An electric drive axle temperature rise early warning and service life prediction method and system

Abstract

The application provides an electric drive axle temperature rise early warning and service life prediction method and system, belonging to the electric drive axle monitoring technical field. The method: collects electric drive axle node temperature data and operating parameter data; calculates temperature rise speed and standard deviation based on temperature data, compares with threshold set to extract temperature rise fluctuation characteristics; analyzes the phase relationship of each node temperature fluctuation based on temperature data, calibrates the optimal zero threshold and outputs the phase characteristics; constructs and drives the digital twin model based on the collected data, simulates the electric drive axle operating state and generates simulation data; uses the collected data and simulation data to train and optimize the threshold set, the optimal zero threshold calibration process and the service life prediction model; based on the temperature rise fluctuation, the phase characteristics and the electric drive axle operating state, the threshold set and the optimal zero threshold are applied for vehicle end instant early warning and the service life prediction model is applied to predict the remaining service life of the electric drive axle. The application early warns temperature rise abnormalities, improves the electric drive axle operating safety and service life.

Description

A method and system for early warning and life prediction of electric drive bridge temperature rise Technical Field

[0001] This application belongs to the field of electric drive bridge monitoring technology, specifically relating to a method and system for early warning of temperature rise and life prediction of electric drive bridge. Background Technology

[0002] As a key power transmission component of a vehicle, the electric drive axle's operating condition directly affects the vehicle's performance, efficiency, and safety. During operation, the electric drive axle experiences temperature rise due to factors such as current flow, load changes, and mechanical friction. Excessive temperature rise can lead to problems such as accelerated aging of insulation materials, lubricant failure, metal material deformation, and changes in mechanical clearances. These issues reduce the electric drive axle's operating efficiency, shorten its service life, and in severe cases, even cause sudden malfunctions, endangering driving safety.

[0003] Currently, most existing electric drive axle temperature rise monitoring systems rely on fixed thresholds for early warning. These systems typically collect temperature data using temperature sensors placed at key locations, triggering an alarm when the measured value exceeds a preset fixed threshold. However, this simple threshold comparison method has the following limitations: First, fixed thresholds cannot adapt to the limitations of electric drive axles or dynamic changes under different conditions, leading to inappropriate early warning timing, often resulting in delayed responses or frequent false alarms. Second, traditional methods struggle to identify early, subtle anomalies during the temperature rise process, which are often precursors to potential failures. Third, existing systems generally lack the ability to predict the lifespan of electric drive axles, failing to provide effective data support for preventative maintenance.

[0004] In recent years, digital twin technology and deep learning algorithms have received widespread attention in the field of industrial equipment condition monitoring. Digital twin technology can construct virtual mappings of physical entities, enabling real-time simulation of operating conditions; deep learning, on the other hand, has demonstrated powerful capabilities in feature extraction and pattern recognition. However, existing technologies have not yet effectively solved the challenges of early warning and lifespan prediction in electric drive bridge temperature rise monitoring, particularly lacking systematic solutions in multi-source data fusion, dynamic threshold adjustment, and weak feature recognition. Summary of the Invention

[0005] In a first aspect, embodiments of this application provide a method for early warning of temperature rise and prediction of lifespan of an electric drive bridge, comprising the following steps:

[0006] S1. Set several monitoring positions as nodes on the electric drive bridge, periodically collect temperature data of each node of the electric drive bridge to obtain temperature time series data of each node, and at the same time collect phase data and operating parameter data of the electric drive bridge.

[0007] S2. Based on the temperature time series data of each node, calculate the temperature rise rate and the standard deviation of the temperature rise rate of each node, compare the standard deviation of the temperature rise rate with the threshold set generated based on historical fault data, and extract the temperature rise fluctuation characteristics based on the comparison results.

[0008] S3. Based on the temperature time series data of each node, analyze the correspondence between temperature fluctuation and phase change of each node, dynamically calibrate the optimal zero-adjustment threshold, and output the phase characteristics of temperature fluctuation between each node based on the optimal zero-adjustment threshold.

