Power distribution terminal alarm system and alarm method based on adaptive filtering

Abstract

A distribution terminal alarm system and method based on adaptive filtering relates to the field of distribution automation technology. To address the shortcomings of existing distribution terminal alarm systems, which rely on a single signal, have a high false alarm rate, and poor anti-interference capabilities, the system collects and fuses multi-source signals; monitors environmental noise, and adaptively and dynamically adjusts the anti-shake time; determines the fault type by matching the time domain characteristics with the frequency domain characteristics, and issues an alarm based on the confidence level of the fault. The present invention is primarily used to improve the detection accuracy of abnormal signals at distribution terminals and reduce false alarms.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution automation, and in particular to a power distribution terminal alarm system and alarm method based on adaptive filtering. Background Art

[0002] Traditional power distribution terminal alarm systems rely on single signal thresholds (e.g., overcurrent and zero current), which are susceptible to transient interference and signal jitter, resulting in high rates of missed and false alarms. Existing technologies for signal de-jittering often employ fixed-delay filtering, which cannot adapt to noise variations under varying operating conditions. Furthermore, the overall fault signal is typically synthesized using simple logic, lacking comprehensive analysis of multiple signal sources (e.g., voltage, current, and equipment status), making it difficult to accurately reflect the true fault.

[0003] Therefore, there is a need for a distribution terminal alarm system and alarm method based on adaptive filtering that can accurately identify distribution terminal faults, has a low false alarm rate, strong anti-interference ability, and a high degree of intelligence. Summary of the Invention

[0004] In order to solve the defects of the existing distribution terminal alarm system that relies on a single signal, has a high false alarm rate and poor anti-interference ability, the present invention provides a distribution terminal alarm system and alarm method based on adaptive filtering that can accurately identify distribution terminal faults, has a low false alarm rate, strong anti-interference ability and a high degree of intelligence.

[0005] The power distribution terminal alarm system based on adaptive filtering described in the present invention includes a multi-source signal fusion module, an anti-shake module and an alarm decision module;

[0006] The multi-source signal fusion module is used to collect multi-source signals, extract time domain features and frequency domain features of the multi-source signals; and fuse the collected time domain features and frequency domain features of the multi-source signals;

[0007] The anti-shake module is used to monitor the environmental noise and adaptively and dynamically adjust the anti-shake time;

[0008] The alarm decision module is used to determine the fault type by matching the time domain features and the frequency domain features, and to issue an alarm based on the confidence level of the fault.

[0009] Furthermore: the multi-source signal fusion module includes a signal acquisition module, a feature extraction module and a fusion model;

[0010] The signal acquisition module is used to collect multi-source signals, and the multi-source signals include current, voltage, switch position and device temperature;

[0011] The feature extraction module is used to extract time domain features and frequency domain features of multi-source signals;

[0012] The fusion model is used to dynamically allocate the weights of the time domain features and frequency domain features of each multi-source signal in the overall accident synthesis according to the signal type and historical data credibility of the multi-source signals, and to fuse the time domain features and frequency domain features of the multi-source signals according to the weights.

[0013] Furthermore: the anti-shake module includes an environmental noise monitoring module and a filtering module;

[0014] The environmental noise monitoring module is used to calculate the noise intensity in real time by sampling the background noise signal and dynamically adjust the anti-shake time;

[0015] The filtering module is used to suppress the interference of instantaneous pulses through the photoelectric isolation module and Kalman filtering.

[0016] Furthermore: the alarm decision module includes a fault pattern matching module, a confidence assessment module and a recording data self-diagnosis module;

[0017] The fault pattern matching module is used to determine the fault type through feature matching based on the built-in typical fault pattern library;

[0018] The confidence evaluation module is used to calculate the alarm confidence of the fault based on signal consistency and duration, and trigger an early warning when the alarm confidence is lower than a preset threshold;

[0019] The recorded data self-diagnosis module is used to automatically select the recording length according to the signal mutation rate corresponding to the early warning, compare the recorded data with the data in the historical waveform library, identify hidden abnormal waveforms, and generate alarm information.

