Traction motor bearing health state evaluation and life prediction method

Abstract

The present application relates to the technical field of bearing state monitoring, in particular to a traction motor bearing health state evaluation and life prediction method, comprising the following steps: acquiring bearing multi-source data and aligning and integrating, extracting features to judge degradation sections, dividing state statistical proportions, converting life consumption to evaluate remaining life, uploading information to complete consistency verification, and generating life state codes; in the present application, by synchronously collecting and aligning multiple types of data such as vibration, flow, speed, temperature, etc., the perception accuracy of bearing state changes is improved, the speed is divided into intervals, the rolling frequency features and vibration offsets are extracted, the identification accuracy is enhanced, the degradation sections are integrated in time continuity and normalized to reside for a long time, the state differentiation is improved, the state distribution proportion and the cumulative running time are combined, the life consumption conversion is refined, the life evaluation granularity is enhanced, the life number and the state proportion are uploaded for remote verification, the state result closed-loop management is realized, and the evaluation credibility and the whole-cycle monitoring efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of bearing condition monitoring technology, and in particular to a method for assessing the health status and predicting the lifespan of traction motor bearings. Background Technology

[0002] The field of bearing condition monitoring technology refers to a collection of technologies related to sensing, analyzing, and judging the operating status of bearings in rotating machinery. The core aspects of this technology include collecting, processing, and comparing vibration, temperature, speed, and load changes generated by the bearing during operation. By setting monitoring parameters, criteria, and evolution patterns, the current state of the bearing is identified. The overall technology employs a method based on time-series data acquisition and comparison with historical operating data to continuously analyze and judge the bearing's state changes. Traditional traction motor bearing health status assessment and life prediction methods address the wear, fatigue, and performance degradation problems of bearings in traction motors under long-term operating conditions. This involves acquiring the vibration amplitude changes, frequency component distribution, temperature rise trends, and the relationship between speed and load under different operating conditions. Based on bearing structural parameters and rated operating conditions, real-time monitoring data is compared with existing operating range division standards. The health status and remaining service life of the bearing are judged by combining the parameter change patterns within a predetermined operating cycle. This type of method relies on bearing type parameters, operating time records, monitoring data change trends, and predetermined judgment rules to process related technical matters.

[0003] Current technologies for assessing bearing condition heavily rely on single-dimensional monitoring data, such as analysis based solely on vibration or temperature parameters. This leads to insufficient information coverage during condition assessment. At the data fusion level, the lack of a time synchronization mechanism limits the effectiveness of collaborative assessment using multi-source data, affecting the accurate capture of abrupt changes in condition. Rotational speed variations are not fully utilized, and the rolling frequency range is not effectively defined, resulting in low accuracy in extracting abnormal signals. Condition classification does not differentiate between different degrees of degradation, only performing coarse-grained judgments, failing to provide effective early warnings for mild degradation or potential fatigue trends. Life assessment methods rely on historical periodic data comparisons and preset rules, making it difficult to cover individual differences under various complex operating conditions, posing a risk of generalization and misjudgment. Monitoring results remain local or single-end processing, lacking a data closed-loop and consistency verification mechanism, which may lead to lagging equipment maintenance strategies or distorted assessment results. For example, slight fluctuations in vibration signals during heavy-load operation may be misjudged as severe degradation by traditional methods, resulting in wasted maintenance resources. These shortcomings are particularly evident in scenarios with frequent changes in the operating environment and hidden early signs of failure, limiting the development of precise and intelligent bearing health management. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a method for assessing the health status and predicting the lifespan of traction motor bearings, comprising the following steps:

[0005] S1: Acquire multi-source monitoring data such as vibration signal, volumetric flow rate, speed and temperature of the bearing assembly inside the traction motor end cover during operation, align and integrate them after unifying the time base, and generate a synchronous data package of bearing operation status.

[0006] S2: Divide the speed range according to the speed information in the bearing operating status synchronization data packet, extract the amplitude change of rolling contact frequency, calculate the vibration offset, count the corresponding flow rate change, and mark the degradation characteristic operating segment based on the joint judgment of vibration limit and flow rate threshold, and generate degradation mark sequence.

[0007] S3: Based on the degradation marker sequence, integrate the continuous degradation segments and divide them into three state types: normal, degradation level one, and degradation level two. Calculate and normalize the residence time of the state intervals to generate a bearing state distribution ratio vector.

[0008] S4: Call the bearing state distribution ratio vector, calculate the life consumption ratio according to the life consumption level, evaluate the life in combination with the cumulative running time of the bearing, and compare it with the rated life to generate the available running time segment.

[0009] S5: Based on the available runtime segment matching lifespan interval standard, generate lifespan segment number, call the vehicle controller host to upload lifespan number and status distribution information to the ground maintenance server, complete consistency verification, and generate lifespan status assessment code.

[0010] As a further aspect of the present invention, the bearing operating status synchronization data package includes vibration signal data, volumetric flow rate data, speed change records, temperature change records, and time alignment identifiers. The degradation marker sequence includes speed range number, vibration degradation discrimination marker, flow abnormality discrimination marker, joint discrimination result marker, and degradation operating segment identifier. The bearing status distribution ratio vector includes normal state ratio, degradation level one state ratio, degradation level two state ratio, normalized dwell time ratio, and state range statistical benchmark. The available operating time segment includes life consumption ratio, cumulative operating time benchmark, rated life comparison margin, available time start point, and available time end point. The life status assessment code includes life segment number, status distribution information verification identifier, server consistency verification result code, life status level code, and upload record index.

[0011] As a further aspect of the present invention, the specific steps of S1 are as follows:

[0012] S101: Acquire vibration signals, volumetric flow rate, speed changes and temperature records of the bearing assembly inside the traction motor end cover during operation, identify and classify the data according to the data acquisition time, remove data whose time interval exceeds the period benchmark, and obtain the screened monitoring dataset.

[0013] S102: Based on the sieved monitoring dataset, time interpolation is performed on the volumetric flow rate, rotation speed and temperature data to complete the time order of the data, and data with synchronization offset exceeding the sampling period are removed to obtain the aligned monitoring data sequence.

[0014] S103: Based on the multiple types of data in the aligned monitoring data sequence, a vector sequence is constructed by combining them according to a unified time reference to form a continuous time matrix, and the data is archived and organized according to the sensor type to generate a synchronous data package of bearing operating status.

[0015] As a further aspect of the present invention, the removal of data whose time interval exceeds the period benchmark is specifically defined as follows: taking the sampling period corresponding to the data acquisition time as the period benchmark, when the time interval between any two adjacent vibration signals, volume flow rate, rotation speed change and temperature records is greater than 1.2 times the sampling period, the corresponding data is determined to be invalid data and removed from the sieved monitoring dataset.

[0016] The specific definition of performing time interpolation to complete the volumetric flow rate, rotation speed and temperature data is as follows: based on the timestamps of adjacent valid data, interpolation is used to generate interpolated data at the missing time points, and the interpolation time interval is not greater than the sampling period.

[0017] Specifically, the removal of data whose synchronization offset exceeds the sampling period is limited to the following: when the offset between the timestamp of the volume flow rate, rotation speed and temperature data and the timestamp of the vibration signal is greater than the sampling period, the corresponding data will not participate in the combination and construction of the vector sequence.

