A method and system for monitoring the insulation condition of a power supply device

Abstract

The application relates to the technical field of power supply device monitoring, and discloses a power supply device insulation state fault monitoring method, which comprises the following steps: acquiring an electrical signal sequence generated in the operation process of a power supply device; extracting an abnormal sequence segment with an insulation performance attenuation trend from the electrical signal sequence; extracting an abnormal feature representing the insulation performance attenuation degree from the abnormal sequence segment, and calculating a risk assessment value according to the abnormal feature; acquiring the time sequence features of all abnormal sequence segments appearing in a preset time window, and obtaining a warning boundary according to the time sequence features; and outputting an insulation performance attenuation warning signal when the risk assessment value is greater than the warning boundary. The application realizes the gradual and cumulative identification of sparse and intermittent insulation abnormalities by constructing a complete monitoring link from abnormality extraction, risk assessment to dynamic boundary determination, solves the problem that the traditional fixed threshold fails in the critical degradation stage, and realizes the accurate and early warning of the sudden breakdown fault of the power supply device.

Description

Technical Field

[0001] This application relates to the field of power supply device monitoring technology, and more specifically, to a method and system for monitoring insulation status faults in power supply devices. Background Technology

[0002] For monitoring the insulation status of power supply equipment, existing conventional methods mainly focus on the judgment and warning boundaries of insulation risk. In healthy or obviously faulty states, abnormal manifestations are either virtually nonexistent or relatively persistent, making it relatively easy for monitoring tasks to determine non-warning or warning status based on established boundaries. However, after long-term operation, the insulation performance of power supply equipment gradually declines. Before formal failure, it typically does not continuously exhibit obvious fault characteristics but rather appears sporadically with short-lived warning signs. When insulation enters this critical degradation stage but has not yet experienced a clear breakdown, the abnormality is no longer a long-term, stable, high-risk state but rather appears sparsely, intermittently, repeatedly, but gradually increasing. In this evolution from an operable state to a critical failure state, the link in recognizing the continuity of risk between multiple short-term anomalies and the boundary formation link between sporadic anomalies accumulating to a warning state become weak points in monitoring. A single short-term anomaly is often insufficient to continuously exceed the warning boundary, but multiple anomaly fragments show a gradual evolutionary trend towards failure over time. When faced with this state where a single anomaly is insufficient for recognition, but continuous neglect would lose its evolutionary significance, the conditions for recognizing short-term anomalies begin to deteriorate. If judged solely on individual segments, most short-term warning signs would be individually zeroed out, failing to accumulate enough to form an early warning system. Conversely, acknowledging all segments together directly could easily lead to misjudging occasional, harmless fluctuations as continuous deterioration. This results in a situation where, when the insulation of a power supply device is in a critical degradation stage, and the precursors to breakdown gradually appear as sparse, intermittent, and repetitive short-term anomalies, the lack of boundary rules for the stable acknowledgment and continuous acceptance of multiple short-term anomalies prevents the early warning system from accurately reflecting the gradual accumulation of insulation risks. Consequently, this severely impacts the early identification of sudden breakdowns.

[0003] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application aims to provide a method and system for monitoring insulation status faults in power supply devices. This system addresses the technical problem that existing technologies, when facing critical insulation degradation stages, cannot reliably acknowledge and continuously accept sparse and intermittent short-term anomalies, resulting in the early warning formation process failing to accurately reflect the gradual accumulation of insulation risks.

[0005] In a first aspect, this application provides a method for monitoring insulation status faults in a power supply device, the method comprising:

[0006] Acquire the electrical signal sequence generated during the operation of the power supply device;

[0007] Anomaly screening is performed on the electrical signal sequence to extract abnormal sequence segments with a trend of insulation performance degradation.

[0008] Anomaly features characterizing the degree of insulation performance degradation are extracted from the anomalous sequence fragments, and a risk assessment value for the anomalous sequence fragments is calculated based on the anomalous features.

[0009] Obtain the temporal characteristics of all the abnormal sequence fragments that appear within a preset time window, and obtain the warning boundary based on the temporal characteristics;

[0010] The risk assessment value is compared with the warning boundary. When the risk assessment value is greater than the warning boundary, an insulation performance degradation warning signal is output.

[0011] Secondly, this application provides a multi-scenario inspection task management system for the industrial internet, used to execute the aforementioned power supply device insulation status fault monitoring system, which includes:

[0012] The signal acquisition module is used to acquire the electrical signal sequence generated during the operation of the power supply device;

[0013] An anomaly identification module is used to screen the electrical signal sequence for anomalies and extract abnormal sequence segments with a trend of insulation performance degradation from the electrical signal sequence.

[0014] The risk assessment module is used to extract abnormal features characterizing the degree of insulation performance degradation from the abnormal sequence fragments, and to calculate the risk assessment value of the abnormal sequence fragments based on the abnormal features.

[0015] The early warning boundary module is used to acquire the temporal characteristics of all the abnormal sequence fragments that appear within a preset time window, and to obtain the early warning boundary based on the temporal characteristics.

[0016] The early warning output module is used to compare the risk assessment value with the early warning boundary. When the risk assessment value is greater than the early warning boundary, it outputs an insulation performance degradation early warning signal.

[0017] In summary, this application provides a method and system for monitoring insulation status faults in power supply devices. By collecting electrical signal sequences from the operation of the power supply device and conducting anomaly screening, it accurately extracts abnormal sequence segments showing a trend of insulation performance degradation. Combining the abnormal characteristics, it quantifies the insulation degradation risk assessment value and dynamically determines the early warning boundary based on the abnormal temporal characteristics within a preset time window. This solution can accurately capture the gradual degradation characteristics of the insulation performance of the power supply device, effectively avoiding the problems of traditional fixed threshold early warning methods being unable to adapt to dynamic changes in operating conditions, and having high rates of missed alarms and false alarms. It can accurately identify the insulation degradation fault risk under different temporal states, fundamentally improving the timeliness and accuracy of insulation fault identification in power supply devices, achieving early warning of insulation performance degradation faults, and ensuring the safe and stable operation of the power supply device. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a method for monitoring the insulation status faults of a power supply device, as provided in an embodiment of this application.

[0019] Figure 2 This is a schematic diagram of the structure of a power supply device insulation status fault monitoring system provided in an embodiment of this application.

[0020] Labeling Explanation: 1. Signal Acquisition Module; 2. Anomaly Identification Module; 3. Risk Assessment Module; 4. Early Warning Boundary Module; 5. Early Warning Output Module. Detailed Implementation

[0021] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0022] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] refer to Figure 1 This application proposes a method for monitoring insulation status faults in a power supply device, the method comprising:

[0024] A1. Obtain the electrical signal sequence generated during the operation of the power supply device;

[0025] A2. Perform anomaly screening on electrical signal sequences and extract abnormal sequence segments with a trend of insulation performance degradation from the electrical signal sequences;

[0026] A3. Extract abnormal features from abnormal sequence fragments to characterize the degree of insulation performance degradation, and calculate the risk assessment value of abnormal sequence fragments based on the abnormal features;

[0027] A4. Obtain the temporal characteristics of all abnormal sequence segments appearing within a preset time window, and obtain the warning boundary based on the temporal characteristics;

[0028] A5. Compare the risk assessment value with the warning boundary. When the risk assessment value is greater than the warning boundary, output an insulation performance degradation warning signal.

[0029] In this application, the electrical signal sequence refers to a digital signal acquired at high speed by a high-frequency current sensor, ultra-high frequency antenna, or built-in partial discharge sensor installed on the grounding lead or cable shield of a power supply device, which carries high-frequency response information of the insulation state.