[0020] The power distribution terminal alarm method based on adaptive filtering disclosed in the present invention adopts the power distribution terminal alarm system based on adaptive filtering, and the specific steps are as follows:

[0021] S1. Collect multi-source signals, extract time domain features and frequency domain features of multi-source signals; and fuse multi-source signals;

[0022] S2. Monitor the ambient noise and dynamically adjust the anti-shake time adaptively;

[0023] S3. Determine the fault type by matching the time domain features and the frequency domain features, and issue an alarm based on the confidence level of the fault.

[0024] Furthermore: in S1, the fusing of multi-source signals comprises the following steps:

[0025] S11, collecting multi-source signals and extracting time domain features and frequency domain features of the multi-source signals;

[0026] S12. Dynamically assigning weights of the time domain features and frequency domain features of each multi-source signal in the accident synthesis based on the signal type and historical data credibility of the multi-source signal; if the fluctuation of the time domain features or frequency domain features of the current multi-source signal exceeds a preset threshold, executing S13; if the fluctuation of the time domain features or frequency domain features of the current multi-source signal does not exceed the threshold, executing S14;

[0027] S13, modifying the weight ratio of the current multi-source signal according to the deviation and credibility exceeding the preset threshold; fusing the time domain features and frequency domain features of the multi-source signal according to the weight;

[0028] S14, repeating S11-S13 to re-collect multi-source signals;

[0029] S15. If the fluctuation range of the multi-source signal received by the system is less than the fixed threshold value compared with the signal received last time, it is determined to be state hold, the CPU maintains the current operating state, and maintains the numerical state of each signal collected last time, and repeats S11-S13 to continue collecting multi-source signals; if the exit condition is met, the collection of multi-source signals ends.

[0030] Further: in S2, the adaptive dynamic adjustment of the anti-shake time includes the following steps:

[0031] S21, collecting noise signals from the background environment;

[0032] S22: Monitor the noise signal and calculate the noise intensity. If the noise intensity exceeds a preset threshold, execute S23; if the noise intensity does not exceed the preset threshold, continue to use the anti-shake time calculated last time.

[0033] S23, dynamically adjust the anti-shake time. If the anti-shake time adjustment is successful, update the determination time of the signal state recognition. If the anti-shake time adjustment fails, repeat S21-S22 to continue collecting the noise signal of the background environment;

[0034] S24. If the adaptive filtering function remains enabled, then S21-S23 are repeatedly executed to continue collecting multi-source signals; if the exit condition is met, then the collection of background environment noise signals ends.

[0035] Further: in S3, the alarming according to the confidence level of the fault includes the following steps:

[0036] S31, collecting recorded wave data and preprocessing the recorded wave data;

[0037] S32, comparing the recorded waveform data with the data in the historical waveform library to identify hidden abnormal waveforms; if the waveform is abnormal, executing S33; if the waveform is normal, repeating S31-S32 to continue collecting multi-source signal recorded waveform data;

[0038] S33, generating alarm information according to the hidden abnormal waveform, lighting an alarm indicator light or performing a self-healing operation;

[0039] S34. If the system self-recovers successfully or the alarm information is manually reset, repeat S31-S33 to continue collecting multi-source signal recording data; if self-recovery fails, an alarm is issued.

[0040] The beneficial effects of the present invention are:

[0041] The adaptive filtering-based power distribution terminal alarm system achieves high-precision fault detection and a low false alarm rate by dynamically adjusting anti-shake parameters and integrating multi-dimensional signal characteristics. This invention is mainly used to improve the detection accuracy of abnormal signals at power distribution terminals and reduce false alarms. The invention has the following advantages:

[0042] High precision: Multi-source signal fusion reduces single signal misjudgment. As shown in the implementation example, it can increase the fault identification accuracy to over 99%.

[0043] Low false alarms and missed alarms: The adaptive anti-shake algorithm reduces noise interference. As described in the embodiment, it can reduce the missed alarm rate by more than 70% and the false alarm rate by more than 60%.