[0018] As a further aspect of the present invention, the specific steps of S2 are as follows:

[0019] S201: Based on the rotational speed record in the bearing operating status synchronization data packet, divide the bearing into multiple stable segments according to the continuous change trend, extract the vibration signal amplitude sequence corresponding to the rolling contact frequency in each segment, call the rate of change and fluctuation range of the vibration signal amplitude in the segment, calculate the frequency amplitude offset index, and generate the amplitude offset sequence.

[0020] S202: Call the volume flow rate record data in the bearing operating status synchronization data packet, statistically analyze the volume flow rate change range of the segment according to the time range corresponding to the speed segment, compare the volume flow rate difference change rate of adjacent time periods, and obtain the segment flow rate change rate sequence.

[0021] S203: Based on the amplitude offset sequence and the section flow rate change sequence, call the vibration limit standard and flow judgment threshold, identify and judge the joint exceedance of the two types of results in the section, mark the operating section with joint offset characteristics, and generate a degradation mark sequence.

[0022] As a further aspect of the present invention, the division of multiple stable segments according to the continuous change trend is specifically defined as follows: taking the speed record in the bearing operating status synchronous data packet as a benchmark, when the change amplitude of adjacent speed records does not exceed the preset speed fluctuation threshold within three consecutive sampling periods, the corresponding time range is divided into the same stable segment.

[0023] The frequency amplitude offset index is specifically defined as follows: based on the vibration signal amplitude sequence corresponding to the rolling contact frequency, the maximum value, minimum value and amplitude change rate of the vibration signal amplitude in the stable section are respectively counted, and the frequency amplitude offset index is generated according to the combination relationship between the maximum value, minimum value and change rate.

[0024] The specific definition of identifying and judging the joint exceedance of the two types of results within the section is as follows: when the frequency amplitude deviation index continuously exceeds the vibration limit standard and the section flow change rate within the corresponding stable section continuously exceeds the flow judgment threshold, the corresponding stable section is marked as an operating section with joint deviation characteristics.

[0025] As a further aspect of the present invention, the specific steps of S3 are as follows:

[0026] S301: Based on the start and end times of the marked segments in the degraded marker sequence, continuously extract the marker points whose adjacent time span difference is less than the segment merging threshold, perform time segment splicing operation accordingly, integrate into multiple running intervals, call the unmarked segments as references, divide all intervals into non-overlapping segments, and obtain the running interval time series.

[0027] S302: Based on the time series of the running interval, call the degradation level identifier value corresponding to the segment in the degradation marker sequence, judge the proportion of degradation state in the running interval, classify the running interval into different state types according to the set state division ratio threshold, and obtain the running state classification label set;

[0028] S303: Based on the set of operating status classification labels and the time series of operating intervals, the dwell time under each status label is statistically calculated, the sum of the dwell time of the status labels is normalized, and the labels are rearranged according to the normalization ratio to generate the bearing status distribution ratio vector.

[0029] As a further aspect of the present invention, the specific steps of S4 are as follows:

[0030] S401: Call the bearing state distribution ratio vector, calculate the life consumption ratio corresponding to each state ratio according to the life consumption level table corresponding to the state, map the state ratio to the consumption level, and summarize the state life consumption ratio in vector order to generate the state life consumption ratio vector.

[0031] S402: Based on the state life consumption ratio vector, collect the cumulative running time record of the bearing, multiply the cumulative running time with the components of the state life consumption ratio vector respectively to obtain the life consumption time component corresponding to the state, and sum and summarize the life consumption time components to obtain the cumulative life consumption time value.

[0032] S403: Based on the cumulative life consumption time value, obtain the bearing rated life reference time, compare the cumulative life consumption time with the rated life reference, calculate the remaining available running time range, and divide it into segments according to time continuity to generate available running time segments.

[0033] As a further aspect of the present invention, the specific steps of S5 are as follows:

[0034] S501: Based on the available runtime segments, and referring to the lifespan interval division standard, the duration value corresponding to the segment is matched with the upper and lower boundaries of the lifespan interval. The corresponding number is assigned according to the matching result, and all numbers are arranged and combined in chronological order to generate a lifespan segment number sequence.

[0035] S502: Based on the lifespan segment number sequence, call the bearing status distribution ratio vector, synchronously package the two types of information according to the current task cycle, transmit them to the vehicle controller host, establish information encapsulation fields inside the controller, and upload them to the ground maintenance server to obtain the remote status upload dataset.

[0036] S503: Based on the remote status upload dataset, call the preset field index rules and verification field generation rules of the ground maintenance server to perform field integrity verification and data format verification on the uploaded data, and recalculate the verification fields based on the uploaded life segment number sequence and bearing status distribution ratio vector, compare the consistency with the uploaded verification fields, and generate life status assessment codes.

[0037] As a further aspect of the present invention, the step of matching the duration value corresponding to the segment with the upper and lower boundaries of the lifespan interval is specifically defined as follows: comparing the duration value corresponding to the available runtime segment with the upper and lower boundaries of two adjacent lifespan intervals in the lifespan interval division standard; when the duration value falls between the corresponding upper and lower boundaries, it is determined that the available runtime segment is matched to the corresponding lifespan interval.

[0038] The specific definition of assigning a corresponding number identifier based on the matching result is as follows: a unique integer number is pre-set for each of the lifespan intervals, and the corresponding integer number is written into the lifespan interval number sequence according to the interval matching result;

[0039] The specific limitation of matching the content of the corresponding position in the uploaded data by field index is as follows: based on the field order index defined in the preset verification field of the ground maintenance server, extract the life segment number sequence content and bearing status distribution ratio vector content of the corresponding index position in the remote status uploaded data one by one, and perform consistency comparison verification under the condition that the index is consistent.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0041] In this invention, by synchronously collecting and aligning multiple types of data such as vibration, flow rate, rotational speed, and temperature, the perception accuracy of bearing condition changes is improved. Based on the rotational speed, the bearing is divided into intervals, and the rolling frequency characteristics and vibration offset are extracted to enhance the identification accuracy. The degradation segment is integrated and the dwell time is normalized to improve the condition differentiation. The life consumption conversion is refined by combining the condition distribution ratio and the cumulative running time to enhance the granularity of life assessment. The life number and condition ratio are uploaded for remote verification to achieve closed-loop management of condition results, thereby improving the reliability of the assessment and the full-cycle monitoring efficiency. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram of the steps of the present invention;

[0044] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0045] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0046] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0047] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0048] Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0050] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0051] Please see Figure 1 This invention provides a method for assessing the health status and predicting the lifespan of traction motor bearings, comprising the following steps:

[0052] S1: Acquire the vibration signal output by the acceleration sensor during the operation of the bearing assembly inside the traction motor end cover, collect the volumetric flow rate data in the lubrication oil supply pipeline during the corresponding time period, read the speed change record output by the encoder sensor, and simultaneously acquire the temperature change record collected by the thermocouple sensor. Align and integrate the multi-source monitoring data according to a unified time reference to generate a bearing operation status synchronization data package.

[0053] S2: Divide the speed range according to the speed record in the synchronous data packet of bearing operation status, extract the rolling contact frequency amplitude change of the range and calculate the vibration offset result, count the volume flow rate change of the corresponding time period, and jointly judge the two types of results according to the vibration limit standard and the flow rate judgment threshold, mark the operating section that meets the degradation characteristics, and generate the degradation mark sequence.

