Contact mechanical wear and fault diagnosis system

Abstract

The present invention relates to the field of fault diagnosis technology, specifically to a contact mechanical wear and fault diagnosis system, which includes a data acquisition module, a wear diagnosis module, a contact surface detection module, a buffer determination module, and a composite diagnosis module. In the present invention, the three-phase opening time series, the instantaneous value of resistance, the pressure signal and the shock wave envelope are aligned and fused at the millisecond time stamp level to improve the multi-source data temporal consistency and linkage feature recognition capabilities, and the statistical correlation between the operation time and the resistance difference is calculated by covariance to enhance the dynamic judgment accuracy of the wear process. Based on the pressure quadrant division and gradient ratio extraction, the spatial positioning characteristics of the contact surface depression are clarified, the ratio of the shock wave change rate to the duration is extracted, the sensitive identification of the buffer performance change is enhanced, and the integration of multiple types of diagnostic results forms an overlapping analysis path to achieve accurate identification and logical attribution of composite fault types.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a contact mechanical wear and fault diagnosis system. Background Art

[0002] The field of fault diagnosis technology encompasses the core technology system for the full lifecycle health management of power equipment, encompassing mechanical structural integrity assessment, research into electrical contact performance degradation mechanisms, dynamic behavior feature extraction, and system reliability prediction methods. This field, based on multi-physics field coupling analysis, utilizes heterogeneous sensor networks such as distributed fiber optic temperature measurement devices, Hall effect current sensors, and vibration accelerometers to construct a multi-dimensional data acquisition system encompassing dynamic contact resistance monitoring, mechanical vibration spectrum analysis, and action timing parameter recording. This complete technology chain encompasses signal noise reduction processing, feature parameter standardization, abnormal pattern library construction, and fault tree inference models.

[0003] The contact mechanical wear and fault diagnosis system is a switchgear condition assessment system based on the correlation analysis of mechanical motion characteristics and electrical parameters. This technology uses a laser displacement sensor to capture the contact opening and closing displacement trajectory curves, and calculates the overtravel time characteristic through stroke velocity differential calculation. A broadband vibration sensor is used to collect the mechanism's motion impact waveform, and a time-domain integration algorithm is used to extract the vibration energy distribution parameters. Four-wire contact resistance measurement technology is used to eliminate lead errors and establish a nonlinear mapping model between contact pressure and resistance. Based on the monitoring data of the opening rebound amplitude, a kinematic inversion method is used to reconstruct the wear state of the transmission connecting rod.

[0004] Existing technologies primarily rely on single-shot sampling for data fusion and lack a unified time alignment mechanism, making it difficult to synchronously analyze information from different sensor channels and limiting the overall restoration of the causal chain of events. In state determination, independent parameters are used as the basis for judgment, and no statistical correlation is formed between indicators. This makes it easy to react slowly to covariation trends in multi-source information, reducing the sensitivity to gradual faults. In contact surface anomaly detection, distributed quadrant division and local gradient indicators are not used, making it impossible to clearly define the boundaries and evolution paths of abnormal areas. In impact feature analysis, although waveform signals are obtained, the ratio of the rate of change to the duration is not introduced, making it difficult to distinguish whether signal intensity fluctuations originate from the buffer attenuation process. In fault type identification, there is a lack of spatial mapping and benchmark comparison between various results, and diagnostic conclusions are mostly based on local data fragments, resulting in insufficient recognition coverage of complex structural anomalies, high misdiagnosis rates, weak correlation, and other problems, limiting the effective control of the global characteristics of the equipment status. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a contact mechanical wear and fault diagnosis system.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: the contact mechanical wear and fault diagnosis system includes:

[0007] The data acquisition module acquires the three-phase opening action time series, the instantaneous value of the contact resistance, the annular array pressure sensor channel signal, and the vibration shock wave envelope waveform, aligns and stores them by timestamp, and generates a multi-source spatiotemporal dataset.

[0008] The wear diagnosis module calls the multi-source spatiotemporal data set, calculates the standard deviation of the operation time, extracts the difference between the contact resistance value and the reference value, calculates the covariance between the two, and generates a contact wear status diagnosis result when the covariance continuously exceeds a threshold;

[0009] The contact surface detection module divides the pressure array data into four detection quadrants based on the multi-source spatiotemporal data set, calculates the ratio of the pressure difference between adjacent sensors to the spacing, and generates the coordinates of the concave area when the gradient in the quadrant continuously exceeds the threshold;

[0010] The buffer determination module intercepts the shock wave envelope waveform based on the multi-source spatiotemporal data set, calculates the ratio of the voltage rising edge change rate to the waveform time span, and generates a buffer performance attenuation determination result;

[0011] The composite diagnostic module is based on the spatial mapping model of the contact wear status diagnosis results and the recessed area coordinates, combined with the time series analysis of the buffer efficiency attenuation judgment results. When the diagnostic results are triggered simultaneously, a conclusion on the type of mechanical fault is generated based on the overlapping area and the calibration parameters.

[0012] As a further solution of the present invention, the multi-source spatiotemporal data set includes a tripping time series, instantaneous resistance value, pressure array signal, and shock wave envelope waveform; the contact wear status diagnosis result includes a time volatility index, a resistance difference index, and a covariance threshold judgment result; the recessed area coordinates include a detection quadrant number, a pressure gradient ratio, and abnormal location information; the buffer efficiency attenuation judgment result includes a voltage change rate, an impact duration, and an efficiency ratio comparison result; the mechanical fault type conclusion includes a wear status result, a recessed space characteristic, and a buffer performance judgment value.

[0013] As a further solution of the present invention, the data acquisition module includes:

[0014] The three-phase trip time submodule extracts the start and end times of the three-phase trip action time series data, obtains the duration of each phase action, calculates the time difference between the three phases, the average time difference, and the time interval between the trip phases, and generates the trip time difference interval;

[0015] The pressure signal decoding submodule calls the opening time difference interval, reads the annular array pressure sensor channel signal, synchronously compares the differential azimuth pressure values ​​at the same time according to the channel number, determines the instantaneous pressure change rate, and selects the channels whose change rate exceeds the pressure change rate threshold, obtains the instantaneous pressure extreme point and corresponding time in the selected channel, and generates a high-variability pressure peak rate;

[0016] The shock wave processing submodule calls the high-variability pressure peak rate, captures the vibration shock wave envelope waveform, divides the envelope waveform into energy segments based on timestamps, extracts the envelope peak, mean, and fluctuation within the segment, and classifies the fluctuation amplitude levels in differentiated time periods through a combined comparison between the fluctuation and the pressure peak rate, obtains the distribution characteristics corresponding to the shock response data, and establishes a multi-source spatiotemporal dataset.