[0044] Intelligence: Fault pattern matching and confidence assessment support graded alarms (early warning / serious alarm).

[0045] Compatibility: Supports hardware modification of existing power distribution terminals. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a system architecture diagram;

[0047] Figure 2 It is a schematic diagram of multi-source signal weight distribution;

[0048] Figure 3 It is the flow chart of the adaptive anti-shake algorithm;

[0049] Figure 4 It is the self-diagnosis flow chart of recorded data. DETAILED DESCRIPTION

[0050] The following are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or replacements that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the scope of protection of the present invention. The embodiments described below are only used to explain the present invention and cannot be interpreted as limiting the present invention. The scope of protection of the present invention should be based on the scope of protection of the claims. The embodiments of the present invention are described in detail below. In order to facilitate the description of the present invention and simplify the description, the technical terms used in the description of the present invention should be interpreted broadly, including but not limited to conventional replacement schemes not mentioned in this application, and also including direct implementation and indirect implementation.

[0051] Example 1

[0052] Combine Figures 1-4 This embodiment describes a power distribution terminal alarm system based on adaptive filtering disclosed in this embodiment, which includes a multi-source signal fusion module, an anti-shake module, and an alarm decision module;

[0053] The multi-source signal fusion module is used to collect multi-source signals, extract time domain features and frequency domain features of the multi-source signals; and fuse the collected time domain features and frequency domain features of the multi-source signals;

[0054] The anti-shake module is used to monitor the environmental noise and adaptively and dynamically adjust the anti-shake time;

[0055] The alarm decision module is used to determine the fault type by matching the time domain features and the frequency domain features, and to issue an alarm based on the confidence level of the fault.

[0056] The multi-source signal fusion module includes a signal acquisition module, a feature extraction module and a fusion model; Figure 1 As shown in the figure, the signal acquisition, fusion model, anti-shake module and alarm decision process are demonstrated.

[0057] The signal acquisition module is used to collect multi-source signals, including current, voltage, switch position and equipment temperature. The signal acquisition module adopts a multi-channel signal acquisition module. Taking the station power distribution terminal as an example, the main control chip is T3, with a built-in double-precision floating-point unit, responsible for signal preprocessing. The 16-bit 16-channel synchronous sampling ADC chip model AD7616BSTZ is used to transmit the exchange data to the main control chip through the SPI interface. The following input channels are configured:

[0058] Current signal: 4 channels (A / B / C / 0 phase), connected through the secondary side of CT, range 0~5A, sampling rate 12.8kHz;

[0059] Voltage signal: 4 channels (A / B / C / 0 phase), input after resistor voltage division, range 0-100V, sampling rate 12.8kHz;

[0060] Switching input: 9 channels (circuit breaker position, grounding switch status, etc.), optocoupler isolation input;

[0061] Temperature signal: 2 PT100 sensors, connected via module communication;

[0062] The feature extraction module is used to extract time domain features (such as amplitude and slope) and frequency domain (such as harmonic content) features of multi-source signals;

[0063] Time domain characteristics: Sliding window calculation (window length 200ms), current effective value:

[0064] ;

[0065] in, is the effective value of current, N is the harmonic number, is the effective value of the nth harmonic current.

[0066] Voltage mutation rate:

[0067] ;

[0068] in, is the voltage mutation rate (voltage change per unit time), is the voltage value of the tth sampling, is the voltage sampling value at time t-1, is the time interval.

[0069] Frequency domain features: FFT (Fast Fourier Transform) analyzes the 0-2kHz spectrum and calculates the THD (Total Harmonic Distortion) value;

[0070] The fusion model is used to dynamically assign weights to the time-domain and frequency-domain features of each multi-source signal in the overall accident synthesis based on the signal type and historical data credibility of the multi-source signal, and to fuse the time-domain and frequency-domain features of the multi-source signal based on the weights. Based on the signal type and historical data credibility, the weight of each signal in the overall accident synthesis is dynamically assigned (e.g., 50% weight for current signals and 20% weight for temperature signals).

[0071] Figure 2 It is a schematic diagram of multi-source signal weight distribution, showing the dynamic weights of signals such as current, voltage, and temperature.