[0054] S3: Based on the temporal continuity of the marked segments in the degradation mark sequence, integrate the operating intervals and divide them into three state types: normal, degradation level one, and degradation level two. Calculate the dwell time of the state intervals and perform normalization processing to generate a bearing state distribution ratio vector.

[0055] S4: Call the bearing state distribution ratio vector, calculate the life consumption ratio by referring to the life consumption level corresponding to the state, perform life assessment by combining the cumulative running time of the bearing, and compare it with the rated life benchmark to generate the available running time segment.

[0056] S5: Based on the available runtime segment matching lifespan interval division standard, generate the corresponding lifespan segment number, call the vehicle controller host to upload the lifespan number and status distribution information to the ground maintenance server for consistency verification, and generate lifespan status assessment code.

[0057] The bearing operating status synchronization data package includes vibration signal data, volumetric flow rate data, speed change records, temperature change records, and time alignment identifiers. The degradation marker sequence includes speed range number, vibration degradation discrimination marker, flow anomaly discrimination marker, joint discrimination result marker, and degradation operating section identifier. The bearing status distribution ratio vector includes normal state ratio, degradation level 1 state ratio, degradation level 2 state ratio, normalized dwell time ratio, and status range statistical benchmark. The available operating time segment includes life consumption ratio, cumulative operating time benchmark, rated life comparison margin, available time start point, and available time end point. The life status assessment code includes life segment number, status distribution information verification identifier, server consistency verification result code, life status level code, and uploaded record index.

[0058] Please see Figure 2 The specific steps of S1 are as follows:

[0059] S101: Acquire vibration signals, volumetric flow rate, speed changes and temperature records of the bearing assembly inside the traction motor end cover during operation, identify and classify the data according to the data acquisition time, remove data whose time interval exceeds the period benchmark, and obtain the screened monitoring dataset.

[0060] Table 1. Sample monitoring data of bearing assembly inside end cover

[0061]

[0062] As shown in Table 1, the sample data includes vibration, volumetric flow rate, rotational speed and temperature, and there are missing values ​​and gaps in the acquisition time. The following paragraphs use these gaps and missing values ​​as the input basis for screening and interpolation calculations.

[0063] When acquiring vibration signals, volumetric flow rate, rotational speed changes, and temperature records, the acquisition channels are triggered in parallel according to the same task cycle. Vibration signals are output as a point-by-point sequence from the accelerometer mounted on the outside of the end cap; volumetric flow rate is output as a point-by-point record from the volumetric flow meter in the cooling circuit; rotational speed changes are output as a point-by-point record from the encoder; and temperature records are output as a point-by-point record from the temperature probe on the end cap. At the start of acquisition, a cycle reference and timestamp rules are written. The cycle reference is set to 1.0 second, and the timestamp accuracy is set to 0.01 seconds. Each record is then assigned an acquisition channel identifier and an acquisition batch identifier. After the identifiers are completed, the records are sorted and archived according to the acquisition time. Subsequently, a time interval filtering operation is performed. The acquisition times of two adjacent records are subtracted. If the difference is greater than 1.2 seconds, the data segment is determined to have crossed the cycle reference, and a rejection rule is triggered. The rejection rule involves deleting one record before and after the gap and writing a gap mark at the gap position to prevent subsequent alignment from mistaking the spanned segment for a continuous one. Taking Table 1 as an example, there is a vibration null value at 08:00:03 on 2023-01-13, and the time difference between 08:00:03 and 08:00:05 is 2.0 seconds, which is greater than 1.2 seconds. After removal, 08:00:02 and 08:00:06 are retained as the two valid starting points, and gap markers are written at the corresponding positions between 08:00:03 and 08:00:05. After the removal is completed, the four types of data in the same batch are stored by channel, and a filtered monitoring dataset is generated. This dataset retains the timestamp, channel identifier, original value, and gap marker of each record as the input data source for subsequent interpolation and alignment operations.

[0064] S102: Based on the sieved monitoring dataset, time interpolation is performed on the volumetric flow rate, rotation speed and temperature data to complete the time order of the data, and data with synchronization offset exceeding the sampling period are removed to obtain the aligned monitoring data sequence.

[0065] When performing time interpolation completion based on the sieved monitoring dataset, the timestamp of the rotational speed record is first used as the unified time axis because the rotational speed channel has continuity and minimal time jitter in traction control. Interpolation completion is performed separately for volumetric flow rate, rotational speed, and temperature data. Before interpolation, data with synchronization offsets exceeding the sampling period are removed. The synchronization offset is obtained by comparing the timestamps of different channels within the same time window. The time window width is 1.0 second, and the allowable offset upper limit is 0.3 seconds. If the offset of a volumetric flow rate or temperature record relative to the rotational speed axis exceeds 0.3 seconds, that record is not included in the interpolation and is marked as offset removal. During interpolation, point-by-point completion is performed for gaps no longer than 3.0 seconds. The completion rule uses a linear transition between adjacent valid points. First, the most recent valid value before and after the gap is read, and then the transition ratio is allocated according to the relative position of each axis moment within the gap. Taking Table 1 as an example, the volumetric flow rate is 5.8 cubic meters per hour at 08:00:01, 5.7 cubic meters per hour at 08:00:02, there is a gap from 08:00:03 to 08:00:05, and 5.6 cubic meters per hour at 08:00:06. After completion according to the main axis, approximately 5.7 cubic meters per hour is written at 08:00:03, approximately 5.6 cubic meters per hour at 08:00:04, and approximately 5.6 cubic meters per hour at 08:00:05, and these completion points are marked as interpolation generation points. After completing the interpolation of volumetric flow rate and temperature, the time sequence of the rotational speed itself is unified, records with reversed time are removed and reordered. The four types of data are re-aligned point by point according to the main axis timestamp. After alignment, it is checked whether multiple records appear in any channel at the same time. If so, the record with the timestamp closer to the main axis is retained and the rest are removed to obtain the aligned monitoring data sequence. Each time in the sequence has a unified timestamp and the corresponding four types of values ​​or interpolation marks.

[0066] S103: Based on the multiple types of data in the aligned monitoring data sequence, a vector sequence is constructed by combining them according to a unified time reference to form a continuous time matrix. The data is then archived and organized according to the sensor type to generate a synchronous data package of bearing operating status.

[0067] When constructing a vector sequence based on the aligned monitoring data sequence, a combination rule under a unified time base is first defined. The current values ​​of vibration acceleration, volumetric flow rate, rotational speed, and temperature are read for each timestamp and concatenated in a fixed order to form a single-time vector. The field order within the vector is fixed as vibration acceleration, volumetric flow rate, rotational speed, and temperature to avoid field misalignment in subsequent calculations. Then, the vectors of consecutive timestamps are stacked in chronological order to form a continuous-time matrix, with rows corresponding to timestamps and columns corresponding to fields. After constructing the matrix, an archiving operation is performed. The matrix column indices are written to an archive table according to sensor type. The archive table records the channel type, installation location, and sampling resolution for each column. The vibration channel sampling resolution is set to 2560 points per second and represented in the matrix as eigenvalues ​​within the same second. The eigenvalues ​​are the root mean square values ​​of the vibration acceleration sequence within that second. The resolutions for volumetric flow rate, rotational speed, and temperature are set to 1 point per second. For example, the original vibration sequence within the second 08:00:02 contains 2560 points. First, the DC component is removed from these 2560 points. Then, the root mean square of the absolute value sequence is calculated to obtain 2.5 m / s², and this value is filled into the vibration field of the corresponding row of the matrix. After filling in all timestamps, the matrix, along with the archive table, gap marker table, and interpolation marker table, is encapsulated into a synchronization data packet. The synchronization data packet is stored by batch number and includes the start time, end time, sampling period, and field order check code. The check code is obtained by summing the field order strings character by character. During subsequent readings, the check code consistency is compared to generate a bearing operating status synchronization data packet.