[0030] An abnormal sequence segment is a short data segment with a clear start and end time cut out from a continuous electrical signal sequence. This segment contains transient pulse characteristics suspected to be caused by insulation degradation.

[0031] Risk assessment value is a comprehensive indicator that quantifies the degree of insulation degradation. It not only reflects the immediate damage caused by a single anomaly, but also includes the residual impact of historical anomalies.

[0032] The warning boundary is a threshold dynamically calculated based on the temporal characteristics of recent anomalies, used to distinguish between tolerable normal fluctuations and critical degradation states requiring alarm. For example, in the scenario of transformer bushing insulation monitoring, two short-term anomalies of the same amplitude, if occurring separately, may both have low risk assessment values; however, if these two anomalies recur within a short period, the risk assessment value corresponding to the latter anomaly will be superimposed with the residual effect of the former. In this case, even if the amplitude of a single anomaly does not increase significantly, it may approach the warning boundary due to the continuous assumption of risk.

[0033] The trend of insulation performance degradation refers to the directional deterioration of certain characteristic quantities in an electrical signal sequence over time, which can be reflected by the continuous change trajectory of increased leakage current, increased dielectric loss value, or increased partial discharge amplitude.

[0034] The degree of insulation performance degradation refers to the result obtained after quantitatively evaluating abnormal sequence segments, reflecting the level to which the insulation performance has deteriorated. It can be reflected by the classification of mild degradation, moderate degradation, and severe degradation or by a continuous degradation index between 0 and 1.

[0035] Anomaly features refer to key feature quantities extracted from anomalous sequence fragments that can characterize the degree of insulation performance degradation. They can be represented by time-domain features, frequency-domain features, trend features, or deep feature vectors extracted from large time-series models.

[0036] Insulation performance degradation early warning signal refers to the alarm information output by the system when the risk assessment value exceeds the warning boundary. It can be reflected in the form of audible and visual alarm, background reminder, work order generation or platform push.

[0037] Acquire the electrical signal sequence generated during the operation of the power supply unit. Specifically, this involves collecting electrical signal data such as voltage, current, or partial discharge generated by the power supply unit during operation using sensors or monitoring equipment. This can be achieved using high-precision current transformers, voltage transformers, or ultra-high frequency partial discharge sensors, which convert electrical signals into digital signals that can be processed by the system.

[0038] Anomaly screening of electrical signal sequences involves extracting anomalous sequence segments exhibiting a trend of insulation performance degradation. Specifically, preliminary analysis of the collected electrical signal data identifies signal segments deviating from normal operating modes. Over a continuous period, the amplitude, frequency, and waveform morphology of these segments are observed to determine if they exhibit continuous deviations from normal values ​​and a progressively worsening trend. If such deviations are observed, the signal segment is deemed to have an insulation performance degradation trend. This process can be implemented using statistical methods, machine learning algorithms, or expert system rules. For example, a baseline threshold can be set; when the amplitude or frequency of partial discharge pulses in an electrical signal exceeds this threshold, and this deviation continues for a continuous period with increasing magnitude, the segment is marked as an anomalous sequence segment. This preliminary screening of the original signals allows focus on potentially problematic signal components, improving the efficiency and accuracy of subsequent analysis.

[0039] This process extracts anomalous features characterizing the degree of insulation performance degradation from abnormal sequence fragments and calculates risk assessment values ​​for these fragments based on these features. Specifically, it involves in-depth analysis of identified anomalous sequence fragments to quantify the severity of their insulation performance degradation and converting it into a risk value. For example, partial discharge pulse counts, average discharge quantity, maximum discharge quantity, and discharge repetition rate can be extracted as anomalous features from the anomalous sequence fragments, and then a risk assessment value is calculated using a weighted summation method. This step transforms abstract electrical signals into quantified risk assessment values. By extracting features that directly reflect the degree of insulation performance degradation and performing calculations based on these features, the risk of the insulation state can be intuitively measured and compared.

[0040] The system acquires the temporal characteristics of all abnormal sequence segments appearing within a preset time window and derives the warning boundary based on these characteristics. Specifically, within a certain time range, it collects time-related information of all identified abnormal sequence segments and dynamically determines a warning threshold based on this information. This can be achieved using time series analysis methods. For example, it can statistically analyze the temporal characteristics of abnormal sequence segments within the preset time window, such as occurrence frequency, duration, and interval, and then use moving average, exponential smoothing, or adaptive thresholding algorithms to calculate the warning boundary. This step, by analyzing the temporal distribution and clustering of abnormal events, allows for dynamic adjustment of the warning threshold. The dynamic generation of the warning boundary enables the system to better adapt to the gradual and cumulative characteristics of insulation performance degradation, avoiding false alarms and missed alarms.

[0041] The system compares the risk assessment value with the warning boundary. When the risk assessment value exceeds the warning boundary, an insulation performance degradation warning signal is output. Specifically, the currently calculated risk assessment value is compared with the dynamically generated warning boundary. Once the risk assessment value exceeds the warning boundary, an insulation performance degradation warning signal is issued. This can be implemented using logical judgment and alarm triggering mechanisms. For example, when the risk assessment value exceeds the warning boundary, it indicates that the risk of insulation performance degradation has reached a level requiring attention. This can trigger an audible and visual alarm, send an SMS or email notification to maintenance personnel, and display the warning level and related data on the monitoring interface.

[0042] Through the above technical solution, this application can achieve real-time and dynamic monitoring of the insulation status of power supply devices, effectively identify the trend of insulation performance degradation, and output early warning signals based on risk assessment values ​​and dynamic early warning boundaries. This method can transform abstract electrical signals into quantitative risk assessments and dynamically adjust the early warning thresholds in conjunction with the time dimension, thereby improving the accuracy and timeliness of early warnings, avoiding the lag of traditional periodic inspections, reducing the probability of insulation faults, and ensuring the safe and stable operation of power supply devices.

[0043] In some preferred embodiments, the step of screening the electrical signal sequence for anomalies and extracting anomalous sequence segments with a trend of insulation performance degradation from the electrical signal sequence includes:

[0044] Time-frequency domain analysis is performed on the electrical signal sequence to obtain its spectral and waveform characteristics.

[0045] Based on the preset normal electrical signal characteristics, the acquired spectral characteristics and waveform characteristics are compared item by item to screen out electrical signal sequence segments with a high proportion of high frequency components in the spectral characteristics or abrupt changes in the waveform characteristics. The screened electrical signal sequence segments are identified as potential abnormal sequence segments.

[0046] Based on the comparison between the spectral and waveform characteristics of potential anomalous sequence segments and the characteristics of normal electrical signals, it is determined whether the potential anomalous sequence segments have a trend of insulation performance degradation.

[0047] If a potential anomalous sequence fragment exhibits a trend of insulation performance degradation, then the potential anomalous sequence fragment is extracted as the final anomalous sequence fragment.

[0048] In this application, time-frequency domain analysis is performed on electrical signal sequences to obtain their spectral and waveform characteristics. Specifically, the electrical signal is transformed from the time domain to the frequency domain, and key information on its waveform is extracted. This allows for a more comprehensive capture of potential anomalies in the signal, providing a rich data foundation for subsequent anomaly screening. Specifically, a Fast Fourier Transform (FFT) can be used to convert the time-domain signal to the frequency-domain signal, obtaining spectral characteristics such as the amplitude and phase of each frequency component. Simultaneously, wavelet transform or Hilbert-Huang transform methods can be used to extract waveform characteristics such as the instantaneous amplitude, instantaneous frequency, instantaneous phase, peak value, root mean square value, waveform factor, and pulse number. For example, for an electrical signal sequence with a sampling frequency of 100kHz, an FFT analysis can be performed every second to obtain the spectral distribution from 0 to 50kHz, and the energy proportion of high-frequency components (e.g., above 10kHz) can be calculated. At the same time, it can monitor the instantaneous amplitude of the signal in real time, and when it changes more than a preset threshold (such as 20% of the normal amplitude) within 1 millisecond, it is recorded as a waveform change.