[0072] Dynamic weight allocation: Establish a historical credibility database and count the number of false alarms for each signal in the past 30 days;

[0073] Weight The calculation formula is:

[0074] ;

[0075] in, is the number of false positives, Score the significance of the current signal mutation, Statistical signal mutation significance score for the past 30 days;

[0076] The anti-shake module includes an environmental noise monitoring module and a filtering module;

[0077] The environmental noise monitoring module is used to calculate the noise intensity in real time by sampling the background noise signal (such as the electromagnetic interference spectrum) and dynamically adjust the anti-shake time (adjustable from 10ms to 1000ms);

[0078] Figure 3 It is a flow chart of the adaptive anti-shake algorithm, including noise monitoring and anti-shake time dynamic adjustment logic.

[0079] The hardware circuit of the anti-shake module is to set double isolation on the switch input channel:

[0080] The front stage uses a high-speed optocoupler model HCPL181 (transmission delay <3μs), and the back stage is equipped with an RC low-pass filter (cut-off frequency 1kHz adjustable);

[0081] Configure the electromagnetic interference monitoring module: The electromagnetic interference monitoring module collects noise in the 0.1GHz-6GHz frequency band in real time and transmits the noise power spectrum data to the main control chip through the communication interface.

[0082] The filtering module is used to suppress the interference of instantaneous pulses through the photoelectric isolation module and Kalman filtering.

[0083] Dynamic adjustment of the parameters of the Kalman filter:

[0084] Process noise covariance Adaptive to the signal rate of change:

[0085] ;

[0086] in, .

[0087] Table 1: Noise intensity classification table;

[0088]

[0089] The alarm decision module is used to determine the fault type by matching the time domain features and the frequency domain features, and to issue an alarm based on the confidence level of the fault.

[0090] Table 2: Example of a typical fault signature library;

[0091]

[0092] Calculation model of confidence C:

[0093] ;

[0094] in, is the time decay factor (default 0.1), is the time interval, Calculate signal strength for decision-making confidence, The weight of the signal participating in the confidence calculation.

[0095] The alarm decision module includes a fault pattern matching module, a confidence assessment module and a recording data self-diagnosis module;

[0096] The fault mode matching module is used to determine the fault type through feature matching based on a built-in typical fault mode library (such as short circuit, grounding, and equipment overheating);

[0097] The confidence evaluation module is used to calculate the alarm confidence of the fault based on signal consistency and duration, and trigger an early warning when the alarm confidence is lower than a threshold;

[0098] The waveform data self-diagnosis module automatically selects the recording length (first 4 cycles + last 8 cycles or a custom one) based on the signal mutation rate corresponding to the warning, and supports Comtrade format storage. It compares the recorded data with a historical waveform library to identify hidden anomalies (such as early signs of insulation degradation) and issues an alarm for waveform signals indicating self-healing failure.

[0099] Figure 4 It is a flow chart for self-diagnosis of recorded wave data, including waveform comparison and latent anomaly identification steps.

[0100] The recorded data is stored in the recording storage unit, which is an external EMMC (embedded multimedia card) for waveform storage, supports Comtrade2013 standard format, and is configured with two recording modes:

[0101] Conventional recording: first 4 cycles (80ms) + last 8 cycles (160ms);

[0102] Long-time recording: User-defined duration (maximum 60s).

[0103] Example 2

[0104] This embodiment is described in conjunction with Example 1. The power distribution terminal alarm method based on adaptive filtering disclosed in this embodiment adopts the power distribution terminal alarm system based on adaptive filtering. The specific steps are as follows:

[0105] S1, collecting multi-source signals, extracting time domain features and frequency domain features of the multi-source signals; fusing the multi-source signals; including the following steps:

[0106] S11, collecting multi-source signals and extracting time domain features and frequency domain features of the multi-source signals;