[0068] Please see Figure 3 The specific steps of S2 are as follows:

[0069] S201: Based on the rotational speed record in the bearing operating status synchronization data packet, divide the bearing into multiple stable segments according to the continuous change trend, extract the vibration signal amplitude sequence corresponding to the rolling contact frequency in each segment, call the rate of change and fluctuation range of the vibration signal amplitude in the segment, calculate the frequency amplitude offset index, and generate the amplitude offset sequence.

[0070] When dividing stable segments based on the rotational speed records in the synchronization data packets, a smoothing operation is first performed on the rotational speed sequence. The smoothing window length is set to 5 seconds, and the median value within the window is used as the smoothing point to avoid segment fragmentation caused by single-point jitter. Then, segments are divided according to continuous change trends. The segmentation rule uses a rotational speed change rate threshold, which is obtained by dividing the difference in rotational speed between two adjacent seconds by 2 seconds. If the absolute value of the change rate does not exceed 20 revolutions per minute per second for 10 consecutive seconds, then that 10 seconds and the subsequent continuous portion meeting the condition are classified as the same stable segment. In the example, if the rotational speed changes slowly between 1780 and 1860 revolutions per minute from 08:10:00 to 08:12:00, and the change does not exceed 15 revolutions per minute per second, then this range forms a stable segment. When extracting the vibration amplitude sequence corresponding to the rolling contact frequency for each stable segment, the contact frequency position of that segment is first calculated based on the average rotational speed within the segment. This contact frequency position is obtained by reading the bearing structural parameter table, which records fields such as the number of rolling elements, pitch circle diameter, and contact angle. The frequency position at that rotational speed is obtained according to a predetermined conversion logic. Subsequently, a spectral transformation is performed on the vibration signal per second within that segment, and the peak amplitude with a bandwidth of 2 Hz near the frequency position is selected to form the amplitude sequence. When calculating the rate of change and fluctuation range of the amplitude sequence, the rate of change is taken as the difference between the first and last amplitude values ​​within the segment divided by the segment duration, and the fluctuation range is taken as the difference between the maximum and minimum amplitude values. The frequency amplitude offset index is generated jointly by the rate of change and the fluctuation range. The generation logic is to weight the rate of change with a preset weight of 0.6 and the fluctuation range with a preset weight of 0.4. The weights are obtained through prior experiments in the range of 0.1 to 0.9 with a step size of 0.1. During the experiment, the accuracy of distinguishing between known healthy bearings and known worn bearings is used as the evaluation. The distinction accuracy corresponding to the combination of weights 0.6 and 0.4 reaches 93.0%, which is higher than other combinations. In the example, a stable segment lasts for 120 seconds, with an initial amplitude of 0.18 m / s² and an ending amplitude of 0.30 m / s². The rate of change is 0.001 m / s², the maximum amplitude is 0.32 m / s², the minimum amplitude is 0.16 m / s², and the fluctuation range is 0.16 m / s². Substituting these two values ​​into the weighted synthesis logic, the amplitude offset of this segment is obtained as 0.065. An amplitude offset sequence is generated according to the segment order.

[0071] S202: Call the volume flow rate record data in the bearing operating status synchronization data packet, calculate the volume flow rate change range of the segment according to the time range corresponding to the speed segment, compare the volume flow rate difference change rate of adjacent time periods, and obtain the segment flow rate change rate sequence.

[0072] Volumetric flow rate was recorded point-by-point within each time interval, and the variation amplitude of volumetric flow rate in each interval was calculated. The variation amplitude was taken as the difference between the maximum and minimum volumetric flow rate in that interval. To avoid deviations caused by interpolation points, points with consecutive interpolation lengths exceeding 2 seconds were removed before calculation, retaining only the supplementary points and the original points with interpolation lengths not exceeding 2 seconds. When comparing the rate of change of volumetric flow rate difference between adjacent time intervals, the difference between the mean volumetric flow rates of two adjacent intervals was first calculated. The difference was taken as the mean of the later interval minus the mean of the earlier interval, and then the difference was divided by the time difference between the starting points of the two intervals to obtain the rate of change. The threshold setting was calibrated experimentally. Twelve sets of traction motor end cover bearing assemblies were selected and operated for 8 hours under ambient temperature of 25 degrees Celsius and rated pump speed of the cooling circuit. The distribution of volumetric flow rate variation amplitude was recorded. The variation amplitude of healthy samples was concentrated between 0.2 and 0.8 cubic meters per hour, while that of worn samples was concentrated between 1.0 and 2.4 cubic meters per hour. Based on this, the warning threshold for the rate of change of flow rate in the interval was set at a change of 0.6 cubic meters per hour. In the example, the volumetric flow rate of segment one fluctuates between 5.6 and 5.9 cubic meters per hour, with a variation range of 0.3 cubic meters per hour. The volumetric flow rate of segment two fluctuates between 4.9 and 5.4 cubic meters per hour, with a variation range of 0.5 cubic meters per hour. The average flow rate of segment two is 5.2 cubic meters per hour, and the average flow rate of segment one is 5.75 cubic meters per hour. The time difference between the starting points of the two segments is 6 minutes. Substituting the difference of 0.55 cubic meters per hour into the rate of change calculation logic yields 0.092 cubic meters per minute, which is recorded as the flow rate change rate of this pair of adjacent segments. Performing the above calculation on all segments and arranging them in segment order yields the sequence of segment flow rate change rates.

[0073] S203: Based on the amplitude offset sequence and the section flow rate change sequence, call the vibration limit standard and flow judgment threshold, identify and judge the joint exceedance of the two types of results in the section, mark the operating section with joint offset characteristics, and generate a degradation mark sequence.