[0049] Using preset normal electrical signal characteristics as a benchmark, the acquired spectral and waveform characteristics are compared item by item to screen out electrical signal sequence segments with a high proportion of high-frequency components in the spectral characteristics or abrupt changes in the waveform characteristics. These screened electrical signal sequence segments are identified as potential abnormal sequence segments. Specifically, by comparing the acquired spectral and waveform characteristics with the normal benchmark, abnormal features that differ significantly from the normal operating state are identified. Among them, a high proportion of high-frequency components is usually a typical feature of early faults such as partial discharge in insulation, while abrupt changes in waveform may indicate abnormal events such as transient overvoltage or partial discharge. These are potential manifestations of insulation performance degradation. Therefore, these electrical signal segments are marked as potential anomalies, narrowing the scope of subsequent analysis. Specifically, a database of electrical signal characteristics under normal operating conditions can be established based on historical experience, containing spectral and waveform characteristics under various operating conditions. When the proportion of high-frequency components in the spectral characteristics of the real-time acquired electrical signal sequence exceeds the average value of the normal benchmark plus three standard deviations, or when the waveform characteristics show that the instantaneous amplitude changes by more than 50% in a short period of time (e.g., 100 microseconds), the signal sequence segment is marked as a potential anomaly. For example, if the average proportion of high-frequency components under normal operating conditions is 5% and the standard deviation is 1%, then when the proportion of high-frequency components reaches 8% in real time, the segment is screened as a potential anomaly.

[0050] Based on the comparison of the spectral and waveform characteristics of potential anomalous sequence segments with those of normal electrical signals, it is determined whether the potential anomalous sequence segments exhibit a trend of insulation performance degradation. Specifically, by analyzing the degree of deviation between the characteristics of potential anomalous segments and normal characteristics, some false anomalies caused by fluctuations in normal operating conditions can be eliminated, thereby improving the accuracy of the judgment. Specifically, deeper characteristics such as the duration, repeatability, and amplitude change rate of potential anomalous sequence segments can be analyzed. For example, if the high proportion of high-frequency components persists for a long time (e.g., exceeding 100 milliseconds) and repeats multiple times within a short period (e.g., within 1 minute), it is more likely to be related to insulation performance degradation. Furthermore, observing whether these anomalous characteristics show a trend of returning to the normal baseline within a certain time period can determine whether the anomaly caused by insulation degradation is a true anomaly. For example, if a regression trend exists, it can be determined to be a false anomaly caused by fluctuations in normal operating conditions that can be recovered without external intervention; otherwise, it is determined to be a true anomaly.

[0051] If a potential anomalous sequence segment exhibits a trend of insulation performance degradation, it is extracted as the final anomalous sequence segment. Specifically, this step confirms and extracts the preceding judgment results, ensuring that the final anomalous sequence segment is truly related to insulation performance degradation, providing reliable input for subsequent risk assessment and early warning. Specifically, when the judgment result clearly indicates that a potential anomalous sequence segment has an insulation performance degradation trend, the segment is extracted from the original electrical signal sequence and marked for processing by the subsequent risk assessment module. For example, if a potential anomalous sequence segment is judged to have an insulation performance degradation trend, its start time, end time, and internal spectral and waveform characteristic data will be packaged, stored as the final anomalous sequence segment, and passed to the subsequent risk assessment module.

[0052] Through the above technical solution, this application can specifically and accurately screen out abnormal sequence segments with insulation performance degradation trends from electrical signal sequences. Time-frequency domain analysis can comprehensively capture abnormal patterns in the signal. By comparing each characteristic with those of normal electrical signals, signal segments that significantly differ from normal operating conditions can be effectively identified. Further judgment of potential abnormal sequence segments can eliminate false anomalies caused by fluctuations in normal operating conditions, thereby improving the accuracy of the judgment. The finally extracted abnormal sequence segments are truly related to insulation performance degradation, providing reliable input for subsequent risk assessment and early warning, avoiding false alarms or missed alarms, and improving the reliability of insulation condition fault monitoring of power supply equipment.

[0053] In some preferred embodiments, the step of determining whether a potential anomalous sequence segment has an insulation performance degradation trend based on a comparison of the spectral and waveform characteristics of the potential anomalous sequence segment with the characteristics of a normal electrical signal includes:

[0054] Obtain information on the power grid operation status in the area where the power supply unit is located;

[0055] Feature parameter analysis was performed on the spectral and waveform features of potential anomalous sequence segments to obtain the corresponding feature parameters;

[0056] Based on the comparison results between the characteristic parameters and the characteristics of normal electrical signals, it is determined whether the occurrence time of the potential abnormal sequence segment coincides with the occurrence time of the event in the power grid operation status information that can cause abnormal electrical signal waveforms.

[0057] When there is a timing overlap, the potential abnormal sequence segment is determined to be a normal fluctuation in the operating condition; otherwise, the potential abnormal sequence segment is determined to have a trend of insulation performance degradation.

[0058] This application involves acquiring power grid operation status information for the area where the power supply device is located. Specifically, it involves acquiring power grid status data related to the operating environment of the power supply device. This can be achieved in various ways, such as by obtaining real-time power grid operation data through monitoring and data acquisition systems or distribution management systems, including voltage fluctuations, current changes, switching operations, load switching, lightning strikes, short-circuit faults, etc. This information can be obtained from data sources such as power grid dispatch centers, substations, or smart meters.

[0059] Feature parameter analysis is performed on the spectral and waveform characteristics of potential anomalous sequence segments to obtain the corresponding feature parameters. Specifically, the identified potential anomalous sequence segments are quantitatively analyzed to extract their key electrical characteristic parameters. This can be achieved in the following ways: for spectral characteristics, parameters such as the dominant frequency component, harmonic content, energy distribution, bandwidth, and center frequency can be calculated. For waveform characteristics, parameters such as peak value, RMS value, waveform distortion rate, rise time, fall time, pulse width, number of zero crossings, and transient duration can be calculated. These parameters can be obtained through signal processing methods such as Fourier transform, wavelet transform, and Hilbert-Huang transform.

[0060] Based on the comparison results between the characteristic parameters and the characteristics of normal electrical signals, it is determined whether the occurrence time of the potential abnormal sequence segment coincides with the occurrence time of events in the power grid operation status information that could cause abnormal electrical signal waveforms. Specifically, the characteristic parameters are compared with the characteristics of normal electrical signals. When the deviation of the characteristic parameters from the reference characteristics of normal electrical signals is within a preset allowable threshold range, it indicates that the suspected abnormal sequence segment does not exhibit obvious abnormal characteristics exceeding the normal disturbance range. Therefore, its occurrence time is further compared with the event time recorded in the power grid operation status information. The preset allowable threshold range refers to a pre-set allowable deviation interval for the characteristic parameters of normal electrical signals. This allowable deviation interval can be determined based on statistical results of historical normal operation data, empirical parameters, or standard reference values, and is used to characterize whether the deviation of the current characteristic parameter from the normal reference value is within an acceptable range. Specifically, this can be achieved in the following way: for example, when the deviation of the characteristic parameter from the normal electrical signal reference characteristic is less than or equal to the allowable threshold range, it is determined that the characteristic parameter is still within the acceptable fluctuation range, and the start timestamp of the potential abnormal sequence segment is compared with the timestamps of various events (such as switching actions, load changes, fault occurrences, etc.) recorded in the power grid operation status information. A time window can be set; if the occurrence time of the potential abnormal sequence segment falls within this time window, it is considered that there is a time overlap.