[0107] S12. Dynamically assigning weights of the time domain features and frequency domain features of each multi-source signal in the accident synthesis based on the signal type and historical data credibility of the multi-source signal; if the fluctuation of the time domain features or frequency domain features of the current multi-source signal exceeds a preset threshold, executing S13; if the fluctuation of the time domain features or frequency domain features of the current multi-source signal does not exceed the threshold, executing S14;

[0108] S13, modifying the weight ratio of the current multi-source signal according to the deviation and credibility exceeding the preset threshold; fusing the time domain features and frequency domain features of the multi-source signal according to the weight;

[0109] S14, repeating S11-S13 to re-collect multi-source signals;

[0110] S15. If the fluctuation range of the multi-source signal received by the system is less than the fixed threshold value compared with the signal received last time, it is determined to be state hold, the CPU maintains the current operating state, and maintains the numerical state of each signal collected last time, and repeats S11-S13 to continue collecting multi-source signals; if the exit condition is met, the collection of multi-source signals ends.

[0111] S2. Monitor the ambient noise and dynamically adjust the anti-shake time adaptively;

[0112] The adaptive dynamic adjustment of the anti-shake time comprises the following steps:

[0113] S21, collecting noise signals from the background environment;

[0114] S22: Monitor the noise signal and calculate the noise intensity. If the noise intensity exceeds a threshold, execute S23; if the noise intensity does not exceed the threshold, continue to use the anti-shake time calculated last time.

[0115] S23, dynamically adjust the anti-shake time. If the anti-shake time adjustment is successful, update the determination time of the signal state recognition. If the anti-shake time adjustment fails, repeat S21-S22 to continue collecting the noise signal of the background environment;

[0116] S24. If the adaptive filtering function remains enabled, then S21-S23 are repeatedly executed to continue collecting multi-source signals; if the exit condition is met, then the collection of background environment noise signals ends.

[0117] S3, determining the fault type by matching the time domain features and the frequency domain features, and issuing an alarm based on the confidence level of the fault. The alarming based on the confidence level of the fault includes the following steps:

[0118] S31, collecting recorded wave data and preprocessing the recorded wave data;

[0119] S32, comparing the recorded waveform data with the data in the historical waveform library to identify hidden abnormal waveforms; if the waveform is abnormal, executing S33; if the waveform is normal, repeating S31-S32 to continue collecting multi-source signal recorded waveform data;

[0120] For example, when the connector temperature curve rises, harmonic current increases, and voltage fluctuations increase, each waveform may appear to be within the normal operating range when viewed individually, but when combined together, it can be inferred that the cable stud is damaged. This type of waveform that is actually operating normally but is analyzed as abnormal when combined individually is defined as an invisible abnormal waveform.

[0121] S33, generating alarm information according to the hidden abnormal waveform, lighting an alarm indicator light or performing a self-healing operation;

[0122] The self-healing operation includes tripping and isolating, disconnecting abnormal lines, or switching to a backup power supply according to a preset logic;

[0123] S34. If the system self-recovers successfully or the alarm information is manually reset, repeat S31-S33 to continue collecting multi-source signal recording data; if self-recovery fails, an alarm is issued.

[0124] Example 3

[0125] This embodiment is described in conjunction with Example 2. The adaptive filtering-based distribution terminal alarm method disclosed in this embodiment takes a K-type station (switch station) distribution automation terminal as an example:

[0126] Hardware deployment: Integrate a multi-channel signal acquisition module into the terminal main control plug-in and connect it to the current transformer, voltage sensor, and temperature probe.

[0127] Algorithm configuration: Set signal weight parameters and select fault mode library (such as cable station and overhead line dedicated library) through the LCD screen menu.

[0128] Operation process:

[0129] Collect signals and extract features in real time to calculate the intensity of ambient noise;

[0130] Dynamically adjust the anti-shake time to filter out instantaneous interference;

[0131] After multi-source signal fusion, the fault mode is matched, the confidence level is evaluated and a graded alarm is triggered;

[0132] In case of abnormality, the waveform is automatically recorded and stored for subsequent analysis.