[0074] When jointly judging the over-limit status based on the amplitude offset sequence and the section flow rate change sequence, two sets of judgment thresholds are first established. The vibration limit standard adopts the long-term operating boundary of the bearing housing vibration velocity. The interval boundary given by the publicly available engineering data is around 2.8 mm / s and 4.5 mm / s. When it exceeds 7.1 mm / s, it needs to be dealt with immediately. This interval boundary is used for the graded judgment in this embodiment. The temperature threshold uses the graded points of 80 degrees Celsius, 90 degrees Celsius, and 100 degrees Celsius as the setting reference for the division between normal, alarm and shutdown. The flow judgment threshold uses the section flow rate change rate of 0.6 cubic meters per minute per hour obtained by the aforementioned experimental calibration as the warning division. Subsequently, the amplitude deviation of each stable segment was converted into an out-of-limit marker under the same dimension. The conversion rule was to first use the upper quartile of the amplitude deviation of the healthy sample as the benchmark value. The benchmark value was obtained by statistically analyzing 12 sets of healthy samples across 200 segments. The upper quartile of the statistical result was set to 0.030. A segment amplitude deviation greater than 0.030 was judged as an out-of-limit vibration frequency amplitude deviation. The judgment of joint out-of-limit was based on the simultaneous fulfillment of two conditions in the same segment. The first condition was that the amplitude deviation exceeded the limit, and the second condition was that the segment flow rate change rate exceeded the limit or the segment volumetric flow rate change amplitude exceeded the limit. The threshold for exceeding the change amplitude was set to 1.0 cubic meters per hour, derived from the maximum change amplitude of 0.8 cubic meters per hour in the healthy sample plus a safety margin of 0.2 cubic meters per hour. In the example, if the amplitude offset of a certain segment is 0.065, which is greater than 0.030, and the volumetric flow rate change of that segment is 1.3 cubic meters per hour, which is greater than 1.0 cubic meters per hour, then that segment is identified as having a joint offset. If another segment has an amplitude offset of 0.028 and a flow rate change of 0.7 cubic meters per minute, then it is not identified because the amplitude offset does not exceed the limit. After identification, a degradation level identifier value is written for each segment. The degradation level is divided into three levels according to the degree of exceeding the limit: amplitude offset between 0.030 and 0.050 is classified as level 1, between 0.050 and 0.080 as level 2, and greater than 0.080 as level 3. At the same time, the flow rate change is classified as level 1 between 0.6 and 1.0 per minute, between 1.0 and 1.6 per minute as level 2, and greater than 1.6 per minute as level 3. The higher level is taken as the segment level, and a degradation marker sequence is generated according to the segment order.

[0075] Please see Figure 4 The specific steps of S3 are as follows:

[0076] S301: Based on the start and end times of the marked segments in the degraded marker sequence, continuously extract the marker points whose adjacent time span difference is less than the segment merging threshold, perform time segment splicing operation accordingly, integrate into multiple running intervals, call the unmarked segments as references, divide all intervals into non-overlapping segments, and obtain the running interval time series.

[0077] When performing time-span splicing and integration on marked segments based on the degradation marker sequence, the start and end times of each marked segment are first read and sorted by start time. Then, the time span difference between adjacent segments is calculated, where the difference is the difference between the start time of the later segment and the end time of the earlier segment. The segment merging threshold is determined by the task cycle and actual operating condition fluctuations, and is set to 30 seconds. The setting process involves comparing train traction operating condition logs and statistically analyzing common intervals during continuous traction and regenerative braking switching. The 80th percentile is approximately 25 seconds, and a 5-second margin is added to arrive at 30 seconds. For adjacent markers with a difference less than 30 seconds, a splicing operation is performed. The splicing rule is to merge two segments into one, with the start time of the merged segment taken as the start time of the earlier segment and the end time taken as the end time of the later segment. Simultaneously, the degradation level of the merged segment is taken as the maximum value of the two segments. In the example, the first labeled segment is from 09:00:00 to 09:02:00, and the second labeled segment is from 09:02:20 to 09:04:10, with an interval of 20 seconds to less than 30 seconds. After concatenation, the operating interval is obtained as 09:00:00 to 09:04:10. When using unlabeled segments as a reference for non-overlapping division, the entire time period of the synchronization data packets is first taken as the total time axis. Then, the concatenated operating interval is projected onto the total time axis. The remaining blank segments are divided into reference intervals according to the boundaries of adjacent intervals. During the division, it is ensured that the interval endpoints are aligned with the sampling timestamps, and the endpoint alignment rule is to round down to the nearest whole second. The complete set of all intervals, including labeled and reference intervals, is obtained, and the operating interval time series is output in chronological order. Each interval in the series records the start and end times, whether it is labeled, the labeling level, and the interval source identifier.

[0078] S302: Based on the time series of the running interval, call the degradation level identifier value corresponding to the segment in the degradation marker sequence, judge the proportion of degradation state in the running interval, classify the running interval into different state types according to the set state division ratio threshold, and obtain the running state classification label set.

[0079] When determining the proportion of degradation states and classifying state types based on the time series of operating intervals, the degradation marker sequence is first mapped to the second-by-second markers within each operating interval. The mapping rule is that for each whole second timestamp within the interval, its original stable segment marker is searched. If the second falls into the marked segment, the corresponding degradation level is written; otherwise, level 0 is written. Then, the proportion of each level within the interval is calculated. The proportion is obtained by counting the number of seconds for each level within the interval and dividing by the total number of seconds in the interval. The state classification ratio threshold is determined through experimental data. Accelerated wear tests are conducted on a test bench using bearing assemblies of the same model. During the wear phase, the bearings are disassembled and inspected every 2 hours, and the actual surface damage level is recorded. Simultaneously, the percentage of seconds corresponding to the degradation level for those 2 hours is saved. Statistical results show that when the sum of the proportions of levels 2 and 3 exceeds 0.30, the disassembly and inspection damage level reaches the interval requiring replacement; when the proportion of level 1 exceeds 0.40 and the proportions of levels 2 and 3 do not exceed 0.30, the disassembly and inspection damage is within the range where operation can continue; the remaining cases correspond to the healthy range. Based on this, the status types are divided into three categories: a severe status is defined as the sum of the proportions of Level 2 and Level 3 being greater than 0.30; a moderate status is defined as the proportion of Level 1 being ≥0.40 and the sum of the proportions of Level 2 and Level 3 not exceeding 0.30; and the rest are classified as mild status. In the example, a certain running interval lasts for 600 seconds, with 240 seconds marked as Level 1, 120 seconds as Level 2, 60 seconds as Level 3, and 180 seconds as Level 0. The proportion of Level 2 and Level 3 is 0.30, which, according to the rule, does not exceed 0.30 and is therefore not classified as severe. Furthermore, the proportion of Level 1 is 0.40, meeting the threshold, and the proportion of Level 2 and Level 3 is not exceeding 0.30, thus classifying it as a moderate status. After classifying all intervals, a running status classification label set is output. Each record in the label set includes the start and end times of the interval, the status type, the proportion of each level, and the total number of seconds used for review.

[0080] S303: Based on the operating status classification label set and the operating interval time series, the dwell time under each status label is counted, the sum of the dwell time of the status labels is normalized, and the bearing status distribution ratio vector is generated by recombining and arranging the labels according to the normalization ratio.

[0081] When calculating dwell time and generating a state distribution ratio vector based on the operational state classification label set, all operational intervals are first grouped by state type. The dwell time of each interval is then accumulated, with the dwell time being the sum of the interval's end time minus its start time in seconds. To avoid confusion with the distribution due to reference intervals, the intervals are filtered by their source identifier during accumulation, including only intervals determined by the degenerate marker sequence. Reference intervals are only used for non-overlapping divisions and do not participate in the distribution statistics. During normalization, the total dwell time of all states is read, and the dwell time of each state is divided by the sum to obtain the normalized ratio. In the example, the mild state accumulated 3600 seconds, the moderate state accumulated 1800 seconds, and the severe state accumulated 600 seconds, totaling 6000 seconds. After normalization, the ratios for mild, moderate, and severe states are 0.60, 0.30, and 0.10, respectively. A recombination and rearrangement operation is then performed, with the arrangement rules consistent with the state order of the subsequent lifespan consumption level table. The state order is fixed as mild, moderate, and severe, and the three proportions are written into a vector in this order. To ensure data flow consistency, the start and end times of the corresponding statistical window and the number of intervals are also written into this vector. The number of intervals is used for subsequent server-side consistency verification. A bearing state distribution proportion vector is generated, and the vector and statistical metadata are appended to the task cycle record as input for lifespan conversion and weighted calculation.