[0061] The system determines whether a potential abnormal sequence segment exhibits a trend of insulation performance degradation. Specifically, under certain conditions, the segment is classified as an event related to insulation performance degradation. This can be achieved in the following way: for example, if the occurrence time of a potential abnormal sequence segment does not coincide with the occurrence time of an event in the power grid operation status information that can cause abnormal electrical signal waveforms, the system will mark it as having a trend of insulation performance degradation.

[0062] The system classifies potentially abnormal sequence segments as normal fluctuations in operating conditions. Specifically, when certain conditions are not met, the segment is categorized as a normal operating phenomenon. This can be achieved in several ways, such as: if the occurrence time of a potentially abnormal sequence segment coincides with the occurrence time of an event in the power grid operating status information that could cause abnormal electrical signal waveforms, the system will mark it as a normal fluctuation in operating conditions. This means that the abnormal electrical signal is caused by a normal change in the power grid operating status, rather than by insulation performance degradation. This effectively distinguishes between electrical signal anomalies caused by changes in power grid operating status and anomalies caused by insulation performance degradation.

[0063] Through the above technical solution, this application improves the accuracy of judging whether potential abnormal sequence segments have a trend of insulation performance degradation, avoids misjudging normal fluctuations in operating conditions as a trend of insulation performance degradation, and thus improves the accuracy of fault monitoring.

[0064] In some preferred embodiments, the step of extracting anomalous features characterizing the degree of insulation performance degradation from the anomalous sequence fragments and calculating the risk assessment value of the anomalous sequence fragments based on the anomalous features includes:

[0065] Extract anomalous features from anomalous sequence fragments and calculate the current risk assessment value based on these features;

[0066] Retrieve historical risk assessment values ​​stored in the power supply unit, and calculate the residual risk values ​​of the historical risk values ​​that remain in the power supply unit at the current time according to the preset risk reduction rules.

[0067] The residual risk value is then added to the current risk value to serve as the risk assessment value for the abnormal sequence fragment.

[0068] In this application, anomalous features are extracted from anomalous sequence fragments, and a current risk assessment value is calculated based on these features. Specifically, features characterizing the degree of insulation performance degradation are first extracted from the identified anomalous sequence fragments. For example, parameters such as the amplitude, frequency, duration, and energy of the anomalous signal can be extracted as anomalous features. The current risk assessment value is then calculated based on these anomalous features. For instance, a risk assessment model can be constructed that takes the extracted anomalous features as input and outputs a quantified current risk assessment value through a pre-defined weighted relationship. This model can be a simple linear weighted model or a complex model based on machine learning algorithms (such as support vector machines or neural networks) that learns the mapping relationship between anomalous features and risk through training.

[0069] The system retrieves historical risk assessment values ​​stored within the power supply unit and calculates the residual risk value remaining in the unit at the current time, according to preset risk reduction rules. Specifically, the power supply unit stores historical risk assessment values, which record past insulation performance degradation risks. These historical risk assessment values ​​are retrieved when calculating the current risk. The preset risk reduction rules simulate the decay pattern of historical risks over time. Specifically, the risk reduction rule can be a mathematical function, such as an exponential decay function or a linear decay function. Its inputs include historical risk assessment values, the time difference between the historical risk occurrence time and the current time, and its output is the residual risk value remaining in the power supply unit at the current time. For example, a simple exponential decay rule can be expressed as: Residual Risk = Historical Risk Assessment Value × e^(-λΔt) × η, where λ is the decay coefficient, Δt is the time difference, e is the base of the natural logarithm, and η is the performance retention factor. The decay coefficient λ can be adjusted based on the characteristics of the insulation material, operating environment, etc. The performance retention factor η can be adjusted based on the rated output capacity of the power supply unit, the current actual output capacity, and the load level, etc. In this way, the ongoing impact of historical risks on the current insulation status can be quantified.

[0070] The residual risk value is added to the current risk value to obtain the risk assessment value for the abnormal sequence segment. Specifically, the calculated current risk assessment value is added to the historical residual risk value to obtain the final risk assessment value for the abnormal sequence segment. That is, the final risk assessment value = current risk assessment value + residual risk value. This allows the risk assessment value to better reflect the true evolution trend of the insulation status of the power supply device, thereby improving the accuracy and reliability of the early warning. In this way, this solution solves the problem that calculating the risk assessment value based solely on the current anomaly characteristics may not fully reflect the insulation status of the power supply device, because the accumulation and decay of historical risks have a significant impact on the current risk assessment, and failing to consider the persistence and decay patterns of historical risks may lead to inaccurate or untimely risk assessments.

[0071] By employing the aforementioned technical solution, this application addresses the problem that calculating risk assessment values ​​solely based on current anomaly characteristics may not fully reflect the insulation status of power supply equipment. This is because the accumulation and attenuation of historical risks significantly impact current risk assessments, and failing to consider the persistence and attenuation patterns of historical risks can lead to inaccurate or untimely risk assessments. This solution introduces historical risk assessment values ​​and risk reduction rules to calculate the residual values ​​of historical risks and adds them to the current risk assessment value, thereby providing a more comprehensive and accurate risk assessment result. This method better reflects the true evolution trend of the insulation status of power supply equipment, improving the accuracy and reliability of insulation performance degradation early warning.

[0072] In some preferred embodiments, the step of retrieving historical risk assessment values ​​stored in the power supply device and calculating the residual risk value of the historical risk value still remaining in the power supply device at the current time according to preset risk reduction rules includes:

[0073] Obtain operating time information of the power supply unit and current aging indicators of the insulation material;

[0074] Based on runtime information and aging indicators, adjust the charge decay time constant (e.g., the decay coefficient λ mentioned above) used to characterize the risk decay rate and the performance retention rate (e.g., the performance retention factor η mentioned above) used to characterize the power supply output capability within the risk reduction rules.

[0075] The adjusted charge decay time constant and performance retention rate are substituted into the risk reduction rule for calculation to obtain the residual risk value of the historical risk assessment at the current moment.

[0076] The application requests information on the operating time of the power supply unit. Specifically, it requests the cumulative operating time since the unit was put into operation. This can be achieved through methods such as recording with a built-in timer, querying from the equipment management system, or monitoring the unit's operating status using sensors and accumulating the time. Operating time information reflects the unit's usage duration and is a crucial factor affecting the aging of insulation materials.

[0077] Obtaining the current aging level index of insulation materials refers to acquiring the degree of performance degradation of the insulation material at the current moment. Specifically, it can be characterized by parameters such as partial discharge, insulation resistance, and breakdown voltage. These parameters can be obtained through online monitoring systems or periodic offline testing. The aging level index directly quantifies the actual degradation of insulation performance, providing a direct basis for subsequent parameter adjustments.

[0078] The charge decay time constant, which characterizes the rate of risk decay within the risk mitigation rules, is adjusted. Specifically, the value of the charge decay time constant is dynamically changed based on operating time information and aging degree indicators. When the operating time is long or the aging degree is high, the performance of the insulation material deteriorates, and the rate of risk decay accelerates. Therefore, the charge decay time constant is adjusted to be smaller to reflect the faster risk decay. This adjustment can be based on a preset lookup table, empirical formula, or machine learning model.

[0079] The performance retention rate, which characterizes power output capability, is adjusted within the risk mitigation rules. Specifically, the performance retention rate is dynamically changed based on operating time information and aging indicators. When the operating time is long or the aging level is high, the output capability of the power supply unit may decrease, so the performance retention rate will be adjusted lower to reflect the actual decline in power output capability. This adjustment can also be based on a preset lookup table, empirical formula, or machine learning model.