[0009] S4. Based on the collected temperature data, phase data and operating parameter data of each node of the electric drive bridge, construct and drive the digital twin model of the electric drive bridge, simulate the operating status of the electric drive bridge in real time and generate simulation data.

[0010] S5. Using the historical values ​​of temperature data, phase data, and operating parameter data of each node of the electric drive bridge, as well as the simulation data generated by the digital twin model of the electric drive bridge, optimize the threshold set, the optimal zero-adjustment threshold calibration process, and train and optimize the pre-built life prediction model.

[0011] S6. Based on the temperature rise fluctuation characteristics, phase characteristics, and the operating status of the electric drive axle simulated by the digital twin model of the electric drive axle, an optimized threshold set and the optimal zero-adjustment threshold are applied to provide real-time vehicle-side warning of the temperature rise of the electric drive axle.

[0012] S7. Based on the temperature rise fluctuation characteristics, phase characteristics, the operating status of the electric drive bridge simulated by the digital twin model of the electric drive bridge, and the optimized life prediction model, predict the remaining service life of the electric drive bridge.

[0013] Furthermore, the specific steps of step S1 are as follows:

[0014] S11. Use the positions of the motor windings, bearings, and gearbox housing of the electric drive bridge as nodes for monitoring, and install temperature sensors;

[0015] S12. Use each temperature sensor to synchronously collect temperature data of each node, and synchronously collect the speed, load torque, working voltage and working current of the electric drive bridge as operating parameter data.

[0016] S13. The collected data is treated as an analog signal and converted into a digital signal using an A / D converter;

[0017] S14. The converted digital signal is transmitted to the central processing unit via the data bus to form temperature timing data and operating parameter dataset.

[0018] Furthermore, the specific steps of step S2 are as follows:

[0019] S21. Preprocess the real-time temperature time series data, using moving average filtering to remove noise and detect and correct abnormal sampled values;

[0020] S22. Based on the preprocessed temperature sequence and corresponding time series Calculate the instantaneous temperature rise rate :

[0021]

[0022] in, and These are the temperature values ​​of the i-th and (i-1)-th sampling points, respectively. and The sampling time is for the i-th and (i-1)-th sampling points;

[0023] S23. Calculate the real-time temperature rise rate sequence within a sliding window of length N. Standard deviation of the rate of temperature rise :

[0024]

[0025] in, This represents the average rate of temperature rise within the window;

[0026] S24. Standard deviation of temperature rise rate With a set of thresholds pre-generated based on historical fault data Compare the results and output the temperature rise fluctuation characteristics:

[0027] like If so, the output temperature rise fluctuation is normal;

[0028] like Then, the first-level early warning feature for temperature rise fluctuations will be output;

[0029] like Then, the output will show the characteristics of a level-two early warning for temperature rise fluctuations;

[0030] like Then, the output will show the three-level early warning characteristics of temperature rise fluctuation.

[0031] Furthermore, the threshold set Generate via the following offline steps:

[0032] Obtain a historical temperature rise dataset with fault labels;

[0033] Calculate the standard deviation of the temperature rise rate for each data point in the historical temperature rise dataset;

[0034] Divide the standard deviation of the temperature rise rate of all historical temperature rise data into continuous intervals;

[0035] For the Calculate the failure probability using a temperature rise rate standard deviation interval. :

[0036]

[0037] in, Let be the number of samples that failed within the k-th standard deviation interval of the temperature rise rate. is the total number of samples within the k-th standard deviation interval of the temperature rise rate;

[0038] Sort the standard deviation intervals of each temperature rise rate in ascending order of value, and calculate the cumulative probability of failure:

[0039] The standard deviation corresponding to the first time the cumulative probability of failure reaches or exceeds the first proportional threshold is set as the first-level warning threshold. ;

[0040] The threshold for the second proportional threshold will be set as the level 2 warning threshold when it is first reached or exceeded. ;

[0041] The threshold for a Level 3 warning will be set when the threshold is first reached or exceeded. .