[0082] Please see Figure 5 The specific steps of S4 are as follows:

[0083] S401: Call the bearing state distribution ratio vector, calculate the life consumption ratio corresponding to each state ratio according to the life consumption level table corresponding to the state, map the state ratio to the consumption level, and summarize the state life consumption ratio in vector order to generate the state life consumption ratio vector.

[0084] When converting the bearing condition distribution ratio vector to the life consumption ratio, the life consumption level table is first read. The level table provides the corresponding life consumption ratio coefficient according to the condition type. The coefficient settings are obtained through bench life tests. Eighteen sets of bearing assemblies from the same batch were selected for the test. Under the same load and lubrication conditions, three combinations of light, medium, and heavy operating conditions were applied respectively, and the cumulative running time when the specified vibration velocity boundary was reached was recorded. Statistical results show that the average life under light operating conditions is in the range of 0.95 to 1.00 of the rated life, under medium operating conditions it is in the range of 0.60 to 0.75, and under heavy operating conditions it is in the range of 0.20 to 0.35. Based on this, the coefficient for light operating conditions is set to 0.05, the coefficient for medium operating conditions to 0.30, and the coefficient for heavy operating conditions to 0.70. The meaning of the coefficients is only used to map the condition ratio to the life consumption ratio, and the values ​​of the three coefficients ensure that the contribution of heavy operating conditions is higher than that of medium operating conditions, which is higher than that of light operating conditions, within the same time window. During the mapping calculation, a product operation is performed on each state. The state proportion is multiplied by the corresponding coefficient to obtain the lifetime consumption proportion component for that state, and these components are summarized in vector order to form the state lifetime consumption proportion vector. In the example, the state distribution proportion vectors are 0.60, 0.30, and 0.10. Substituting these into the mapping logic, we obtain a mild lifetime consumption proportion of 0.030, a moderate lifetime consumption proportion of 0.090, and a severe lifetime consumption proportion of 0.070. Summarizing these in order, we obtain the state lifetime consumption proportion vectors of 0.030, 0.090, and 0.070. This vector is also written to a verification field. The verification field is obtained by concatenating the three components into a string after rounding to the nearest thousandth, and is used for subsequent consistency comparison during upload, thus completing the generation of the state lifetime consumption proportion vector.

[0085] S402: Based on the state life consumption ratio vector, collect the cumulative running time record of the bearing, multiply the cumulative running time with the components of the state life consumption ratio vector respectively to obtain the life consumption time component corresponding to the state, and sum up the life consumption time components to obtain the cumulative life consumption time value.

[0086] When collecting cumulative runtime based on the state-life consumption ratio vector and calculating the weighted lifespan consumption time component, the cumulative runtime record of the bearing is first read from the traction motor operation log. The operation log is accumulated in seconds, and the data is truncated according to the end time of the same task cycle to avoid duplicate entries across cycles. Then, the cumulative runtime is weighted and calculated with the state-life consumption ratio vector. The weighted calculation performs a product operation on each component to obtain the corresponding lifespan consumption time component. Finally, all components are summed to obtain the cumulative lifespan consumption time value. In the example, the cumulative runtime is 1200 hours, and the state-life consumption ratio vectors are 0.030, 0.090, and 0.070. Substituting these values ​​into the weighting logic, we obtain a mild lifespan consumption time component of 36 hours, a moderate lifespan consumption time component of 108 hours, and a severe lifespan consumption time component of 84 hours, summing up to a cumulative lifespan consumption time value of 228 hours. To ensure this value comes from verifiable data, the step simultaneously outputs a list of components and the corresponding start and end times of the statistical window. It also removes downtime intervals appearing in the logs, with the rule being that intervals with continuous 0 RPM exceeding 300 seconds are not included in the cumulative runtime. If the cumulative runtime after removal is adjusted from 1200 hours to 1180 hours, it is then re-substituted into the weighting logic to obtain a cumulative lifetime consumption time value of 224.2 hours, which is then output to ensure that the cumulative lifetime consumption time value is consistent with the proportion of states in the same period.

[0087] S403: Based on the cumulative life consumption time value, obtain the bearing rated life reference time, compare the cumulative life consumption time with the rated life reference, calculate the remaining available running time range, and divide it into segments according to time continuity to generate available running time segments.

[0088] When calculating the remaining usable operating time range and dividing it into segments based on the cumulative lifespan consumption time, the bearing's rated lifespan reference time is first obtained. This reference time is taken from the unified value of the factory lifespan calibration record and bench verification record of the same model bearing; in this example, it is taken as 50,000 hours. Verification is performed on 6 sets of samples randomly selected from the same batch. The verification method is the average time corresponding to operation under rated load until the vibration velocity reaches the boundary of 7.1 mm / s. The average value falls within the range of 49,500 to 50,500 hours, consistent with 50,000 hours. During comparison, a difference operation is performed, subtracting the cumulative lifespan consumption time from the rated lifespan reference time to obtain the remaining usable operating time. In the example, the rated lifespan reference time is 50,000 hours, and the cumulative lifespan consumption time is 224.2 hours. Substituting these values ​​into the difference operation, the remaining usable operating time is obtained as 49,775.8 hours. When dividing sections based on time continuity, the remaining available running time is first mapped to future operating plans. These plans are estimated from daily train schedules and task cycles; for example, an average daily running time of 12 hours. The 49,775.8 hours are converted to approximately 4,147.98 days (approximately 4,148 days) at 12 hours / day, and the planned running time for each future day is written continuously according to the calendar. The section division rule uses the number of consecutive days of planned operation. If the interval between consecutive operating days is no more than one day, they are grouped into the same section. If there is a continuous shutdown exceeding one day, the section is broken and a new section is formed. In the example, if the train runs continuously for the next 10 days, stops on the 11th day, and resumes operation on the 12th day, then Section 1 will accumulate 120 hours over the next 10 days, and Section 2 will accumulate remaining time starting from the 12th day. Generate available runtime segments. Each segment records the start and end dates and times, available runtime and equivalent days, and writes 49,775.8 hours into the remaining fields of the first segment for subsequent lifetime range matching operations.

[0089] Please see Figure 6 The specific steps of S5 are as follows:

[0090] S501: Based on the available runtime segments and referring to the lifespan interval division standard, the duration value corresponding to the segment is matched with the upper and lower boundaries of the lifespan interval. The corresponding number is assigned according to the matching result, and all numbers are arranged and combined in chronological order to generate a lifespan segment number sequence.