[0017] As a further solution of the present invention, the wear diagnosis module includes:

[0018] The trip time series extraction submodule extracts the trip action information corresponding to the time based on the multi-source spatiotemporal dataset, identifies the time record points, and establishes a continuous trip time series. When extracting data, the records are arranged and filtered in time series, and data with abnormal time marks and missing records are eliminated to generate a time series trip data column.

[0019] The operation time fluctuation calculation submodule calculates the difference of each operation time based on the time-series tripping data column, obtains the operation time series, calls the series to calculate the standard deviation, uses the sliding window method to estimate the variance within a certain time interval and performs square root processing to generate the operation time standard deviation series;

[0020] The contact resistance covariance analysis submodule calls the operation time standard deviation sequence, obtains the operation time standard deviation value at the current moment, detects the contact resistance measurement value at the corresponding moment and the preset contact resistance reference value, calculates the absolute difference, and constructs a covariance function model with the current standard deviation value using the formula:

[0021] ;

[0022] Calculate and obtain the current covariance value, compare the current covariance value in a continuous period with the threshold, determine whether it is continuously exceeded, obtain the continuous trigger state interval, and generate the contact wear state diagnosis result;

[0023] in, represents the current covariance value, Representative The contact resistance value at the moment, Representative The contact resistance reference value at the moment, Representative The standard deviation of the operation time at each moment, is the mean of the standard deviation series of operation time, is the sample size.

[0024] As a further solution of the present invention, the contact surface detection module includes:

[0025] The quadrant division submodule obtains the pressure array data from the multi-source spatiotemporal data set, constructs a two-dimensional coordinate distribution map based on the sensing spatial position, divides the entire detection area with the center point as the origin, and evenly divides it into four quadrant areas, marked as quadrants 1 to 4 respectively, extracts the corresponding sensing data spatial coordinates within the quadrant, and establishes a quadrant coordinate set;

[0026] The pressure gradient calculation submodule calls the coordinate set in the quadrant, selects two adjacent points in the quadrant as a unit group, performs difference calculation on the pressure values ​​between each set of sensor information, and calculates the distance between the corresponding sensor coordinates using the formula:

[0027] ;

[0028] Obtain the gradient value within the quadrant by calculation, compare the continuous gradient value with the set gradient threshold, retain the coordinate points corresponding to the continuous exceeding area, and generate the continuous exceeding area;

[0029] in, Indicates sensor and The gradient value between 、 are the pressure values ​​of the two sensors respectively, 、 、 、 is the spatial coordinate, For quadrant The average pressure value of all sensors in the system;

[0030] The sunken area identification submodule determines whether the number of continuous point sets exceeds the ratio threshold of the number of sensors in the quadrant based on the continuous over-limit area. If the condition is met, all coordinate point sets in the continuous area are selected, the boundary range is extracted to construct a closed space, the center point of the area is calculated and the spatial coordinates are marked to obtain the coordinates of the sunken area.

[0031] As a further solution of the present invention, the buffer determination module includes:

[0032] The waveform interception submodule obtains the shock wave envelope waveform at the start of the tripping phase in the multi-source spatiotemporal dataset, extracts the waveform segment between the start point and the end point of each waveform according to the trigger time window, removes segments with insufficient signal fluctuation amplitude, and performs boundary alignment to obtain the tripping waveform segment;

[0033] The voltage rate calculation submodule calls the trip waveform segment, identifies the voltage value sequence corresponding to the time point, locates the starting and ending positions of the voltage rising segment, calculates the ratio of the voltage change amplitude to the corresponding time interval, and records the start and end time difference of the entire segment waveform, using the formula:

[0034] ;

[0035] Calculate the rising amplitude characteristic value, take the coupled change of the waveform rising rate and time span as the quantization result, and generate the amplitude change result;

[0036] in, represents the characteristic value of the rising amplitude, Indicates the waveform The voltage change rate of the point, Indicates the time interval between the start and end of the waveform. Indicates the Point voltage value, represents the mean of all voltage values, Indicates the total number of waveform data points;

[0037] The performance evaluation submodule calls the buffer performance benchmark value provided by the equipment based on the amplitude change results, performs ratio judgment, compares the characteristic value with the benchmark value, marks an abnormality if the difference exceeds the limit continuously, collects the corresponding time point and stores the abnormal characteristic vector, and obtains the buffer performance attenuation judgment result.

[0038] As a further solution of the present invention, the composite diagnostic module includes:

[0039] The state mapping submodule obtains the contact wear state diagnosis results and the coordinates of the recessed area, matches the spatial coordinate point set of the recessed area with the corresponding wear state time node, selects areas with overlapping distribution within the same time period and records geometric contour information, establishes a dual index structure of space and time, and generates a spatial mapping relationship set;

[0040] The benchmark calibration submodule calls the buffer performance attenuation determination result, extracts the time series information corresponding to the determination, identifies the starting time point and duration of the diagnostic event, matches it with the time segments of the area in the spatial mapping relationship concentration, extracts the overlapping intervals, and updates the indicator value range according to the internal standard calibration parameters to obtain the calibration benchmark parameter set;

[0041] The fault judgment submodule identifies the area value of the spatial overlapping area within the diagnosis time based on the spatial mapping relationship set and the calibration reference parameter set, combines and compares the calibration parameters associated with the area, couples and screens the area index of the spatial area with the corresponding calibration parameter set, and obtains the conclusion of the mechanical fault type when the parameter combination relationship meets the abnormal judgment condition.

[0042] Compared with the prior art, the advantages and positive effects of the present invention are:

[0043] In the present invention, the three-phase opening time series, instantaneous resistance value, pressure signal and shock wave envelope are aligned and fused according to millisecond timestamps to improve the temporal consistency of multi-source data and the linkage feature recognition capability. The statistical correlation between the operation time and the resistance difference is calculated by covariance to enhance the dynamic judgment accuracy of the wear process. Based on the pressure quadrant division and gradient ratio extraction, the spatial positioning characteristics of the contact surface depression are clarified, the ratio of the shock wave change rate to the duration is extracted, and the sensitive recognition of the buffering performance change is enhanced. The multiple diagnostic results are integrated to form an overlapping analysis path to achieve accurate identification and logical attribution of complex fault types. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a system flow chart of the present invention;

[0045] Figure 2 This is a flow chart of the data acquisition module of the present invention;

[0046] Figure 3 This is a flow chart of the wear diagnosis module of the present invention;

[0047] Figure 4 This is a flow chart of the contact surface detection module of the present invention;

[0048] Figure 5 This is a flow chart of the buffer determination module of the present invention;

[0049] Figure 6 This is a flow chart of the composite diagnostic module of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0051] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0052] See also Figure 1 , the contact mechanical wear and fault diagnosis system includes:

[0053] The data acquisition module acquires the time series data of three-phase opening actions, collects the instantaneous measurement value of contact resistance, reads the channel signal of the ring array pressure sensor, captures the envelope waveform of the vibration shock wave, and aligns and stores the four types of data with millisecond-level timestamps to generate a multi-source spatiotemporal dataset.