[0080] The adjusted charge decay time constant and performance retention rate are substituted into the risk reduction rule for calculation. Specifically, the dynamically adjusted charge decay time constant and performance retention rate are used as input parameters and substituted into a preset risk reduction mathematical model for calculation. This risk reduction rule is typically a function describing the decay of risk over time, such as an exponential decay model. In this way, the residual risk value of historical risk assessments at the current moment can be obtained. This makes the risk reduction rule no longer static but adaptively optimized according to the actual operating conditions of the power supply equipment and the aging degree of the insulation materials. For example, when the operating time is long or the aging degree is high, the charge decay time constant may be adjusted to be smaller, indicating a faster rate of risk decay; the performance retention rate may also be adjusted to be lower, indicating a decrease in power output capability. In this way, the decay process of insulation performance can be simulated more realistically.

[0081] Through the above technical solution, this application addresses the problem that the charge decay time constant used to characterize the risk decay rate and the performance retention rate used to characterize the power supply output capability are fixed within the risk reduction rules, failing to consider the actual operating time information of the power supply device and the current aging degree index of the insulation material. This may lead to inaccurate calculated residual risk values, failing to truly reflect the insulation performance degradation of the power supply device, thus affecting the accuracy of risk assessment. This solution can dynamically adjust the key parameters in the risk reduction rules based on the actual operating time information of the power supply device and the current aging degree index of the insulation material, making the calculated residual risk values ​​more accurate, thus truly reflecting the insulation performance degradation of the power supply device and improving the accuracy of risk assessment.

[0082] In some preferred embodiments, the step of obtaining the temporal characteristics of all the abnormal sequence segments appearing within a preset time window, and obtaining the warning boundary based on the temporal characteristics, includes:

[0083] Obtain the temporal characteristics of all abnormal sequence segments appearing within a preset time window, and extract the start and end times of each abnormal sequence segment from each temporal characteristic;

[0084] Based on the time intervals of each abnormal sequence segment obtained at each start and end time, multiple abnormal sequence segments with consecutive time intervals all less than a preset time threshold are divided into the same group as an abnormal cluster group.

[0085] The number of anomalous sequence fragments within each anomalous cluster is counted, and the density of the corresponding anomalous cluster is calculated based on the number of anomalous sequence fragments and the time span of the anomalous cluster.

[0086] The warning boundary is calculated based on the number and density of abnormal sequence fragments corresponding to each abnormal cluster group.

[0087] The application involves acquiring the temporal characteristics of all anomalous sequence fragments appearing within a preset time window, and extracting the start and end times of each anomalous sequence fragment from these characteristics. Specifically, preliminary temporal analysis is performed on the raw anomalous data to lay the foundation for subsequent aggregation analysis. This can be achieved by continuously monitoring the electrical signals of the power supply unit using sensors or a data acquisition system. When anomalous sequence fragments exhibiting a trend of insulation degradation are identified, their start and end times are recorded. For example, an anomalous sequence fragment might start at time T1 and end at time T2; therefore, T1 and T2 are its start and end times. These temporal characteristics can be stored in a database for subsequent processing.

[0088] The time intervals of each anomalous sequence segment are obtained based on their start and end times. Specifically, the time difference between adjacent anomalous sequence segments is calculated. This can be achieved as follows: For two consecutive anomalous sequence segments A and B, if the end time of A is T_A_end and the start time of B is T_B_start, then the time interval between them is T_B_start - T_A_end. Multiple consecutive anomalous sequence segments with time intervals all less than a preset time threshold are grouped into the same group as anomalous clusters. This means classifying temporally closely related anomalous events into an anomalous cluster. The preset time threshold can be set according to the actual application scenario and experience; for example, it can be set to 1 minute, 5 minutes, or 10 minutes. If the time intervals between consecutive anomalous sequence segments are all less than this threshold, they are considered to belong to the same anomalous cluster. For example, if there are three anomalous sequence segments A, B, and C, with a time interval of 3 minutes between A and B, a time interval of 2 minutes between B and C, and a preset time threshold of 5 minutes, then A, B, and C will be classified into the same anomalous cluster. This reflects that insulation performance may continuously deteriorate or frequently experience problems over a certain period of time, rather than being an isolated, occasional event. This clustering analysis can better capture the continuity and concentration characteristics of insulation performance degradation.

[0089] The number of anomalous sequence segments within each anomalous cluster is counted. Specifically, the number of anomalous sequence segments contained in each anomalous cluster is calculated. This can be achieved as follows: for an existing anomalous cluster, all anomalous sequence segments it contains are traversed and counted. For example, an anomalous cluster might contain 5 anomalous sequence segments. The density of the corresponding anomalous cluster is calculated based on the number of anomalous sequence segments and the time span of the cluster. This density quantifies the severity and duration of each anomalous cluster. The time span refers to the length of time between the start time of the first anomalous sequence segment and the end time of the last anomalous sequence segment in the cluster. The density can be defined as the number of anomalous sequence segments divided by the time span. For example, if an anomalous cluster contains 5 anomalous sequence segments and the time span is 10 minutes, then the density is 0.5 segments / minute. The number of anomalous sequence segments reflects the frequency of anomalous events, while the density, combining the number and the time span, more comprehensively characterizes the concentration of anomalous events within a specific time period, providing richer information for subsequent early warning boundary calculations.

[0090] The warning boundary is calculated based on the number and density of anomalous sequence fragments corresponding to each anomalous cluster. Specifically, the number and density of anomalous clusters are comprehensively considered to more accurately reflect the overall deterioration trend of insulation performance. This can be achieved by designing a function or model that takes the number and density of anomalous sequence fragments as input and outputs the warning boundary. For example, the warning boundary can be negatively correlated with the number and density of anomalous sequence fragments; that is, the more numerous and denser the fragments, the lower the warning boundary, indicating more severe insulation performance deterioration. By comprehensively considering the number and density of anomalous clusters, the overall deterioration trend of insulation performance can be more accurately reflected. This cluster-based warning boundary calculation method, compared to simply relying on a single anomalous event, can more effectively identify potential and persistent insulation performance degradation risks, thereby improving the timeliness and accuracy of warnings.

[0091] Through the above technical solution, this application can process the temporal characteristics of abnormal sequence segments more precisely. By dividing temporally closely linked abnormal events into abnormal clusters and comprehensively considering their quantity and density, the warning boundary can be calculated more accurately. This helps to better reflect the deterioration trend of insulation performance, improve the accuracy and timeliness of insulation performance degradation warning, and avoid false alarms or missed alarms caused by isolated sporadic events, thereby providing a more reliable basis for the maintenance and repair of power supply equipment.

[0092] In some preferred embodiments, the step of calculating the warning boundary based on the number and density of abnormal sequence fragments corresponding to each abnormal cluster group includes:

[0093] Based on the number and density of abnormal sequence fragments corresponding to each abnormal cluster, the local deterioration rate factor corresponding to each abnormal cluster is calculated.

[0094] For each local deterioration rate factor, its weight is determined by the number and density of abnormal sequence fragments in the corresponding abnormal cluster group. The weighted operation is performed on all local deterioration rate factors to obtain the global deterioration rate factor.

[0095] The warning boundary is calculated based on the global deterioration rate factor.

[0096] In the application, the local degradation rate factor refers to an index that reflects the rate of insulation performance deterioration in a local area. Specifically, it can be implemented by multiplying or weighting the number and density of anomalous sequence fragments. For example, a local degradation rate factor can be set. * * ,in The number of abnormal sequence fragments. For density, The weighting coefficients correspond to the number of abnormal sequence fragments. This refers to the weighting coefficient corresponding to the density. The weighting coefficient can be predetermined based on historical sample data, equipment operating experience, or optimization results, and is used to adjust the contribution of each feature in the weighting calculation of the local deterioration rate factor.

