the collected UHF time domain signals, extracts all pulses that exceed an initial pulse amplitude threshold (the threshold is higher than the average background noise level measured by the system in a quiet environment), calculates the average amplitude (unit: dBm) of the extracted pulses in 10 consecutive power frequency cycles (0.2 seconds), and requires that the measured average amplitude exceeds a third preset threshold in at least three consecutive sampling periods (such as 0.2 seconds of sampling every 5 seconds). The threshold is pre-set according to the voltage level and type of the switch cabinet. For example, for a 10kV switch cabinet, the typical value can be 50-60dBm. The phase pattern has the following regularity. The extracted pulse sequence is subjected to phase analysis to generate an initial UHF PRPD pattern. Even in high noise, a real discharge will show a certain phase aggregation. The regularity index of the pattern can be calculated. The lower the entropy value, the stronger the phase aggregation. The index is required to be lower than a preset value. If the amplitude continuously exceeds the threshold and the phase pattern has regularity, it is preliminarily determined that the UHF signal is not random interference, suspected to be a real discharge, and triggers the next step of multi-feature verification.

[0109] Specifically, the second diagnostic module further includes:

[0110] The result generation unit is configured to output a diagnosis result of surface or near-surface discharge in response to the ultrasonic signal being synchronized with the UHF signal and having the same phase characteristics, and the SF6 decomposition product concentration being less than a preset concentration threshold; and output a diagnosis result of high-energy discharge having caused insulation chemical decomposition in response to the SF6 decomposition product concentration exceeding the preset concentration threshold and continuously rising.

[0111] In the method, the concentration growth rate is greater than a preset growth rate threshold, and it is determined that the SF6 decomposition product concentration is continuously rising.

[0112] In the embodiment, when the UHF signal is preliminarily verified, the result generating unit collects and fuses the ultrasonic wave and chemical evidence to perform rapid classification diagnosis, including ultrasonic wave signal synchronization verification and SF6 decomposition characteristic verification. The process of the ultrasonic wave signal synchronization verification is as follows: the ultrasonic wave sensor is immediately controlled to detect the cabinet position with the strongest UHF signal, and the collected ultrasonic wave pulse time sequence and the UHF pulse time sequence are cross-correlated or time difference histogram statistics. If the pulse occurrence times of the two signals are significantly higher than the random probability within a time window of ±1 microsecond, it is determined that the time domain is synchronized; the PRPD spectrum of the ultrasonic wave is generated, and the cross-correlation coefficient or the Euclidean distance of the characteristic parameters such as bimodality and symmetry of the UHF PRPD spectrum is calculated. If the similarity exceeds the preset threshold (such as a correlation coefficient > 0.7), it is determined that the phase characteristics are the same; the process of SF6 decomposition characteristic verification is as follows: for the gas-filled cabinet, the system reads the real-time concentration (unit: μL / L or ppm) of SO2 and H2S from the online monitoring unit of the gas chamber or through a portable detector, judges whether the concentration exceeds the preset concentration threshold, which is usually referred to the equipment manufacturer's standard or industry guide, for example, the SO2 warning threshold is 1-2 μL / L, and the attention value is 5 μL / L; the decomposition concentration data of the gas chamber in the past 7-30 days is called from the historical database, and the daily growth rate or the weekly growth rate of the current concentration is calculated; if the ultrasonic wave and the UHF signal are synchronized and the phase characteristics are the same, and the SF6 decomposition concentration is less than the preset concentration threshold, and the SF6 decomposition concentration growth rate is less than or equal to the preset growth rate threshold, it indicates that the discharge occurs in the surface or near-surface area of the equipment, and the energy is not high enough to cause electric arc, serious burn of insulation and a large amount of SF6 decomposition, so it is determined that there is active surface or near-surface discharge, and the monitoring frequency needs to be strengthened, and the relevant parts are cleaned, tightened or replaced when the next power-off opportunity occurs; if the ultrasonic wave and the UHF signal are synchronized and the phase characteristics are the same, the SF6 decomposition concentration is greater than or equal to the preset concentration threshold, or the concentration growth rate is greater than the preset growth rate threshold, it indicates that the discharge energy is high, and an electric arc or a continuous spark has been generated, resulting in decomposition of SF6 gas and solid insulation material, so it is determined that the high-energy discharge has caused chemical decomposition of the insulation material, and immediate power-off maintenance is required to avoid rapid degradation of insulation performance leading to breakdown; even in a harsh high-noise environment, the reliability of the diagnosis can be maximized.