[0054] The wear diagnosis module uses the tripping time series from the multi-source spatiotemporal data set to calculate the standard deviation of the operation time, extract the absolute difference between the current contact resistance value and the reference value, and calculate the covariance value of the time standard deviation and the resistance difference. When the covariance value continuously exceeds the set threshold, a diagnosis result of the contact wear status is generated;

[0055] The contact surface detection module is based on the pressure array data of the multi-source spatiotemporal dataset. It divides the detection quadrants into four detection quadrants according to the annular distribution and calculates the ratio of the pressure difference between adjacent sensors to the spacing. When the gradient value in the quadrant continuously exceeds the set threshold, the coordinates of the concave area are generated.

[0056] The buffer determination module, based on a multi-source spatiotemporal dataset, intercepts the shock wave envelope waveform at the start of the tripping phase, calculates the maximum rate of change along the voltage rise edge, measures the time span from the start to the end of the waveform, compares the ratio of the rate value to the time span with the device benchmark parameters, and generates a buffer performance attenuation determination result.

[0057] The composite diagnosis module is based on the spatial mapping model of the contact wear status diagnosis results and the coordinates of the recessed area, and calls the buffer efficiency attenuation judgment results to calibrate the time series analysis benchmark. When the diagnosis results are triggered at the same time, the mechanical fault type conclusion is generated according to the combined relationship between the area of ​​the spatial overlapping area and the calibrated benchmark parameters.

[0058] The multi-source spatiotemporal data set includes the tripping time series, instantaneous resistance value, pressure array signal, and shock wave envelope waveform. The contact wear status diagnosis results include the time volatility index, resistance difference index, and covariance threshold judgment result. The coordinates of the depression area include the detection quadrant number, pressure gradient ratio, and abnormal location information. The buffer efficiency attenuation judgment results include the voltage change rate, impact duration, and efficiency ratio comparison results. The mechanical fault type conclusion includes the wear status result, depression space characteristics, and buffer performance judgment value.

[0059] See also Figure 2 , the data acquisition module includes:

[0060] The three-phase trip time submodule extracts the start and end times of the three-phase trip action time series data, obtains the duration of each phase action, calculates the time difference between the three phases, the average time difference, and the time interval between the trip phases, and generates the trip time difference interval;

[0061] First, the start and end time of each phase action needs to be extracted from the device, which is usually recorded using a high-precision timestamp collected synchronously. After the system detects the current jump feature, it records the start time of phase A as 1.025 seconds, the start time of phase B as 1.027 seconds, and the start time of phase C as 1.030 seconds. The corresponding end times are 1.072 seconds, 1.076 seconds, and 1.074 seconds respectively. Subtracting the start time from the end time, the duration of phase A is 0.047 seconds, the duration of phase B is 0.049 seconds, and the duration of phase C is 0.044 seconds. Further, the start time of the three phases is compared pairwise, and the start time difference of phase AB is 0.002 seconds, BC is 0.003 seconds, and AC is 0.005 seconds. The three values ​​are added up and divided by three to obtain the average time difference (0.002 +0.003+0.005) / 3=0.0033 seconds, and then determine the average trip delay index. The time interval between the minimum trip phase, i.e., phase A with the earliest start time, and the maximum trip phase, i.e., phase C with the latest start time, is 0.005 seconds. This interval is the maximum time offset among the three phases. The same treatment is applied to the end time. Finally, the time difference interval division standard in the trip action is constructed, and the interval division value is 0.002 seconds. 0–0.002 is defined as a low difference, 0.002–0.004 as a medium difference, and 0.004 and above as a high difference. Based on the dense distribution of samples, the difference frequency under repeated sampling is accumulated and statistically analyzed. If the frequency of a time difference interval exceeds 25% of the total number of samples, the interval is set as the dominant difference range and used accordingly for the next sub-module call.

[0062] The pressure signal decoding submodule calls the opening time difference interval, reads the annular array pressure sensor channel signal, and synchronously compares the differential azimuth pressure values ​​at the same time according to the channel number, determines the instantaneous pressure change rate, and selects the channels whose change rate exceeds the pressure change rate threshold. It obtains the instantaneous pressure extreme point and corresponding time in the selected channel, and generates the high-variability pressure peak rate;

[0063] The time difference interval generated above is called to read the instantaneous data of all pressure channels within a window of 0.01 seconds before and after the opening time point. The channels are numbered from 1 to 16, corresponding to pressure sensors arranged in different directions on the circular array. The system accurately compares the pressure changes of each channel at the same time according to the timestamp, and samples are taken according to the time step of 0.0001 seconds. The total sampling time is 0.05 seconds, and a total of 500 sets of pressure data are obtained. The pressure difference between adjacent time points is calculated and divided by the time step to obtain the change rate sequence. Taking channel number 3 as an example, its pressure rises rapidly from 10.25MPa to 12.00MPa, the change time is 0.003 seconds, and the change rate is (12.00–10.25) / 0.003=58 3.33MPa / s, which is compared with the maximum pressure change rate threshold of 450MPa / s set by the system, and channel 3 is determined to be an over-threshold channel. The setting of the threshold of 450MPa / s is based on the average value of the maximum instantaneous pressure change of 520MPa / s in the nearly 20 groups of historical sampling, and is set downward by 15% as a safety threshold deviation, which is representative. By screening out all channel numbers that exceed the threshold, the time point of the maximum pressure change and its corresponding value are obtained in the rate change curve of each channel. For example, the maximum rate point of channel 3 appears at 1.031 seconds, and the pressure is 12.00MPa. This pressure point is recorded as the instantaneous pressure extreme point. This type of extreme point will be used to generate a high-variability pressure peak rate data group for processing by the shock wave module.