[0133] Take the cable branch box overload warning as an example;

[0134] Scenario: The temperature of cable joints in an industrial park rises abnormally;

[0135] System Response:

[0136] 15:00:00 The temperature sensor detected that the joint temperature rose from 45°C to 58°C (temperature rise rate 13K / h);

[0137] 15:00:05 The current harmonic content increased from 5% to 18%, and the voltage fluctuation rate increased to 2% / s;

[0138] Fusion computing:

[0139] Temperature weight is 30%, current weight is 40%, and voltage weight is 30%;

[0140] Confidence level C = 0.78 (warning threshold 0.7);

[0141] Triggering the second-level warning, recording the waveform for 10 seconds and uploading it to the main station;

[0142] Result: The operation and maintenance personnel found that the connector was oxidized during on-site inspection, thus avoiding the cable burning accident.

[0143] Take the overhead line lightning interference processing as an example;

[0144] Scenario: Thunderstorm weather causes instantaneous pulse interference;

[0145] System Response:

[0146] The adaptive anti-shake module detected that the noise in the 6GHz band reached -95dBm / Hz;

[0147] Automatically adjust the anti-shake time to 80ms;

[0148] The current transient pulse (amplitude 8In, duration 2ms) is suppressed by double filtering;

[0149] No false alarms were triggered, only event logs were recorded;

[0150] Comparison: Traditional fixed 50ms anti-shake systems produce false alarms in this scenario.

[0151] Parameter configuration method: Enter the "Advanced Settings → Signal Weight" menu through the terminal maintenance software or LCD;

[0152] Table 3: Three preset modes;

[0153]

[0154] Supports user customization, and the sum of all weights must meet the 100% constraint.

[0155] Performance test data: In a provincial power grid pilot project, 10 sites were tested comparatively:

[0156] Table 4: Performance test data statistics

[0157]

[0158] Test results show that the system has significantly improved key performance indicators, especially in high-noise environments (areas where electromagnetic interference exceeds standards in distribution substations), with the false alarm rate reduced by 72%.

[0159] The present invention is also applicable to scenarios such as new energy stations and rail transit power supply systems. It can adapt to different application environments by simply adjusting the characteristic thresholds in the fault mode library.

Claims

1. The power distribution terminal alarm system based on multi-source signal fusion and adaptive filtering is characterized by: Including multi-source signal fusion module, anti-shake module and alarm decision module; The multi-source signal fusion module is used to collect multi-source signals and extract time domain features and frequency domain features of the multi-source signals; Dynamically assign weights to the time domain and frequency domain features of each multi-source signal in the overall accident synthesis based on the signal type and historical data credibility of the multi-source signal. If the fluctuation of the time domain or frequency domain features of the current multi-source signal exceeds the preset threshold, modify the weight ratio of the current multi-source signal based on the deviation and credibility exceeding the preset threshold. Fuse the collected time domain and frequency domain features of the multi-source signal. The anti-shake module is used to monitor the environmental noise and adaptively and dynamically adjust the anti-shake time; The alarm decision module is used to determine the fault type by matching the time domain features and the frequency domain features, and to issue an alarm based on the confidence level of the fault; The alarm decision module includes a confidence assessment module and a wave recording data self-diagnosis module; The confidence evaluation module is used to calculate the alarm confidence of the fault based on signal consistency and duration, and trigger an early warning when the alarm confidence is lower than a preset threshold; The recorded data self-diagnosis module is used to automatically select the recording length according to the signal mutation rate corresponding to the warning, compare the recorded data with the data in the historical waveform library, identify hidden abnormal waveforms, and generate alarm information; The multi-source signal fusion module includes a signal acquisition module, a feature extraction module and a fusion model; The signal acquisition module is used to collect multi-source signals, and the multi-source signals include current, voltage, switch position and device temperature; The feature extraction module is used to extract time domain features and frequency domain features of multi-source signals; The fusion model is used to dynamically allocate the weights of the time domain features and frequency domain features of each multi-source signal in the overall accident synthesis according to the signal type of the multi-source signal and the credibility of historical data, and to fuse the time domain features and frequency domain features of the multi-source signal according to the weights.