[0113] The setting method of the preset growth rate threshold in the embodiment is as follows, including base value setting, dynamic adjustment and comprehensive judgment, the base value setting process is as follows, for a typical 110kV GIS device, the preset daily growth threshold of SO2 concentration is set to 0.2 μL / L / day, the preset daily growth threshold of H2S is set to 0.1 μL / L / day, and the relative daily growth rate threshold is set to 10% / day; the dynamic adjustment process is as follows, the threshold is optimized according to the actual operation data of the device, in the initial stage of operation of the device, the system learns the background fluctuation level of the concentration of the decomposition product, and automatically sets the threshold to 3 times the standard deviation, with the increase of the running time, the system establishes a concentration segmentation threshold model: when the SO2 concentration is lower than 1 μL / L, a strict threshold of 0.1 μL / L / day is adopted; when the concentration is between 1-5 μL / L, a threshold of 0.2 μL / L / day is adopted; when the concentration is higher than 5 μL / L, a threshold of 0.3 μL / L / day is adopted, but a more stringent relative growth rate threshold of 5% / day is simultaneously started for double judgment; the comprehensive judgment rule is that when the SF6 decomposition product concentration meets the following two conditions at the same time, the system determines that the condition A (absolute growth) is met: the daily concentration growth value is greater than or equal to the preset daily growth threshold; the condition B (trend confirmation) is met: the sliding average of the concentration in the last 3 days is greater than or equal to twice the preset daily growth threshold than the sliding average in the last 3 days; for example, the SO2 concentration of the GIS gas chamber is measured as 1.2, 1.5 and 1.9 μL / L for three consecutive days. The preset daily growth threshold is 0.2 μL / L, and the single-day growth is 0.3 and 0.4 μL / L, respectively, both of which exceed 0.2 μL / L (satisfying condition A), and through trend confirmation, the three-day sliding average (1.53) is greater than the three-day sliding average (assuming 1.1) by 0.43 μL / L, which exceeds 0.4 μL / L (satisfying condition B); therefore, the system determines that the SF6 decomposition product concentration of the gas chamber is continuously rising, and combined with other characteristics, a serious alarm of high-energy discharge causing chemical decomposition is triggered.

[0114] Specifically, the second diagnosis module further comprises a deep confirmation unit, configured to:

[0115] When the UHF signal is strong but the ultrasonic wave signal is not synchronized and the SF6 decomposition product concentration is normal, triggering the pulse current method for confirmation;

[0116] The deep confirmation unit controls the installation of the differential Rogowski coil, applies an alternating test voltage, collects a pulse current signal, and analyzes the PRPD spectrum characteristics and the discharge amount to diagnose and locate suspected internal air gap discharge or deep fault, and outputs the diagnosis result.