[0064] The shock wave processing submodule uses the high-variability pressure peak rate to capture the vibration shock wave envelope waveform, divides the envelope waveform into energy segments based on timestamps, extracts the envelope peak value, mean value, and fluctuation within the segment, and classifies the fluctuation amplitude level in different time periods through the combined comparison between the fluctuation and the pressure peak rate. The distribution characteristics corresponding to the shock response data are obtained, and a multi-source spatiotemporal dataset is established.

[0065] The aforementioned high-variability pressure peak rate array is called, and the marked peak occurrence time is selected as the center. 0.01 seconds are taken before and after to form an analysis window. The pressure waveform envelope within the window is extracted, and the peak locking method is used to identify the maximum value and average value for each waveform. A fixed time interval of 0.002 seconds is set in this segment. A total of 10 energy segments are divided. The envelope waveform in each segment is scanned point by point to extract the maximum peak and the corresponding time. For example, in the third segment, the maximum value of the envelope waveform is 2.75MPa, the corresponding time is 1.034 seconds, and the average is 2.22MPa. The fluctuation is obtained by calculating the difference between the maximum and minimum pressures in this segment. 0.91MPa, cross-comparison is made between the fluctuation of each section and the peak pressure rate value of the corresponding channel. When the fluctuation is greater than 0.8MPa and the peak rate exceeds 500MPa / s, it is defined as a high-fluctuation and high-variability section. The system presets a fluctuation critical value of 0.8MPa. Based on 200 historical waveform analysis samples, it is determined by adopting the 1.5 times standard deviation delimiter value of the center area of ​​the Gaussian distribution curve tail data set. The number of such sections is counted as a representative indicator of the impact response amplitude of this section. In the same time period, the multi-temporal and spatial data sets formed by all channel data eventually constitute the response intensity distribution feature set, which is used for response decision-making in subsequent stages.

[0066] See also Figure 3 , the wear diagnosis module includes:

[0067] The trip time series extraction submodule extracts trip action information corresponding to each moment based on multi-source spatiotemporal datasets, identifies time record points, and establishes a continuous trip time series. When extracting data, the records are arranged and filtered in chronological order, data with abnormal time stamps and missing records are removed, and a time series trip data column is generated.

[0068] Based on the multi-source spatiotemporal dataset, the operation is performed. First, the original operation log dataset uploaded by the circuit breaker control unit is called. Each data contains three fields: action type, operation flag and trigger timestamp. The system extracts the action records with the "opening" mark one by one, and determines whether it is a valid opening event by checking whether the action flag field is equal to "OPEN". After filtering and extracting such records, a preliminary operation sequence is formed. Then, the timestamp field value is called to sort the extracted records in ascending order, and the initial time series is constructed by sorting by the trigger time. Then, it is determined whether there are abnormal records in the time series. The interval value of each adjacent time point is calculated. If the interval between two time points is less than the set minimum opening interval, If the threshold is 10 seconds, the latter is judged as a false trigger and is removed. For example, if the trigger time of the first record is 08:00:00 and the second record is 08:00:05, the interval is 5 seconds. If it is less than 10 seconds, it is considered an anomaly. After removing the second record, the cleaned time series is regenerated. In addition, the system verifies the integrity of the timestamp field. If the timestamp field in the record is empty or the format is illegal, such as "08-00-ss", the record is marked as an invalid record and removed. After completing the screening, the final time series tripping data column is obtained. For example, the tripping times finally retained are 08:00:00, 08:10:15, 08:20:30, 08:31:00, and 08:41:45.

[0069] The operation time fluctuation calculation submodule calculates the difference of each operation time based on the time-series tripping data column, obtains the operation time series, calls the series to calculate the standard deviation, uses the sliding window method to estimate the variance within a certain time interval and performs square root processing to generate the operation time standard deviation series;

[0070] Calculate the time difference between adjacent moments one by one to obtain the operation time interval sequence. According to the time points in the above example, the calculated time intervals are: the difference between the first and second points is 10 minutes and 15 seconds, that is, 615 seconds; the difference between the second and third points is 10 minutes and 15 seconds, that is, 615 seconds; the difference between the third and fourth points is 10 minutes and 30 seconds, that is, 630 seconds; the difference between the fourth and fifth points is 10 minutes and 45 seconds, that is, 645 seconds. Therefore, the operation time interval sequence is:

[0071] ;

[0072] Then the standard deviation of the series is calculated by first calculating the mean , and then calculate the variance: the first term is , the second term is also 126.5625, and the third term is , the fourth item is , the sum is 618.75, divided by the number of samples, the variance is 154.6875, and the square root of it is the standard deviation of the operation time. , then the sliding window calculation method is used, the window size is 3, and the first window is

[0073] ;

[0074] Its mean is 620 and its standard deviation is , the second window is

[0075] ;

[0076] Its mean is 630 and its standard deviation is , the standard deviation sequence of the generated operation time is

[0077] .

[0078] The contact resistance covariance analysis submodule calls the operation time standard deviation sequence, obtains the operation time standard deviation value at the current moment, detects the contact resistance measurement value at the corresponding moment and the preset contact resistance reference value, calculates the absolute difference, and constructs a covariance function model with the current standard deviation value using the formula:

[0079] ;

[0080] Calculate and obtain the current covariance value, compare the current covariance value in a continuous period with the threshold, determine whether it is continuously exceeded, obtain the continuous trigger state interval, and generate the contact wear state diagnosis result;

[0081] in, represents the current covariance value, Representative The contact resistance value at the moment, Representative The contact resistance reference value at the moment, Representative The standard deviation of the operation time at each moment, is the mean of the standard deviation series of operation time, is the sample size;

[0082] Call the operation time standard deviation sequence and take the operation time standard deviation values ​​corresponding to the two moments as follows: and , and the contact resistance measurement values ​​at the two moments are called respectively 、 , the contact resistance reference values ​​are set as 、 , calculate the mean of the standard deviation sequence as , bring into the covariance calculation formula:

[0083] ;

[0084] The specific calculation is as follows:

[0085] The absolute difference of the first term is 2.0, and the standard deviation shift is , the product is ;

[0086] The absolute difference of the second term is 2.5, the standard deviation shift is 2.59, and the product is 6.475;

[0087] The sum of the two is , the covariance is:

[0088] ;

[0089] This value is used to compare with the wear trigger threshold. If the threshold is set to 0.6, the current value 0.6475 is greater than the threshold, and the current is determined to be the initial stage of contact wear, and the moment is marked as the starting point of potential wear. If three consecutive covariance values ​​are greater than the threshold, it is identified as a continuous wear trend interval. This process does not require calling the model, and only compares the current point by point. The benefit of this formula, which combines absolute deviation with standard deviation offset, is that it simultaneously assesses the linkage between contact resistance change and operational fluctuation, providing physical interpretability. This result indicates that when the amplification of operational time fluctuations coincides with resistance deviation, their multiplicative effect increases the covariance. Exceeding the set threshold indicates a trend of contact performance degradation.