[0097] Weights refer to the numerical values ​​used in weighted calculations to measure the importance of each local deterioration rate factor. Specifically, they can be determined by a combination of the number and density of anomalous sequence fragments. For example, weights can be set... * * ,in For the service life of regional equipment, This is a statistical analysis of the frequency of similar insulation faults in the region throughout history. and A preset weighting coefficient is used; this coefficient can be predetermined based on historical sample data, equipment operating experience, or optimization results, and is used to adjust the contribution of each feature in the global deterioration rate factor weight calculation. Higher equipment aging levels and more historical faults correspond to higher weights. The larger the value, the more severe and concentrated the anomalies are in the region, and the greater the weight of the local deterioration rate factor in the calculation of the global deterioration rate factor.

[0098] Weighted calculation refers to multiplying each local deterioration rate factor by its corresponding weight, summing the results, and then dividing by the total weight to obtain the global deterioration rate factor. The global deterioration rate factor is an indicator that comprehensively reflects the overall deterioration trend of insulation performance within the entire preset time window. Specifically, it can be implemented using a weighted average method. For example, the global deterioration rate factor... Σ(Local deterioration rate factor) *Weight ) / Σ weight w, global deterioration rate factor The higher the value, the more severe the overall deterioration trend of insulation performance.

[0099] The warning boundary refers to the threshold used to determine whether the insulation performance has deteriorated to the level requiring a warning. Specifically, it can be calculated based on the global degradation rate factor through a preset functional relationship or lookup table. For example, the warning boundary... g(global deterioration rate factor) ), where g is a decreasing function.

[0100] Specifically, this solution aims to improve the accuracy of early warning boundaries through more refined calculation methods, thereby more effectively monitoring the insulation status of power supply devices.

[0101] First, based on the number and density of anomalous sequence fragments corresponding to each anomalous cluster, the local deterioration rate factor for each cluster is calculated. This step, by comprehensively considering the number of anomalous sequence fragments and their temporal density, can more accurately reflect the rate of insulation performance deterioration in a local area. A greater number and higher density of anomalous sequence fragments generally indicates more severe insulation performance deterioration, thus resulting in a larger calculated local deterioration rate factor.

[0102] Secondly, for each local deterioration rate factor, its weight is determined by the number and density of anomalous sequence fragments in the corresponding anomalous cluster. A weighted average is then applied to all local deterioration rate factors to obtain the global deterioration rate factor. This step, by assigning different weights to different local deterioration rate factors, can more comprehensively reflect the overall deterioration trend of insulation performance within the entire preset time window. The weight allocation is based on the number and density of anomalous sequence fragments in the anomalous cluster. This means that areas with more severe and concentrated anomalous conditions will have a greater weight in the calculation of the global deterioration rate factor, thus enabling the global deterioration rate factor to more accurately reflect the overall risk.

[0103] Finally, the warning boundary is calculated based on the global deterioration rate factor. By integrating local deterioration trends into a global deterioration trend and using this as a basis for calculating the warning boundary, the setting of the warning boundary becomes more scientific and reasonable, and more closely reflects the actual degradation of the insulation performance of the power supply equipment. When the risk assessment value exceeds this warning boundary calculated based on the global deterioration rate factor, an insulation performance degradation warning signal can be output in a timely manner, thus providing an important basis for decision-making in the maintenance and repair of the power supply equipment.

[0104] Through the above technical solution, this application can more accurately calculate the warning boundary to better reflect the deterioration trend of insulation performance, thereby improving the accuracy and timeliness of the warning.

[0105] In some preferred embodiments, after the step of calculating the warning boundary based on the number and density of abnormal sequence fragments, the following steps are included:

[0106] Obtain the reference warning boundary of the power supply device; and compare the warning boundary with the reference warning boundary;

[0107] When the warning boundary is smaller than the reference warning boundary, the warning boundary will be adjusted to the reference warning boundary.

[0108] Otherwise, based on the time difference between the current time and the end time of the most recent abnormal cluster, calculate the boundary rebound damping coefficient, which characterizes the degree of rebound constraint of the warning boundary; adjust the warning boundary according to the boundary rebound damping coefficient, and update the reference warning boundary to the adjusted warning boundary.

[0109] In this application, a reference warning boundary for the power supply device is obtained. Specifically, a preset warning boundary value used to measure the safety baseline of insulation performance is read from the power supply device's storage unit or external configuration. This can be achieved by reading a preset fixed value from the power supply device's non-volatile memory, or by obtaining a dynamic value set based on historical operating data and expert experience from a remote server.

[0110] The warning boundary is compared with a reference warning boundary. If the warning boundary is smaller than the reference warning boundary, the warning boundary is adjusted to the reference warning boundary. Specifically, if the currently calculated warning boundary is lower than the reference warning boundary, the value of the warning boundary is forcibly set to the value of the reference warning boundary, which can be achieved through an assignment operation. This effectively avoids the warning boundary being too low due to local data fluctuations or calculation errors, thereby improving the reliability of the warning and preventing missed reports.

[0111] Based on the time difference between the current moment and the end time of the most recent anomalous cluster activity, a boundary rebound damping coefficient, characterizing the degree of rebound constraint on the warning boundary, is calculated. Specifically, the time interval between the current point in time and the end time of the most recent anomalous cluster activity is calculated, and a damping coefficient is determined based on this time interval using a preset function or lookup table. This coefficient controls the speed and magnitude of the warning boundary rebound. It can be implemented using timestamp calculations and function mapping; for example, a larger time difference results in a smaller damping coefficient, indicating a lower degree of rebound constraint and a faster rebound of the warning boundary.

[0112] The warning boundary is adjusted based on the boundary rebound damping coefficient. Specifically, the current warning boundary is calculated with the boundary rebound damping coefficient to achieve dynamic adjustment of the warning boundary. This can be achieved using mathematical operations such as multiplication or weighted averaging. For example, the warning boundary can be multiplied by a damping coefficient less than 1 to lower it, or a weighted average can be used, averaging the current warning boundary with a lower value, with the weights determined by the damping coefficient. The boundary rebound damping coefficient is calculated based on the time difference between the current time and the end time of the most recent anomaly cluster, reflecting the time process of the system returning to normal after a risk event. The larger the time difference, the smaller the impact of the risk event may be or the situation may have been mitigated. Therefore, the boundary rebound damping coefficient will be adjusted accordingly to allow the warning boundary to rebound more flexibly. This damping coefficient adjustment mechanism based on time difference enables the warning boundary to more accurately reflect the actual attenuation trend of the power supply device's insulation performance, avoiding the warning boundary remaining too high after the risk has been mitigated, thereby reducing false alarms and making the warning system more adaptable.

[0113] The reference warning boundary is updated to the adjusted warning boundary. Specifically, the warning boundary value adjusted by the boundary rebound damping coefficient is stored as the new reference warning boundary value for subsequent warning boundary calculations and comparisons. This can be achieved through variable assignment or data writing to memory. This step enables the warning system to continuously learn and adapt to changes in the operating status of the power supply unit, further improving the long-term effectiveness and accuracy of the warnings.

[0114] Through the above technical solution, this application solves the problems of unstable or inaccurate early warning boundary calculation, particularly the inability to effectively reflect the actual risk situation when the calculated early warning boundary is lower than the reference early warning boundary, and the inflexibility of the early warning boundary rebound mechanism when the risk is temporarily mitigated, failing to adjust in time to adapt to the new operating state, thus affecting the accuracy and timeliness of early warnings. This solution introduces a reference early warning boundary to ensure that the early warning boundary does not fall below the safety baseline, improving the reliability of early warnings and preventing missed alarms. Simultaneously, by introducing a boundary rebound damping coefficient, the early warning boundary can be dynamically adjusted according to the degree of risk mitigation, improving the adaptability and accuracy of early warnings and reducing false alarms.