2. The power distribution terminal alarm system based on multi-source signal fusion and adaptive filtering according to claim 1 is characterized in that: The anti-shake module includes an environmental noise monitoring module and a filtering module; The environmental noise monitoring module is used to calculate the noise intensity in real time by sampling the background noise signal and dynamically adjust the anti-shake time; The filtering module is used to suppress the interference of instantaneous pulses through the photoelectric isolation module and Kalman filtering.

3. The power distribution terminal alarm system based on multi-source signal fusion and adaptive filtering according to claim 1 is characterized in that: The alarm decision module includes a fault pattern matching module; The fault pattern matching module is used to determine the fault type through feature matching based on the built-in typical fault pattern library.

4. A power distribution terminal alarm method based on multi-source signal fusion and adaptive filtering is characterized in that: The distribution terminal alarm system based on multi-source signal fusion and adaptive filtering as described in any one of claims 1 to 3 is adopted, and the specific steps are as follows: S1. Collect multi-source signals and extract time domain features and frequency domain features of the multi-source signals; Fusion of multi-source signals; S2. Monitor the ambient noise and dynamically adjust the anti-shake time adaptively; S3. Determine the fault type by matching the time domain features and the frequency domain features, and issue an alarm based on the confidence level of the fault.

5. The power distribution terminal alarm method based on multi-source signal fusion and adaptive filtering according to claim 4 is characterized in that: In S1, the fusion of multi-source signals includes the following steps: S11, collecting multi-source signals and extracting time domain features and frequency domain features of the multi-source signals; S12. Dynamically assigning weights of the time domain features and frequency domain features of each multi-source signal in the accident synthesis based on the signal type and historical data credibility of the multi-source signal; if the fluctuation of the time domain features or frequency domain features of the current multi-source signal exceeds a preset threshold, executing S13; if the fluctuation of the time domain features or frequency domain features of the current multi-source signal does not exceed the threshold, executing S14; S13, modifying the weight ratio of the current multi-source signal according to the deviation and credibility exceeding the preset threshold; fusing the time domain features and frequency domain features of the multi-source signal according to the weight; S14, repeating S11-S13 to re-collect multi-source signals; S15. If the fluctuation range of the multi-source signal received by the system is less than the fixed threshold value compared with the signal received last time, it is determined to be state hold, the CPU maintains the current operating state, and maintains the numerical state of each signal collected last time, and repeats S11-S13 to continue collecting multi-source signals; if the exit condition is met, the collection of multi-source signals ends.

6. The power distribution terminal alarm method based on multi-source signal fusion and adaptive filtering according to claim 4 is characterized in that: In S2, the adaptive dynamic adjustment of the anti-shake time includes the following steps: S21, collecting noise signals from the background environment; S22: Monitor the noise signal and calculate the noise intensity. If the noise intensity exceeds a preset threshold, execute S23; if the noise intensity does not exceed the preset threshold, continue to use the anti-shake time calculated last time. S23, dynamically adjust the anti-shake time. If the anti-shake time adjustment is successful, update the determination time of the signal state recognition. If the anti-shake time adjustment fails, repeat S21-S22 to continue collecting the noise signal of the background environment; S24. If the adaptive filtering function remains enabled, then S21-S23 are repeatedly executed to continue collecting multi-source signals; if the exit condition is met, then the collection of background environment noise signals ends.

7. The power distribution terminal alarm method based on multi-source signal fusion and adaptive filtering according to claim 4 is characterized in that: In S3, the alarming according to the confidence level of the fault includes the following steps: S31, collecting recorded wave data and preprocessing the recorded wave data; S32, comparing the recorded waveform data with the data in the historical waveform library to identify hidden abnormal waveforms; if the waveform is abnormal, executing S33; if the waveform is normal, repeating S31-S32 to continue collecting multi-source signal recorded waveform data; S33, generating alarm information according to the hidden abnormal waveform, lighting an alarm indicator light or performing a self-healing operation; S34. If the system self-recovers successfully or the alarm information is manually reset, repeat S31-S33 to continue collecting multi-source signal recording data; if self-recovery fails, an alarm is issued.

Images