[0117] In the present embodiment, when the UHF signal is strong but the ultrasonic wave is not synchronized, and the SF6 decomposition concentration is normal, it may be a pure internal air gap discharge or a deep fault, which is difficult to diagnose by conventional live detection means, and requires pulse current method. The depth confirmation unit is triggered, suggesting power-off detection, and applying for planned power-off of the switch cabinet (or the entire interval). On the grounding downlead of the switch cabinet (or the corresponding GIS gas chamber) that has been powered off and has safety measures in place, install a differential Rogowski coil. The bandwidth of the coil needs to cover the partial discharge pulse frequency (usually tens of kHz to several MHz). Install a wireless synchronous data acquisition terminal to ensure communication with the central host. According to IEC60270 standard, use a partial discharge test transformer to apply AC voltage from 0.7 to 1.2 times the rated phase voltage to the equipment. Synchronously collect pulse current signals, record the partial discharge inception voltage (PDIV) and extinction voltage (PDEV), and at 1.1 times the rated voltage, measure stably for at least 1 minute to obtain sufficient pulse data for statistical analysis. Calibrate the system by injecting a standard pulse with a known charge amount, directly read the apparent discharge quantity in picocoulombs (pC), and use it to quantify the severity of the defect. Draw a PRPD spectrum of the pulse current method and perform a deep analysis of the PRPD spectrum. The typical internal air gap discharge spectrum features are: a "double rabbit ear" or "half moon" distribution that is symmetrical in the positive and negative half cycles, and a very concentrated pulse phase distribution with stable discharge quantity. Compare it with the previous UHF PRPD spectrum to verify its consistency. If the measured discharge quantity q is greater than several hundred pC and the PRPD spectrum shows typical internal discharge characteristics, it is diagnosed as a serious internal insulation deterioration, such as air gap or crack. Combining the measurement results of multiple Rogowski coils, precise positioning can be achieved in meters or even decimeters by using the pulse arrival time difference, and the specific insulation piece or conductor connection site can be clearly pointed out. Output the final report containing accurate discharge quantity, defect type, and positioning information as the core basis for maintenance decision-making.

[0118] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A switch cabinet partial discharge on-line intelligent detection system, characterized in that, The application relates to a TEV detection system for switch cabinets, which comprises: a preliminary screening module for scanning each switch cabinet by using a TEV detector to obtain corresponding real-time TEV detection values, calculating real-time TEV value increments and average TEV detection values according to the real-time TEV detection values, and determining suspected abnormal cabinets and abnormal cabinets in each switch cabinet according to the real-time TEV value increments and the average TEV detection values; an environment monitoring module connected with the preliminary screening module, used for monitoring the environmental background noise level and determining whether the current environment is at a first noise level or a second noise level; a first diagnosis module connected with the preliminary screening module and the environment monitoring module respectively, used for synchronously performing ultrasonic detection and UHF detection on the abnormal cabinet in response to the current environment being at the first noise level, outputting an abnormal diagnosis result according to wave signal characteristics and atlas analysis; a second diagnosis module connected with the preliminary screening module and the environment monitoring module respectively, used for performing UHF detection on the switch cabinet to be detected in response to the current environment being at the second noise level, obtaining an amplitude and phase atlas of a UHF signal, and judging whether the amplitude of the UHF signal continuously exceeds a third preset threshold value and the phase atlas has regularity; if the judgment is yes, further obtaining ultrasonic signal characteristics and SF6 decomposition characteristics to determine whether the ultrasonic signal and the UHF signal are synchronous and have the same phase characteristics, and outputting an abnormal diagnosis result according to SF6 decomposition concentration and a pulse current method; a feedback adjustment module connected with the environment monitoring module, the first diagnosis module and the second diagnosis module respectively, used for dynamically adjusting a preset noise threshold value in the environment monitoring module for distinguishing the first noise level from the second noise level according to actual verification results of historical diagnosis events; the preliminary screening module comprises: a baseline establishment unit for establishing dynamic baseline values and normal fluctuation ranges of each switch cabinet at different time periods according to historical TEV detection data; a real-time calculation unit for comparing the real-time TEV detection values with the dynamic baseline values of the corresponding time periods to calculate the real-time TEV value increments; a horizontal comparison unit for calculating relative differences between the real-time TEV detection values of each switch cabinet and average TEV detection values of adjacent switch cabinets at the same detection position; a threshold value determination unit for determining a first preset threshold value based on the dynamic baseline values, a signal attenuation coefficient and environmental parameters; a comprehensive judgment unit for judging that the corresponding switch cabinet is an abnormal cabinet when the real-time TEV value increment is greater than the first preset threshold value, judging that the corresponding switch cabinet is an abnormal cabinet when the real-time TEV value increment is less than or equal to the first preset threshold value and the relative difference exceeds a second preset threshold value, and otherwise judging that the corresponding switch cabinet is a suspected abnormal cabinet; the first diagnosis module comprises: a joint detection unit for synchronously collecting ultrasonic signals and UHF signals of the abnormal cabinet; a feature extraction unit for extracting a first phase atlas and sound signal intensity distribution characteristics from the ultrasonic signals and extracting a second phase atlas and electromagnetic signal intensity distribution characteristics from the UHF signals; The fusion analysis unit is configured to compare and analyze the first phase pattern and the second phase pattern, and determine a discharge type and locate an abnormal area according to consistency of intensity distribution of the acoustic signal and the electromagnetic signal.