[0090] See also Figure 4 , the contact surface detection module includes:

[0091] The quadrant division submodule obtains the pressure array data from the multi-source spatiotemporal data set, constructs a two-dimensional coordinate distribution map based on the sensor spatial position, divides the entire detection area with the center point as the origin, and evenly divides it into four quadrants, marked as quadrants 1 to 4. The corresponding sensor data spatial coordinates are extracted within the quadrant to establish the coordinate set within the quadrant;

[0092] The quadrant division submodule obtains the pressure array data from the multi-source spatiotemporal data set. In this process, the system first extracts the two-dimensional coordinates and pressure value information of each sensor point, constructs a two-dimensional coordinate map and uses the regional center point as the center point. As the coordinate origin, the entire detection plane is divided into four quadrants through the origin, that is, the plane is divided equally into quadrants 1 to 4 by horizontal and vertical lines respectively. The sensing area is a square area of ​​10 cm × 10 cm, where quadrant 1 corresponds to the upper right corner area, quadrant 2 to the upper left corner area, quadrant 3 to the lower left corner area, and quadrant 4 to the lower right corner area. If the coordinates of a group of sensors are 、 、 、 , the system judges the coordinates of each point, and Assigned to quadrant one, and Assigned to quadrant two, and Classified into quadrant three, and Assigned to quadrant 4, so sensor 1 belongs to quadrant 1, sensor 2 belongs to quadrant 2, sensor 3 belongs to quadrant 3, and sensor 4 belongs to quadrant 4. The system collects the corresponding coordinates in each quadrant to form a quadrant coordinate set. For example, the corresponding coordinate set of quadrant 1 is , quadrant four corresponds to ,After this processing is completed, an independent sensor spatial ,distribution data set is formed in each quadrant.

[0093] The pressure gradient calculation submodule calls the coordinate set in the quadrant, selects two adjacent points in the quadrant as a unit group, performs difference calculation on the pressure values ​​between each set of sensor information, and calculates the distance between the corresponding sensor coordinates using the formula:

[0094] ;

[0095] Obtain the gradient value within the quadrant by calculation, compare the continuous gradient value with the set gradient threshold, retain the coordinate points corresponding to the continuous exceeding area, and generate the continuous exceeding area;

[0096] in, Indicates sensor and The gradient value between 、 are the pressure values ​​of the two sensors respectively, 、 、 、 is the spatial coordinate, For quadrant The average pressure value of all sensors in the system;

[0097] The gradient value is used to quantify the degree of pressure change between adjacent sensors. The calculation process is divided into three steps: First, calculate the distance between the two sensor coordinates. ; Then calculate the relative change of pressure difference and average pressure (ε is the minimum value to prevent zero); the final gradient value This value reflects the rate of pressure change per unit distance. When the gradient values ​​of multiple consecutive sensor pairs exceed the set threshold, it indicates that there is an abnormal pressure mutation in the area, which may correspond to a physical depression on the mechanical contact surface.

[0098] The pressure gradient calculation submodule calls the quadrant coordinate set, selects two sensor points in each quadrant as analysis objects, and calculates the pressure gradient for them. Assume that there is a sensor in quadrant 1. and , and its corresponding coordinates are 、 , and the corresponding pressure values ​​are 、 The average pressure in the quadrant is , then according to the gradient calculation formula:

[0099] ;

[0100] First calculate the coordinate distance:

[0101] ;

[0102] Then calculate the pressure difference term:

[0103] ;

[0104] Then calculate the mean deviation term:

[0105] ;

[0106] The final gradient value is:

[0107] ;

[0108] If the gradient threshold set by the system is 1.5, then the gradient value 1.8 is greater than the threshold, indicating that the pressure gradient between the coordinate point pair is out of limit. The coordinates of this pair of sensors should be retained and added to the out-of-limit area mark list. If multiple groups of sensors meet this judgment at the same time, a continuous out-of-limit area is formed. The system records the out-of-limit values ​​of each group by serial number and stores them in the continuous out-of-limit array for subsequent sag judgment.

[0109] The sunken area recognition submodule determines whether the number of continuous point sets exceeds the ratio threshold of the number of sensors in the quadrant based on the continuous over-limit area. If the condition is met, all coordinate point sets in the continuous area are selected, the boundary range is extracted to construct a closed space, the center point of the area is calculated and the spatial coordinates are marked to obtain the coordinates of the sunken area;

[0110] According to the set of coordinate points extracted from the continuous over-limit area, the system first counts the total number of sensors in each quadrant and the number of points in the continuous over-limit area. Assuming that the total number of sensors in quadrant 1 is 4 and the number of points in the continuous over-limit area is 3, the ratio is , to determine whether the value exceeds the set ratio threshold. For example, the system sets the threshold to 70%, that is, 0.7, then 0.75 is greater than 0.7, indicating that the condition is met. Then the system extracts all coordinate points in the continuous area and constructs a closed boundary. The maximum and minimum coordinate values ​​of the boundary points are extracted to construct an enclosing rectangular area. Let the current coordinate point set be ,but 、 、 、 , calculate the center point of the region:

[0111] ;

[0112] Finally, the coordinates of the sunken area are obtained as The coordinates are output as the center point of the depression and marked as the center point of the suspected physical subsidence area in the quadrant.

[0113] See also Figure 5 , the buffer determination module includes:

[0114] The waveform interception submodule obtains the shock wave envelope waveform at the start of the tripping phase from the multi-source spatiotemporal dataset. For each waveform, the waveform segment between the start and end points is extracted according to the trigger time window. Segments with insufficient signal fluctuation amplitude are removed and boundary alignment is performed to obtain the tripping waveform segment.