[0042] Furthermore, the specific steps of step S3 are as follows:

[0043] S31. Use the electric drive bridge speed signal as a synchronization reference to perform time alignment of the temperature timing data of each node.

[0044] S32. Filter and detrend the aligned temperature time series data of each node to obtain the fluctuation signal. ;

[0045] S33. Use Fourier transform to transform the wave signal. Transform to the frequency domain and extract the instantaneous phase. ;

[0046] S34. Taking the motor winding temperature node as reference node i, calculate the instantaneous phase difference between node j and reference node i. ;

[0047] S35. Calculate the average phase difference between nodes based on historical data of the electric drive bridge's health status. As an inherent phase bias ;

[0048] S36. Select to set or calibrate the optimal zeroing threshold;

[0049] If not set, proceed to step S37;

[0050] If it is for calibration, proceed to step S38;

[0051] S37. Calculate the standard deviation of the phase difference sequence of health data. Standard deviation based on health data phase difference sequence Set the optimal zeroing threshold ;

[0052] Where K is a constant selected based on the characteristics of the normal distribution;

[0053] S38. Based on the inherent phase offset Determine the initial zeroing threshold based on the preset initial proportional tolerance;

[0054] During operation, the current instantaneous phase difference and the inherent phase offset are calculated in real time. Real-time deviation;

[0055] When the preset period or event triggering condition is met, the current zeroing threshold is iteratively adjusted based on the real-time deviation.

[0056] When the real-time deviation meets the preset convergence condition, the current zeroing threshold is determined as the optimal zeroing threshold. ;

[0057] S39. Calculate the original instantaneous phase difference based on real-time data. With inherent phase bias Perform zeroing calibration:

[0058]

[0059] in, The corrected phase difference;

[0060] like If the condition is met, it is determined to be a valid phase deviation feature, and a warning message is output.

[0061] Furthermore, the specific steps of step S4 are as follows:

[0062] S41. Construct a three-dimensional digital model that is consistent with the geometric parameters, materials and connection relationships of the physical electric drive bridge, and add temperature simulation logic and motion simulation logic to generate a digital twin model of the electric drive bridge;

[0063] S42. Receive the collected temperature, speed, and phase data of the electric drive bridge in real time through the data interface, and drive the digital twin model of the electric drive bridge to run synchronously;

[0064] S43. Perform temperature field simulation in the digital twin model of the electric drive bridge, and use the finite element analysis method to calculate the internal temperature distribution of the electric drive bridge based on real-time data;

[0065] S44. Perform kinematic simulation in the digital twin model of the electric drive bridge, and simulate the motion state of each component of the electric drive bridge based on phase and speed data;

[0066] S45. By using the working condition simulation function, different combinations of load and speed parameters are set to simulate the operating state of the electric drive axle under different preset working conditions, including full-load climbing and high-speed driving, and generate simulation data. The generation frequency of the simulation data is consistent with the actual sampling frequency.

[0067] Furthermore, the specific steps of step S5 are as follows:

[0068] S51. Optimize the threshold set:

[0069] Based on the acquired historical operating data and simulated operating data, the first neural network model is trained with the standard deviation sequence of temperature rise rate under different fault types as input and the verified optimal threshold corresponding to each fault type as the expected output, so as to generate a threshold set that can be dynamically adjusted according to real-time temperature rise characteristics.

[0070] S52. Optimize the optimal zero-adjustment threshold calibration process:

[0071] Based on the phase characteristics of each node obtained by analysis and the historical data of the health status of the electric drive bridge, the standard deviation of the phase difference sequence under the health status is calculated offline.

[0072] Based on the standard deviation of the phase difference sequence under healthy conditions, the optimal zeroing threshold is calculated and determined by setting a fixed multiple.