2. The partial discharge on-line intelligent detection system of switch cabinet according to claim 1, characterized in that, The threshold determination unit comprises: a threshold calculation sub-unit configured to determine the first preset threshold based on a statistical standard deviation multiple of the dynamic baseline value; a threshold correction sub-unit configured to adaptively correct the first preset threshold according to a signal attenuation coefficient of the detection point position and an environmental parameter.

3. The partial discharge on-line intelligent detection system of switch cabinet according to claim 1, characterized in that, The environmental monitoring module comprises: an environmental determination unit configured to determine a current environmental background noise level; if the background noise is less than a preset noise threshold, it is determined that the current environment is at a first noise level, and if the background noise is greater than or equal to the preset noise threshold, it is determined that the current environment is at a second noise level.

4. The partial discharge on-line intelligent detection system of switch cabinet according to claim 1, characterized in that, The feedback adjustment module is configured to: record diagnostic events completed at the first noise level and the second noise level and their final confirmation results, and calculate a first misjudgment ratio and a second misjudgment ratio; if the first misjudgment ratio is greater than a first ratio threshold, the preset noise threshold is increased; if the second misjudgment ratio exceeds a second ratio threshold, the preset noise threshold is decreased; wherein the first misjudgment ratio is a proportion of events that are finally confirmed as interference or false positives in the diagnoses completed at the first noise level; and the second misjudgment ratio is a proportion of events that are finally confirmed as real discharge defects but are initially diagnosed as interference or events to be observed in the diagnoses completed at the second noise level.

5. The partial discharge on-line intelligent detection system of switch cabinet according to claim 1, characterized in that, The fusion analysis unit is specifically configured to: if the acoustic signal intensity of the abnormal area located by the ultrasonic signal presents a gradient distribution, and the first phase pattern and the second phase pattern both present symmetrical double-peak characteristics related to the power frequency, it is determined that the discharge type is an insulation defect type discharge; if the second phase pattern presents a single-peak characteristic that only appears in the negative half cycle of the power frequency, it is determined that the discharge type is a corona discharge.

6. The partial discharge on-line intelligent detection system of switch cabinet according to claim 1, characterized in that, The second diagnostic module comprises: a UHF dominant detection unit configured to perform UHF detection on the switch cabinet to be detected at the second noise level to obtain an amplitude and phase spectrum of a UHF signal; a verification analysis unit configured to determine whether the amplitude of the UHF signal continuously exceeds a third preset threshold and the phase spectrum has regularity; if the determination is yes, further obtain ultrasonic signal characteristics and SF6 decomposition characteristics to determine whether the ultrasonic signal and the UHF signal are synchronized and have the same phase characteristics, and output a diagnosis result according to the SF6 decomposition concentration and the pulse current method.

7. The partial discharge on-line intelligent detection system of switch cabinet according to claim 6, characterized in that, The second diagnostic module further comprises: a result generation unit configured to output a diagnosis result of surface or near-surface discharge in response to the ultrasonic signal and the UHF signal being synchronized and having the same phase characteristics, and the SF6 decomposition concentration being less than a preset concentration threshold; and output a diagnosis result that high-energy discharge has caused insulation chemical decomposition in response to the SF6 decomposition concentration exceeding the preset concentration threshold and continuously rising.

8. The partial discharge on-line intelligent detection system of switch cabinet according to claim 7, characterized in that, The second diagnostic module further comprises a depth confirmation unit configured to: trigger the pulse current method for confirmation when the UHF signal is strong but the ultrasonic signal is not synchronized and the SF6 decomposition concentration is normal. The deep confirmation unit controls the installation of the differential Rogowski coil, applies an alternating test voltage, collects a pulse current signal, and diagnoses and locates suspected internal air gap discharge or deep faults by analyzing PRPD pattern characteristics and discharge capacity, and outputs a diagnosis result.

Images