[0115] The shock wave envelope waveform at the initial stage of tripping is obtained from the multi-source spatiotemporal data set. The system first extracts the trigger starting point based on the tripping trigger signal marked in the equipment work record. Assuming that the waveform record is a continuous signal stream with a duration of 10 milliseconds and a sampling frequency of 1kHz, each waveform segment contains 1000 data points. The system extracts the 2-millisecond time windows before and after the trigger index value as the interception boundary, that is, the waveform data within 4 milliseconds from 2 milliseconds before the trigger point to 2 milliseconds after the trigger point. If the trigger point is at the 500th sampling point, the waveform segment is extracted from the 480th point to the 520th point, and the waveform sequence in this interval is traversed. The amplitude difference of adjacent points is calculated point by point. For example, the amplitude difference between points 481 and 482 is If the difference is less than the set minimum fluctuation amplitude of 0.05V, it will be recorded as an invalid point segment. Three or more consecutive invalid point segments will be eliminated as a whole. For example, in a certain waveform segment, point segments with consecutive differences of 0.01, 0.03, and 0.02 will be completely excluded. The remaining waveform segments will use the first valid point as the alignment starting point. All waveform segments will be uniformly reset to zero through shift processing. If the first valid point in the waveform is the 485th point, the system will set the time corresponding to this point to 0ms, and all subsequent points will recalculate the time index relative to this point. The final output segment is such as a voltage value sequence.

[0116] ;

[0117] The corresponding time is

[0118] ;

[0119] milliseconds, forming a breaking waveform segment.

[0120] The voltage rate calculation submodule calls the trip waveform segment, identifies the voltage value sequence corresponding to the time point, locates the starting and ending positions of the voltage rising segment, calculates the ratio of the voltage change amplitude to the corresponding time interval, and records the start and end time difference of the entire segment waveform using the formula:

[0121] ;

[0122] Calculate the rising amplitude characteristic value, take the coupled change of the waveform rising rate and time span as the quantization result, and generate the amplitude change result;

[0123] in, represents the characteristic value of the rising amplitude, Indicates the waveform The voltage change rate of the point, Indicates the time interval between the start and end of the waveform. Indicates the Point voltage value, represents the mean of all voltage values, Indicates the total number of waveform data points;

[0124] This characteristic value is used to characterize the steepness of the voltage waveform during the circuit breaker opening process. The calculation process includes: calculating the voltage change rate of each sampling point in the intercepted waveform segment ; Extract the maximum rate of change ; Calculate the overall time span of the waveform ; Final eigenvalue ,in is the mean value of the waveform voltage. This value comprehensively reflects the relationship between the voltage rise rate and duration. An abnormal increase in the value indicates that the mechanical action impact caused by the degradation of the buffer performance has intensified.

[0125] Call the trip waveform segment and calculate the rising amplitude characteristic value. Assume that the input waveform segment length is 6 sampling points, and the corresponding voltage value sequence is

[0126] ;

[0127] (unit is volt), the time interval is equidistant 0.2 milliseconds, and the total time span of the entire waveform rising segment is , first solve the voltage change rate and define ,have to:

[0128] ;

[0129] ;

[0130] ;

[0131] ;

[0132] ;

[0133] Calculate the maximum voltage rise rate as , so the first term is:

[0134] ;

[0135] Then calculate the average voltage:

[0136] ;

[0137] Next, calculate the absolute deviation of each point from the mean:

[0138] ;

[0139] ;

[0140] ;

[0141] ;

[0142] ;

[0143] ;

[0144] The total deviation is , the second item is:

[0145] ;

[0146] The final calculated voltage rise characteristic value is:

[0147] ;

[0148] This value represents the quantitative result of the waveform's rising speed and its relative stability.

[0149] The performance evaluation submodule uses the buffer performance benchmark value provided by the equipment based on the amplitude change results to perform a ratio judgment, comparing the characteristic value with the benchmark value. If the difference exceeds the limit continuously, an abnormality mark is marked, and the corresponding time point is collected and stored. The abnormal feature vector is used to obtain the buffer performance attenuation judgment result;

[0150] According to the amplitude change results obtained from the above calculation To make a judgment, the system first reads the design performance benchmark value of the current device buffer , the difference between the two is obtained as the deviation value The system presets the deviation threshold of abnormal identification to 0.4. After comparison, it is found that the deviation value is greater than the threshold, so the current waveform data is marked as a performance abnormal point. The system further determines whether the abnormality is continuous. For example, the calculation results of three consecutive waveform segments are 1.83, 1.87, and 1.91 respectively, and the differences from the benchmark value of 1.2 are 0.63, 0.67, and 0.71, which are all greater than the set threshold and meet the continuous over-limit judgment condition. The system records the corresponding three segment time periods and binds them to abnormal labels. The system simultaneously constructs the abnormal feature vector and assigns it

[0151] ;

[0152] The sequence is stored in the abnormal database for subsequent equipment performance status analysis, and finally a conclusion is output that the current buffer has performance degradation.

[0153] See also Figure 6 , the composite diagnostic module includes:

[0154] The state mapping submodule obtains the contact wear state diagnosis results and the coordinates of the concave area, matches the spatial coordinate point set of the concave area with the corresponding wear state time node, selects the areas with overlapping distribution in the same time period and records the geometric contour information, establishes a dual index structure of space and time, and generates a spatial mapping relationship set;

[0155] After the state mapping submodule obtains the contact wear state diagnosis results and the coordinates of the recessed area, the system first extracts the two-dimensional coordinate point set of each recessed area and its corresponding start and end time period in turn. During the extraction process, the system indexes the recessed area number, coordinate configuration and time interval. For example, area R1 is composed of the point set (5.0, 6.0), (5.5, 6.2), (6.0, 5.9), with a start time of 120 milliseconds and an end time of 190 milliseconds. The system pairs the state time nodes recorded in the contact wear diagnosis results. For example, the "mild wear" state occurs between 140 milliseconds and 180 milliseconds. The system calculates the time overlap interval and determines whether it is a valid intersection. If the intersection is greater than the set minimum overlap threshold of 10 milliseconds, the time match is considered valid. In this case, the overlap area is 40 milliseconds, and if it exceeds the 10 millisecond threshold, the time matching condition is passed. ,Then the system extracts the spatial contour of the concave area within the time period and determines the boundary using the minimum circumscribed rectangle method. For example, the minimum value of the boundary in the X direction is 5.0 and the maximum value is 6.0, and the minimum value of the boundary in the Y direction is 5.9 and the maximum value is 6.2. The contour width is 1.0 cm, the height is 0.3 cm, and the calculated area is 0.3 square centimeters. The system constructs a composite index with this spatial information, the time tag, and the wear status tag, and binds the three-dimensional structure of space, time, and status to form a spatial mapping item. The data structure records the region number, spatial contour, time range, and corresponding diagnostic level. The final output mapping structure is, for example: region R1, corresponding to the time period 120 to 190 milliseconds, corresponding to the status level "mild", and the corresponding spatial boundaries are X: 5.0 to 6.0, Y: 5.9 to 6.2. The system stores this structure in the status mapping cache.