[0091] When performing lifetime interval matching and generating a number sequence based on available runtime segments, the lifetime interval division standard is first read. The standard divides the remaining available runtime into 5 intervals. The interval boundaries are determined through maintenance strategies and bench risk statistics. Example boundaries are 0 to 5,000 hours, 5,000 to 12,500 hours, 12,500 to 25,000 hours, 25,000 to 37,500 hours, and 37,500 to 50,000 hours. The matching operation compares the remaining runtime value of each segment with the interval boundaries. The comparison rule is that if the remaining runtime of a segment falls into a certain interval, that interval is assigned a number. The numbers use Arabic numerals from 1 to 5 to correspond to the aforementioned intervals. In the example, the remaining runtime of the segment, 49,775.8 hours, falls into the 37,500 to 50,000 hour interval and is assigned the number 5. If the remaining duration of a subsequent task cycle is recalculated to be 37,480.6 hours, it falls within the range of 25,000 to 37,500 hours, is assigned the number 4, and the number change record is written into the sequence. To maintain the chronological order, a number record is written for each task cycle within the step. The record includes the task cycle end time and the number value. If multiple task cycles exist within the same day, they are sorted by timestamp. In the example, 2023-01-13 08:30 corresponds to number 5, 2023-01-13 20:30 corresponds to number 5, and 2023-01-20 20:30 corresponds to number 4. Therefore, the lifespan segment number sequence is output in chronological order as 5, 5, 4. To avoid abnormal jumps, a consistency check is performed. If the number increases by more than one level in two adjacent cycles, the input logs of the cumulative lifespan consumption time and the rated lifespan baseline time are checked for duplicate entries or missing entries. If the input is confirmed to be correct after the check, the jump record is retained and marked as a check point, and the lifespan segment number sequence is generated.

[0092] S502: Based on the lifespan segment number sequence, the bearing status distribution ratio vector is called to synchronously package the two types of information according to the current task cycle and transmit them to the vehicle controller host. An information encapsulation field is established inside the controller and uploaded to the ground maintenance server to obtain the remote status upload dataset.

[0093] When synchronously packaging and uploading bearing condition distribution ratio vectors based on lifespan segment number sequences, the latest lifespan segment number and corresponding condition distribution ratio vector are first retrieved within the current task cycle. These two types of information are then encapsulated according to a fixed field order: task cycle end time, lifespan segment number, mild ratio, moderate ratio, severe ratio, and a verification field. During encapsulation, each field is formatted: the time field uses a string to the second, the number field uses an integer, the ratio field uses a decimal string with three decimal places, and the verification field uses the aforementioned concatenated string. Subsequently, a transmission preparation operation is performed, writing the encapsulation result to the transmission buffer and generating a transmission batch number. The batch number consists of the date and the current day's sequence number; for example, 20230113 concatenated with 003 results in 20230113003. During transmission, the data is first written to the vehicle-side local record. After the record is written, the write-back content is read and compared field by field. Once consistency is confirmed, the data frame is sent to the remote end. To ensure reproducibility of the link, the sending timestamp, receiving acknowledgment timestamp, and retransmission count are simultaneously written within each step. The retransmission rule is to retransmit every 10 seconds if no acknowledgment is received, with a maximum of 3 retransmissions. In the example, a data frame is sent at the task cycle end time of 2023-01-13 20:30:00, with a lifespan segment number of 5, a ratio of 0.600, 0.300, and 0.100, and a checksum field of 030090070. After being written to the buffer, it is sent and an acknowledgment is received during the second retransmission. The acknowledged data frames are aggregated into a remote status upload dataset, which is archived by batch number and includes the field values ​​and acknowledgment flags for each upload record.

[0094] Table 2 Critical Thresholds and Experimental Calibration Results

[0095]

[0096] Table 2 lists the key thresholds used for screening, splicing, judgment, and grading in the embodiments, and provides sample size and calibration summary. Subsequent paragraphs directly reference the selected values ​​in the table when calling the thresholds.

[0097] S503: Based on the remote status upload dataset, call the preset field index rules and verification field generation rules of the ground maintenance server to perform field integrity verification and data format verification on the uploaded data, and recalculate the verification fields based on the uploaded life segment number sequence and bearing status distribution ratio vector, compare the consistency with the uploaded verification fields, and generate life status assessment codes.

[0098] When performing consistency comparison verification and generating evaluation codes based on the remote status upload dataset, a preset verification field index table is first read. This index table specifies the position order and length rules of each field in the uploaded record, and stipulates that the generation logic of the verification fields must be consistent with the vehicle-mounted enclosure. The verification operation performs field index matching on each uploaded record. First, it extracts the lifespan segment number, status ratio, and two other items according to the index table. Then, it recalculates the verification fields according to the same formatting rules. Subsequently, it performs a consistency comparison between the recalculated verification fields and the verification fields in the uploaded record. The comparison result generates a verification result code, which consists of three segments. The first segment indicates field completeness (1 for complete, 0 for incomplete). The second segment indicates verification field consistency (1 for consistent, 0 for inconsistent). The third segment indicates the reasonableness of the numerical range: 1 for lifespan segment numbers between 1 and 5, all three ratios between 0 and 1, and the sum of the three ratios between 0.995 and 1.005; otherwise, 0. In the example, a record numbered 5 has proportions of 0.600, 0.300, and 0.100. The verification fields are consistent, and the sum of the proportions is 1.000, resulting in a verification result code of 111. If another record has proportions of 0.650, 0.300, and 0.100, and the sum of the proportions is 1.050, exceeding 1.005, then the third segment is 0, the code is 110, and this record is added to the exception list. After generating codes for all records, the codes are associated with the corresponding task cycle end time to form a lifetime status assessment code sequence.

[0099] Table 3. Results of Evaluation Coding Verification and Comparison Experiments

[0100]

[0101] Table 3 shows the comparison results of uploaded data consistency verification. The results show that after adopting the verification code in this embodiment, the verification pass rate increased by 5.1 percentage points compared with the comparison group, the average time for anomaly location was reduced by 37 seconds, and the false alarm rate decreased from 6.8% to 2.1%, which verifies the effectiveness of lifetime status assessment code in data consistency verification.

[0102] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for assessing the health status and predicting the lifespan of traction motor bearings, characterized in that, Includes the following steps: S1: Acquire multi-source monitoring data of vibration signal, volumetric flow rate, speed and temperature of the bearing assembly inside the traction motor end cover during operation, align and integrate them after unifying the time base, and generate a synchronous data package of bearing operation status. S2: Divide the speed range according to the speed information in the bearing operation status synchronization data packet, calculate the rolling contact frequency based on the vibration signal, extract the amplitude change, calculate the vibration offset, count the corresponding flow change, and mark the degradation characteristic operation segment according to the joint judgment of vibration limit and flow threshold, and generate degradation mark sequence. S3: Based on the degradation marker sequence, integrate the continuous degradation segments and divide them into three state types: normal, degradation level one, and degradation level two. Calculate and normalize the residence time of the state intervals to generate a bearing state distribution ratio vector. S4: Call the bearing state distribution ratio vector, calculate the life consumption ratio according to the life consumption level table corresponding to the state, evaluate the life in combination with the cumulative running time of the bearing, and compare it with the rated life to generate the available running time segment. S5: Based on the available runtime segment matching lifespan interval standard, generate lifespan segment number, call the vehicle controller host to upload lifespan number and status distribution information to the ground maintenance server, and the ground maintenance server performs consistency comparison and verification to generate lifespan status assessment code.

2. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 1, characterized in that, The bearing operating status synchronization data package includes vibration signal data, volumetric flow rate data, speed change records, temperature change records, and time alignment identifiers. The degradation marker sequence includes speed range number, vibration degradation discrimination marker, flow anomaly discrimination marker, joint discrimination result marker, and degradation operating segment identifier. The bearing status distribution ratio vector includes normal state ratio, degradation level 1 state ratio, degradation level 2 state ratio, normalized dwell time ratio, and state range statistical benchmark. The available operating time segment includes life consumption ratio, cumulative operating time benchmark, rated life comparison margin, available time start point, and available time end point. The life status assessment code includes life segment number, status distribution information verification identifier, server consistency verification result code, life status level code, and uploaded record index.

3. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Acquire vibration signals, volumetric flow rate, speed changes and temperature records of the bearing assembly inside the traction motor end cover during operation, identify and classify the data according to the data acquisition time, remove data whose time interval exceeds the period benchmark, and obtain the screened monitoring dataset. S102: Based on the sieved monitoring dataset, time interpolation is performed on the volumetric flow rate, rotation speed and temperature data to complete the time order of the data, and data with synchronization offset exceeding the sampling period are removed to obtain the aligned monitoring data sequence. S103: Based on the multiple types of data in the aligned monitoring data sequence, a vector sequence is constructed by combining them according to a unified time reference to form a continuous time matrix, and the data is archived and organized according to the sensor type to generate a synchronous data package of bearing operating status.

4. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 3, characterized in that, The limitation of removing data whose time interval exceeds the period benchmark is as follows: taking the sampling period corresponding to the data acquisition time as the period benchmark, if the time interval between any two adjacent vibration signals, volume flow rate, rotation speed change and temperature records is greater than 1.2 times the sampling period, the corresponding data is determined to be invalid data and removed from the screened monitoring dataset. Among them, 1.2 times is the time jitter tolerance margin of the sampling period, which is used to distinguish between normal acquisition jitter and sampling gaps, and to avoid mistaking cross-period missing data as continuous valid data; The time interpolation completion of volumetric flow rate, rotation speed and temperature data is limited to: based on the timestamps of adjacent valid data, interpolation is used to generate interpolated data at missing time points, and the interpolation time interval is not greater than the sampling period; The exclusion of data whose synchronization offset exceeds the sampling period is limited to the following: when the offset between the timestamp of the volume flow rate, rotation speed and temperature data and the timestamp of the vibration signal is greater than the sampling period, the corresponding data will not participate in the combination and construction of the vector sequence.

5. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 3, characterized in that, The specific steps of S2 are as follows: S201: Based on the rotational speed record in the bearing operating status synchronization data packet, divide the bearing into multiple stable segments according to the continuous change trend, extract the vibration signal amplitude sequence corresponding to the rolling contact frequency in each segment, call the rate of change and fluctuation range of the vibration signal amplitude in the segment, calculate the frequency amplitude offset index, and generate the amplitude offset sequence. S202: Call the volume flow rate record data in the bearing operating status synchronization data packet, statistically analyze the volume flow rate change range of the segment according to the time range corresponding to the speed segment, compare the volume flow rate difference change rate of adjacent time periods, and obtain the segment flow rate change rate sequence. S203: Based on the amplitude offset sequence and the section flow rate change sequence, call the vibration limit standard and flow judgment threshold, identify and judge the joint exceedance of the two types of results in the section, mark the operating section with joint offset characteristics, and generate a degradation mark sequence.

6. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 5, characterized in that, The specific definition of dividing multiple stable segments according to the continuous change trend is as follows: taking the speed record in the bearing operation status synchronization data packet as the benchmark, when the change amplitude of adjacent speed records does not exceed the preset speed fluctuation threshold within three consecutive sampling periods, the corresponding time range is divided into the same stable segment. The frequency amplitude offset index is specifically defined as follows: based on the vibration signal amplitude sequence corresponding to the rolling contact frequency, the maximum value, minimum value and amplitude change rate of the vibration signal amplitude in the stable section are respectively counted, and the frequency amplitude offset index is generated by weighting the difference between the change rate and the maximum and minimum values ​​according to a preset weight. The specific definition of identifying and judging the joint exceedance of the two types of results within the section is as follows: when the frequency amplitude deviation index continuously exceeds the vibration limit standard and the section flow change rate within the corresponding stable section continuously exceeds the flow judgment threshold, the corresponding stable section is marked as an operating section with joint deviation characteristics.

7. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 5, characterized in that, The specific steps for S3 are as follows: S301: Based on the start and end times of the marked segments in the degraded marker sequence, continuously extract the marker points whose adjacent time span difference is less than the segment merging threshold, perform time segment splicing operation accordingly, integrate into multiple running intervals, call the unmarked segments as references, divide all intervals into non-overlapping segments, and obtain the running interval time series. S302: Based on the time series of the running interval, call the degradation level identifier value corresponding to the segment in the degradation marker sequence, judge the proportion of degradation state in the running interval, classify the running interval into different state types according to the set state division ratio threshold, and obtain the running state classification label set; S303: Based on the set of operating status classification labels and the time series of operating intervals, the dwell time under each status label is statistically calculated, the sum of the dwell time of the status labels is normalized, and the labels are rearranged according to the normalization ratio to generate the bearing status distribution ratio vector.

8. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 7, characterized in that, The specific steps of S4 are as follows: S401: Call the bearing state distribution ratio vector, calculate the life consumption ratio corresponding to each state ratio according to the life consumption level table corresponding to the state, map the state ratio to the consumption level, and summarize the state life consumption ratio in vector order to generate the state life consumption ratio vector. S402: Based on the state life consumption ratio vector, collect the cumulative running time record of the bearing, multiply the cumulative running time with the components of the state life consumption ratio vector respectively to obtain the life consumption time component corresponding to the state, and sum and summarize the life consumption time components to obtain the cumulative life consumption time value. S403: Based on the cumulative life consumption time value, obtain the bearing rated life reference time, compare the cumulative life consumption time with the rated life reference, calculate the remaining available running time range, and divide it into segments according to time continuity to generate available running time segments.

9. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 8, characterized in that, The specific steps of S5 are as follows: S501: Based on the available runtime segments, and referring to the lifespan interval division standard, the duration value corresponding to the segment is matched with the upper and lower boundaries of the lifespan interval. The corresponding number is assigned according to the matching result, and all numbers are arranged and combined in chronological order to generate a lifespan segment number sequence. S502: Based on the lifespan segment number sequence, call the bearing status distribution ratio vector, synchronously package the two types of information according to the current task cycle, transmit them to the vehicle controller host, establish information encapsulation fields inside the controller, and upload them to the ground maintenance server to obtain the remote status upload dataset. S503: Based on the remote status upload dataset, call the preset field index rules and verification field generation rules of the ground maintenance server to perform field integrity verification and data format verification on the uploaded data, and recalculate the verification fields based on the uploaded life segment number sequence and bearing status distribution ratio vector, compare the consistency with the uploaded verification fields, and generate life status assessment codes.

10. The method for assessing the health status and predicting the lifespan of traction motor bearings according to claim 9, characterized in that, The specific definition of matching the duration value corresponding to the segment with the upper and lower boundaries of the lifespan interval is as follows: comparing the duration value corresponding to the available runtime segment with the upper and lower boundaries of two adjacent lifespan intervals in the lifespan interval division standard; when the duration value falls between the corresponding upper and lower boundaries, it is determined that the available runtime segment is matched to the corresponding lifespan interval. The specific definition of assigning a corresponding number identifier based on the matching result is as follows: a unique integer number is pre-set for each of the lifespan intervals, and the corresponding integer number is written into the lifespan interval number sequence according to the interval matching result; The specific limitation of matching the content of the corresponding position in the uploaded data by field index is as follows: based on the field order index defined in the preset verification field of the ground maintenance server, extract the life segment number sequence content and bearing status distribution ratio vector content of the corresponding index position in the remote status uploaded data one by one, and perform consistency comparison verification under the condition that the index is consistent.

Images