[0115] In some preferred embodiments, the step of comparing the risk assessment value with the warning boundary, and outputting an insulation performance degradation warning signal when the risk assessment value is greater than the warning boundary, includes:

[0116] Load pre-configured reference information; the reference information includes several warning levels and the corresponding amplitude range for each warning level;

[0117] Calculate the extent by which the risk assessment value exceeds the warning boundary, compare this extent with the range in the reference information, determine the corresponding warning level, and output the corresponding level of insulation performance degradation warning signal.

[0118] The application loads pre-configured reference information. Specifically, before classifying the warning, the system preloads data containing the warning level and its corresponding amplitude range. This can be achieved by storing the reference information in a database, configuration file, or memory. For example, the amplitude range for "slight attenuation" can be set to 0-10%, for "moderate attenuation" to 10%-30%, and for "severe attenuation" to above 30%.

[0119] Calculate the extent to which the risk assessment value exceeds the warning boundary. Specifically, when the risk assessment value is greater than the warning boundary, calculate the difference between the two and convert it into a relative or absolute magnitude value. The relative magnitude can be calculated as (risk assessment value - warning boundary) / warning boundary * 100%, or the absolute magnitude can be calculated directly as (risk assessment value - warning boundary).

[0120] The calculated exceedance is compared with the ranges in the reference information to determine the corresponding warning level. Specifically, the calculated exceedance is compared with the ranges corresponding to each pre-defined warning level to determine which range the exceedance falls into, thus determining the appropriate warning level. This can be achieved by iterating through the ranges in the reference information and checking if the calculated exceedance falls within a certain range.

[0121] The system outputs an early warning signal indicating the corresponding level of insulation performance degradation. Specifically, after determining the warning level, the system generates and sends an early warning signal containing information about that level. This can be achieved by displaying the warning signal through a user interface, sending an SMS or email, or pushing the warning information to other systems via an API interface.

[0122] This solution combines the steps of comparing risk assessment values ​​with warning boundaries, and outputting an insulation performance degradation warning signal when the risk assessment value exceeds the warning boundary. By introducing a warning grading mechanism, the original single warning signal is enhanced, making the warning information more refined. By loading reference information, calculating the exceedance range, and comparing the results, the severity of insulation performance degradation can be accurately quantified into different warning levels. This solves the problem that simply outputting a single warning signal cannot reflect the severity of insulation performance degradation, and provides more precise guidance for subsequent maintenance and repair.

[0123] Through the above technical solution, this application solves the problem that simply outputting a warning signal cannot reflect the severity of insulation performance degradation, and provides more refined guidance for subsequent maintenance and repair. By classifying the warning signals, the degree of insulation performance degradation can be reflected more accurately, enabling maintenance personnel to rationally arrange maintenance and repair plans according to the severity of the warning level, avoiding over-maintenance or under-maintenance, thereby improving the efficiency and accuracy of insulation condition fault monitoring of power supply equipment.

[0124] In some preferred embodiments, a specific example is given below. In a high-voltage switchgear insulation condition monitoring scenario, an ultra-high frequency partial discharge sensor is deployed to continuously collect partial discharge signals inside the switchgear at a frequency of 1000 sampling points per second, forming an electrical signal sequence.

[0125] The system first performs real-time Fourier transform on the acquired electrical signal sequence to analyze its spectral characteristics, and then combines this with wavelet transform to analyze its waveform characteristics. The preset characteristics of normal electrical signals are no obvious high-frequency components and a stable waveform. When the proportion of high-frequency components in the spectrum continuously exceeds 20% or when the waveform shows abrupt changes such as spikes or glitches, the corresponding signal segment is marked as a potential abnormal sequence segment.

[0126] Furthermore, the system acquires information on the power grid operation status of the area where the switchgear is located, such as whether lightning strikes, switch operations, or short-circuit faults have occurred. Simultaneously, it performs characteristic parameter analysis on the spectral and waveform characteristics of potential abnormal sequence segments, such as calculating the pulse amplitude and repetition frequency of partial discharges. If the characteristic parameters conform to normal electrical signal characteristics, but the timing of its occurrence overlaps with events in the power grid operation status information that could cause abnormal electrical signal waveforms (such as lightning strikes), then the potential abnormal sequence segment is determined to be a normal fluctuation in operating conditions and is not considered a trend of insulation performance degradation. Conversely, if the characteristic parameters do not conform to normal electrical signal characteristics and do not overlap with power grid events in time, then it is determined to have a trend of insulation performance degradation and is extracted as the final abnormal sequence segment.

[0127] For extracted abnormal sequence fragments, the system calculates the average discharge amount, maximum discharge amount, and number of discharges as abnormal characteristics. Based on these characteristics, a preset risk assessment model is used to calculate the current risk assessment value. Simultaneously, the system retrieves historical risk assessment values ​​stored within the switchgear and adjusts the charge decay time constant and performance retention rate within the risk reduction rules based on the switchgear's operating time information and the current aging level of the insulation material (e.g., characterized by the rate of change of the dielectric loss factor tanδ). Substituting the adjusted parameters into the risk reduction rules, the system calculates the residual risk value remaining in the switchgear at the current time, and adds it to the current risk value to obtain the final risk assessment value.

[0128] The system sets a preset time window of 30 minutes. Within this window, it collects the start and end times of all anomalous sequence segments. If the time interval between multiple consecutive anomalous sequence segments is less than 5 minutes, they are grouped into the same anomalous cluster. The system counts the number of anomalous sequence segments within each cluster and calculates their time span to obtain the density. Based on the number and density of anomalous sequence segments, the system calculates a local deterioration rate factor for each cluster. Then, it determines the weight of each local deterioration rate factor based on the number and density of anomalous sequence segments corresponding to it, performs a weighted average of all local deterioration rate factors to obtain the global deterioration rate factor, and calculates the warning boundary accordingly.

[0129] Finally, the system compares the calculated risk assessment value with the warning boundary. If the risk assessment value exceeds the warning boundary, the system loads pre-configured reference information, which includes warning levels such as "minor warning," "moderate warning," and "serious warning," along with their corresponding amplitude ranges. The system calculates the extent to which the risk assessment value exceeds the warning boundary and compares it with the amplitude range in the reference information to determine the corresponding warning level. It then outputs a warning signal for the corresponding level of insulation performance degradation, for example, displaying "Moderate Warning: Insulation performance degradation in phase C of the switchgear, please pay attention" on the monitoring interface.

[0130] refer to Figure 2 This application provides a power supply device insulation condition fault monitoring system, the system comprising:

[0131] Signal acquisition module 1 is used to acquire the electrical signal sequence generated during the operation of the power supply device (the specific process can be referred to above).

[0132] Anomaly identification module 2 is used to screen the electrical signal sequence for anomalies and extract abnormal sequence segments with a trend of insulation performance degradation from the electrical signal sequence (the specific process can be referred to above).

[0133] Risk assessment module 3 is used to extract abnormal features that characterize the degree of insulation performance degradation from the abnormal sequence fragments, and calculate the risk assessment value of the abnormal sequence fragments based on the abnormal features (the specific process can be found above).

[0134] The early warning boundary module 4 is used to obtain the temporal characteristics of all the abnormal sequence segments that appear within a preset time window, and to obtain the early warning boundary based on the temporal characteristics (the specific process can be referred to above).

[0135] The early warning output module 5 is used to compare the risk assessment value with the early warning boundary. When the risk assessment value is greater than the early warning boundary, it outputs an insulation performance attenuation early warning signal (the specific process can be referred to above).