[0156] The benchmark calibration submodule calls the buffer performance attenuation determination result, extracts the time series information corresponding to the determination, identifies the starting time point and duration of the diagnostic event, matches it with the time segments of the area in the spatial mapping relationship concentration, extracts the overlapping intervals, and updates the indicator value range according to the internal standard calibration parameters to obtain the calibration benchmark parameter set;

[0157] After the benchmark calibration submodule calls the buffer performance attenuation judgment result, it extracts the time series segment corresponding to the abnormal judgment. For example, the system records that a certain buffer performance attenuation occurs between 240 milliseconds and 305 milliseconds. After locating the time period, the system calls all time period items in the spatial mapping structure and compares them one by one to determine whether there is a time intersection. Assuming that the time period of area R2 is 250 milliseconds to 320 milliseconds, the system determines whether the two time periods intersect. The intersection segment is 250 milliseconds to 305 milliseconds, with a length of 55 milliseconds, which exceeds the set minimum effective comparison length threshold of 30 milliseconds and is determined to be a valid matching area. The system further extracts the spatial area of ​​the area and sets it to 6.1 square centimeters. It is compared with the internal calibration parameters. The default internal benchmark area of ​​the device is set to 5.5 square centimeters. This value is set by the statistical average value of the normal operating area measured when the equipment leaves the factory. The system statistics show that the average area of ​​the spatial contour in 100 samples without abnormalities is 5.5 square centimeters, with a standard deviation of 0.2 square centimeters. The system stipulates that the threshold deviation is set to three times the standard deviation, that is, 0.6 square centimeters. Therefore, an area exceeding 6.1 square centimeters is out of limit. The current area is 6.1 square centimeters, which exceeds the baseline value by 0.6 square centimeters and meets the over-limit condition. The system continues to extract the response time parameters associated with the area. The response time is the duration from the start to the end of the abnormality, which is 65 milliseconds in this segment. The system defaults that the effective response time of the buffer should be less than 40 milliseconds. This value comes from the average response time of 38.5 milliseconds for 100 operations of the equipment during new installation and debugging. The maximum value is taken and a 5% margin is added to obtain 40 milliseconds as the threshold. Therefore, the current response time is also abnormal. The system updates the area and response time parameters and writes them into the baseline parameter set of the area, and marks it as an object to be compared.

[0158] The fault determination submodule identifies the area value of the spatial overlapping region within the diagnosis time based on the spatial mapping relationship set and the calibration reference parameter set, combines and compares the calibration parameters associated with the region, and couples and screens the area index of the spatial region with the corresponding calibration parameter set. When the parameter combination relationship meets the abnormality determination conditions, the conclusion of the mechanical fault type is obtained;

[0159] First, determine whether the area in the mapping structure has synchronous data of time and parameters. Analyze the area and response time combination of the areas that meet the conditions one by one to see if they trigger the fault judgment condition. Suppose the current matching area number is R2, the area is 6.1 square centimeters, and the response time is 65 milliseconds. According to the internal rules of the system, if the area of ​​a certain area is larger than the set value of 5.5 square centimeters and the response time exceeds 40 milliseconds, it will be marked as a potential abnormal area. The judgment standard is based on the fact that the experimental equipment recorded a total of 150 sets of data at the initial stage of performance degradation, of which 95% of the effective pre-fault areas were not less than 5.6 square centimeters and the response time exceeded 40 milliseconds. The response time is longer than 42 milliseconds, so the lower limit of 5.5 square centimeters and 40 milliseconds are taken as the minimum combined threshold for real-time online judgment. The current area meets both conditions, that is, the area is 6.1 greater than 5.5, and the response time is 65 milliseconds greater than 40 milliseconds. The system judges it as an abnormal area. Next, combined with the regional status information and historical diagnostic label analysis, if a similar judgment has occurred in the same area in the previous cycle and the area increase this time exceeds 1.5 times the previous one, for example, the previous area was 4.0 square centimeters, then the area growth ratio is 6.1 / 4.0=1.5256.1 / 4.0=1.5256.1 / 4.0=1.525;

[0160] The system determines that it is an area expansion type fault, and further identifies whether the contact wear level in the area is "severe" or "moderate". If the level is "severe", the system confirms that there is a "buffer mechanical structure damage type fault" in the area and outputs the fault conclusion label. At the same time, it records the fault number, associated time period, spatial contour boundary and key parameters for maintenance retrieval.

[0161] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. Contact mechanical wear and fault diagnosis system, characterized in that: The system comprises: The data acquisition module acquires the three-phase opening action time series, the instantaneous value of the contact resistance, the annular array pressure sensor channel signal, and the vibration shock wave envelope waveform, aligns and stores them by timestamp, and generates a multi-source spatiotemporal dataset. The wear diagnosis module calls the multi-source spatiotemporal data set, calculates the standard deviation of the operation time, extracts the difference between the contact resistance value and the reference value, calculates the covariance between the two, and generates a contact wear status diagnosis result when the covariance continuously exceeds a threshold; The contact surface detection module divides the pressure array data into four detection quadrants based on the multi-source spatiotemporal data set, calculates the ratio of the pressure difference between adjacent sensors to the spacing, and generates the coordinates of the concave area when the gradient in the quadrant continuously exceeds the threshold; The buffer determination module intercepts the shock wave envelope waveform based on the multi-source spatiotemporal data set, calculates the ratio of the voltage rising edge change rate to the waveform time span, and generates a buffer performance attenuation determination result; The composite diagnostic module generates a conclusion on the type of mechanical fault based on a spatial mapping model of the contact wear state diagnosis results and the coordinates of the recessed area, combined with a time series analysis of the buffer efficiency attenuation determination results. When the diagnostic results are triggered simultaneously, the composite diagnostic module generates a conclusion on the type of mechanical fault based on a combination of the overlapping area and calibration parameters. The buffer determination module includes: The waveform interception submodule obtains the shock wave envelope waveform at the start of the tripping phase in the multi-source spatiotemporal dataset, extracts the waveform segment between the start point and the end point of each waveform according to the trigger time window, removes segments with insufficient signal fluctuation amplitude, and performs boundary alignment to obtain the tripping waveform segment; The voltage rate calculation submodule calls the trip waveform segment, identifies the voltage value sequence corresponding to the time point, locates the starting and ending positions of the voltage rising segment, calculates the ratio of the voltage change amplitude to the corresponding time interval, and records the start and end time difference of the entire segment waveform, using the formula: ； Calculate the rising amplitude characteristic value, take the coupled change of the waveform rising rate and time span as the quantization result, and generate the amplitude change result; in, represents the characteristic value of the rising amplitude, Indicates the waveform The voltage change rate of the point, Indicates the time interval between the start and end of the waveform. Indicates the Point voltage value, represents the mean of all voltage values, Indicates the total number of waveform data points; The performance evaluation submodule, based on the amplitude change result, calls the buffer performance benchmark value provided by the device, performs a ratio judgment, compares the characteristic value with the benchmark value, and marks an abnormality mark if the difference exceeds the limit continuously. The corresponding time point is collected and the abnormal characteristic vector is stored to obtain the buffer performance attenuation judgment result; The composite diagnostic module comprises: The state mapping submodule obtains the contact wear state diagnosis results and the coordinates of the recessed area, matches the spatial coordinate point set of the recessed area with the corresponding wear state time node, selects areas with overlapping distribution within the same time period and records geometric contour information, establishes a dual index structure of space and time, and generates a spatial mapping relationship set; The benchmark calibration submodule calls the buffer performance attenuation determination result, extracts the time series information corresponding to the determination, identifies the starting time point and duration of the diagnostic event, matches it with the time segments of the area in the spatial mapping relationship concentration, extracts the overlapping intervals, and updates the indicator value range according to the internal standard calibration parameters to obtain the calibration benchmark parameter set; The fault judgment submodule identifies the area value of the spatial overlapping area within the diagnosis time based on the spatial mapping relationship set and the calibration reference parameter set, combines and compares the calibration parameters associated with the area, couples and screens the area index of the spatial area with the corresponding calibration parameter set, and obtains the conclusion of the mechanical fault type when the parameter combination relationship meets the abnormal judgment condition.

2. The contact mechanical wear and fault diagnosis system according to claim 1, characterized in that: The multi-source spatiotemporal data set includes the tripping time series, instantaneous resistance value, pressure array signal, and shock wave envelope waveform; the contact wear status diagnosis results include the time volatility index, resistance difference index, and covariance threshold judgment result; the recessed area coordinates include the detection quadrant number, pressure gradient ratio, and abnormal location information; the buffer efficiency attenuation judgment results include the voltage change rate, impact duration, and efficiency ratio comparison results; the mechanical fault type conclusion includes the wear status result, recessed space characteristics, and buffer performance judgment value.

3. The contact mechanical wear and fault diagnosis system according to claim 2, characterized in that: The data acquisition module includes: The three-phase trip time submodule extracts the start and end times of the three-phase trip action time series data, obtains the duration of each phase action, calculates the time difference between the three phases, the average time difference, and the time interval between the trip phases, and generates the trip time difference interval; The pressure signal decoding submodule calls the opening time difference interval, reads the annular array pressure sensor channel signal, synchronously compares the differential azimuth pressure values ​​at the same time according to the channel number, determines the instantaneous pressure change rate, and selects the channels whose change rate exceeds the pressure change rate threshold, obtains the instantaneous pressure extreme point and corresponding time in the selected channel, and generates a high-variability pressure peak rate; The shock wave processing submodule calls the high-variability pressure peak rate, captures the vibration shock wave envelope waveform, divides the envelope waveform into energy segments based on timestamps, extracts the envelope peak, mean, and fluctuation within the segment, and classifies the fluctuation amplitude levels in differentiated time periods through a combined comparison between the fluctuation and the pressure peak rate, obtains the distribution characteristics corresponding to the shock response data, and establishes a multi-source spatiotemporal dataset.

4. The contact mechanical wear and fault diagnosis system according to claim 3, characterized in that: The wear diagnosis module includes: The trip time series extraction submodule extracts the trip action information corresponding to the time based on the multi-source spatiotemporal dataset, identifies the time record points, and establishes a continuous trip time series. When extracting data, the records are arranged and filtered in time series, and data with abnormal time marks and missing records are eliminated to generate a time series trip data column. The operation time fluctuation calculation submodule calculates the difference of each operation time based on the time-series tripping data column, obtains the operation time series, calls the series to calculate the standard deviation, uses the sliding window method to estimate the variance within a certain time interval and performs square root processing to generate the operation time standard deviation series; The contact resistance covariance analysis submodule calls the operation time standard deviation sequence, obtains the operation time standard deviation value at the current moment, detects the contact resistance measurement value at the corresponding moment and the preset contact resistance reference value, calculates the absolute difference, and constructs a covariance function model with the current standard deviation value using the formula: ； Calculate and obtain the current covariance value, compare the current covariance value in a continuous period with the threshold, determine whether it is continuously exceeded, obtain the continuous trigger state interval, and generate the contact wear state diagnosis result; in, represents the current covariance value, Representative The contact resistance value at the moment, Representative The contact resistance reference value at the moment, Representative The standard deviation of the operation time at each moment, is the mean of the standard deviation series of operation time, is the sample size.

5. The contact mechanical wear and fault diagnosis system according to claim 4, characterized in that: The contact surface detection module includes: The quadrant division submodule obtains the pressure array data from the multi-source spatiotemporal data set, constructs a two-dimensional coordinate distribution map based on the sensing spatial position, divides the entire detection area with the center point as the origin, and evenly divides it into four quadrant areas, marked as quadrants 1 to 4 respectively, extracts the corresponding sensing data spatial coordinates within the quadrant, and establishes a quadrant coordinate set; The pressure gradient calculation submodule calls the coordinate set in the quadrant, selects two adjacent points in the quadrant as a unit group, performs difference calculation on the pressure values ​​between each set of sensor information, and calculates the distance between the corresponding sensor coordinates using the formula: ； Obtain the gradient value within the quadrant by calculation, compare the continuous gradient value with the set gradient threshold, retain the coordinate points corresponding to the continuous exceeding area, and generate the continuous exceeding area; in, Indicates sensor and The gradient value between 、 are the pressure values ​​of the two sensors respectively, 、 、 、 is the spatial coordinate, For quadrant The average pressure value of all sensors in the system; The sunken area identification submodule determines whether the number of continuous point sets exceeds the ratio threshold of the number of sensors in the quadrant based on the continuous over-limit area. If the condition is met, all coordinate point sets in the continuous area are selected, the boundary range is extracted to construct a closed space, the center point of the area is calculated and the spatial coordinates are marked to obtain the coordinates of the sunken area.

Images