[0136] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for monitoring insulation status faults in a power supply device, characterized in that, include: Acquire the electrical signal sequence generated during the operation of the power supply device; Anomaly screening is performed on the electrical signal sequence to extract abnormal sequence segments with a trend of insulation performance degradation. Anomaly features characterizing the degree of insulation performance degradation are extracted from the anomalous sequence fragments, and a risk assessment value for the anomalous sequence fragments is calculated based on the anomalous features. Obtain the temporal characteristics of all the abnormal sequence fragments that appear within a preset time window, and obtain the warning boundary based on the temporal characteristics; The risk assessment value is compared with the warning boundary. When the risk assessment value is greater than the warning boundary, an insulation performance degradation warning signal is output.

2. The method for monitoring insulation status faults in a power supply device according to claim 1, characterized in that, The step of performing anomaly screening on the electrical signal sequence and extracting abnormal sequence segments with a trend of insulation performance degradation from the electrical signal sequence includes: Time-frequency domain analysis is performed on the electrical signal sequence to obtain its spectral and waveform characteristics. Based on the preset normal electrical signal characteristics, the acquired spectral characteristics and waveform characteristics are compared item by item to screen out electrical signal sequence segments whose spectral characteristics show a high proportion of high frequency components or whose waveform characteristics show abrupt changes. The screened electrical signal sequence segments are identified as potential abnormal sequence segments. Based on the comparison between the spectral and waveform characteristics of the potential abnormal sequence fragment and the characteristics of the normal electrical signal, it is determined whether the potential abnormal sequence fragment has a trend of insulation performance degradation. If the potential anomalous sequence fragment has a trend of insulation performance degradation, then the potential anomalous sequence fragment is extracted as the final anomalous sequence fragment.

3. The method for monitoring insulation status faults in a power supply device according to claim 2, characterized in that, The step of determining whether the potential abnormal sequence segment has an insulation performance degradation trend based on the comparison results of the spectral and waveform characteristics of the potential abnormal sequence segment with the characteristics of the normal electrical signal includes: Obtain the power grid operation status information of the area where the power supply device is located; The spectral and waveform characteristics of the potential abnormal sequence segments are analyzed to obtain the corresponding characteristic parameters. Based on the comparison results between the characteristic parameters and the characteristics of normal electrical signals, it is determined whether the occurrence time of the potential abnormal sequence segment coincides with the occurrence time of the event in the power grid operation status information that can cause abnormal electrical signal waveforms; When there is a timing overlap, the potential abnormal sequence segment is determined to be a normal fluctuation in operating conditions; otherwise, the potential abnormal sequence segment is determined to have a trend of insulation performance degradation.

4. The method for monitoring insulation status faults in a power supply device according to claim 1, characterized in that, The step of extracting anomalous features characterizing the degree of insulation performance degradation from the anomalous sequence fragments and calculating the risk assessment value of the anomalous sequence fragments based on the anomalous features includes: The abnormal features are extracted from the abnormal sequence fragments, and the current risk assessment value is calculated based on the abnormal features; Retrieve historical risk assessment values ​​stored in the power supply unit, and calculate the residual risk values ​​of the historical risk values ​​that remain in the power supply unit at the current time according to the preset risk reduction rules. The residual risk value is then added to the current risk value to serve as the risk assessment value for the abnormal sequence fragment.

5. A method for monitoring insulation status faults in a power supply device according to claim 4, characterized in that, The step of retrieving historical risk assessment values ​​stored in the power supply device and calculating the residual risk values ​​of the historical risk values ​​still remaining in the power supply device at the current time according to preset risk reduction rules includes: Obtain operating time information of the power supply unit and current aging indicators of the insulation material; Based on the running time information and the aging degree index, adjust the charge decay time constant used to characterize the risk decay rate and the performance retention rate used to characterize the power supply output capability within the risk reduction rule; The adjusted charge decay time constant and the performance retention rate are substituted into the risk reduction rule for calculation to obtain the residual risk value of the historical risk assessment value at the current moment.

6. The method for monitoring insulation status faults in a power supply device according to claim 1, characterized in that, The step of obtaining the temporal characteristics of all the abnormal sequence segments appearing within a preset time window, and obtaining the warning boundary based on the temporal characteristics, includes: Obtain the temporal characteristics of all the abnormal sequence segments that appear within a preset time window, and extract the start and end times of each abnormal sequence segment from each temporal characteristic; The time intervals of each abnormal sequence segment are obtained based on the start and end times. Multiple abnormal sequence segments with consecutive time intervals less than a preset time threshold are divided into the same group as an abnormal cluster group. The number of anomalous sequence fragments within each anomalous cluster is counted, and the density of the corresponding anomalous cluster is calculated based on the number of anomalous sequence fragments and the time span of the anomalous cluster. The warning boundary is calculated based on the number of abnormal sequence fragments corresponding to each of the aforementioned abnormal clusters and the density.

7. The method for monitoring insulation status faults in a power supply device according to claim 6, characterized in that, The step of calculating the warning boundary based on the number of abnormal sequence fragments corresponding to each of the abnormal clusters and the density includes: Based on the number of abnormal sequence fragments corresponding to each abnormal cluster group and the density, calculate the local deterioration rate factor corresponding to each abnormal cluster group; For each local deterioration rate factor, its weight is determined by the number and density of abnormal sequence fragments in the corresponding abnormal cluster group. The weighted operation is performed on all local deterioration rate factors to obtain the global deterioration rate factor. The warning boundary is calculated based on the global deterioration rate factor.

8. A method for monitoring insulation status faults in a power supply device according to claim 6, characterized in that, The step of calculating the warning boundary based on the number and density of the abnormal sequence fragments includes: Obtain the reference warning boundary of the power supply device; and compare the warning boundary with the reference warning boundary; When the warning boundary is smaller than the reference warning boundary, the warning boundary is adjusted to the reference warning boundary; Otherwise, based on the time difference between the current time and the end time of the most recent abnormal cluster, a boundary rebound damping coefficient characterizing the degree of rebound constraint of the warning boundary is calculated; the warning boundary is adjusted according to the boundary rebound damping coefficient, and the reference warning boundary is updated to the adjusted warning boundary.

9. A method for monitoring insulation status faults in a power supply device according to claim 1, characterized in that, The step of comparing the risk assessment value with the warning boundary, and outputting an insulation performance degradation warning signal when the risk assessment value is greater than the warning boundary, includes: Load pre-configured reference information; the reference information includes several warning levels and the amplitude range corresponding to each warning level; The magnitude by which the risk assessment value exceeds the warning boundary is calculated, and this magnitude is compared with the magnitude range in the reference information to determine the corresponding warning level, and the corresponding level of insulation performance degradation warning signal is output.

10. A power supply device insulation condition fault monitoring system, used to execute the power supply device insulation condition fault monitoring method according to any one of claims 1 to 9, characterized in that, include: The signal acquisition module is used to acquire the electrical signal sequence generated during the operation of the power supply device; An anomaly identification module is used to screen the electrical signal sequence for anomalies and extract abnormal sequence segments with a trend of insulation performance degradation from the electrical signal sequence. The risk assessment module is used to extract abnormal features characterizing the degree of insulation performance degradation from the abnormal sequence fragments, and to calculate the risk assessment value of the abnormal sequence fragments based on the abnormal features. The early warning boundary module is used to acquire the temporal characteristics of all the abnormal sequence fragments that appear within a preset time window, and to obtain the early warning boundary based on the temporal characteristics. The early warning output module is used to compare the risk assessment value with the early warning boundary. When the risk assessment value is greater than the early warning boundary, it outputs an insulation performance degradation early warning signal.

Images