[0073] S53. Constructing and optimizing the lifetime prediction model:

[0074] Based on the acquired historical operating data and simulated operating condition data, the second long short-term memory network model is trained with the operating time series, temperature rise data, phase data and maintenance records of the electric drive bridge as input and the actual lifespan as output, in order to build and optimize the lifespan prediction model and control the prediction error within the set error range.

[0075] Furthermore, step S6 is detailed as follows:

[0076] S61. Based on the operating status simulated by the temperature rise fluctuation characteristics, phase characteristics, and digital twin model of the electric drive bridge, and applying the optimized threshold set and the optimal zero-adjustment threshold, multi-level early warning judgment is performed.

[0077] S62. When the conditions for a first-level warning are met, a first-level warning is triggered, and an intermittent alert sound at the first frequency is emitted through the audible and visual alarm device;

[0078] S63. When the conditions for a second-level warning are met, a second-level warning is triggered, and an intermittent prompt sound at a second frequency is emitted through an audible and visual alarm device, supplemented by a first-level voice prompt message; the second frequency is higher than the first frequency;

[0079] S64. When the conditions for a Level 3 warning are met, a Level 3 warning is triggered, and a continuous alert is issued through the audible and visual alarm device, supplemented by Level 2 voice prompts.

[0080] The first-level warning condition is any of the following:

[0081] Temperature fluctuations exceeded the Level 1 warning threshold.

[0082] The phase characteristic deviates from the optimal zeroing threshold to the first preset range;

[0083] The operating parameters simulated by the digital twin model exceed the first normal range threshold.

[0084] The conditions for a Level 2 warning are any of the following:

[0085] Temperature fluctuations exceeded the level-two warning threshold.

[0086] The phase characteristic deviates from the optimal zero-adjustment threshold and reaches the second preset range;

[0087] The operating parameters simulated by the digital twin model exceed the second normal range threshold.

[0088] The conditions for a Level 3 warning are any of the following:

[0089] Temperature fluctuations exceed the level-three warning threshold;

[0090] The phase characteristic deviates from the optimal zero-adjustment threshold and reaches the third preset range;

[0091] The operating parameters simulated by the digital twin model exceed the third normal range threshold.

[0092] Furthermore, the deviations of the first preset range, the second preset range, and the third preset range increase sequentially, and the strictness of the first normal range threshold, the second normal range threshold, and the third normal range threshold increases sequentially.

[0093] Furthermore, the specific steps of step S7 are as follows:

[0094] S71. Obtain simulation data of temperature rise fluctuation characteristics, phase characteristics, and digital twin model of electric drive bridge, and input them into the optimized life prediction model;

[0095] S72. The optimized life prediction model uses deep learning to fit the degradation curve, predicts the remaining life of the electric drive bridge, and displays the predicted remaining life according to a preset time unit.

[0096] S73. When the predicted remaining service life is less than a preset time threshold, it is highlighted on the vehicle display screen, and reference information on the impact of usage intensity on service life is provided.

[0097] Secondly, embodiments of this application also provide an electric drive bridge temperature rise early warning and life prediction system, including:

[0098] The data acquisition module is used to set several monitoring positions as nodes on the electric drive bridge, periodically collect temperature data of each node of the electric drive bridge, obtain temperature time series data of each node, and at the same time collect phase data and operating parameter data of the electric drive bridge.

[0099] The weak feature extraction module is used to calculate the temperature rise rate and temperature rise rate standard deviation of each node based on the temperature time series data of each node, compare the temperature rise rate standard deviation with the threshold set generated based on historical fault data, and extract temperature rise fluctuation features based on the comparison results.

[0100] The multi-node phase analysis module is used to analyze the correspondence between temperature fluctuations and phase changes of each node based on the temperature time series data of each node, dynamically calibrate the optimal zero-adjustment threshold, and output the phase characteristics of temperature fluctuations between each node based on the optimal zero-adjustment threshold.

[0101] The digital twin module is used to build and drive a digital twin model of the electric drive bridge based on the collected temperature data, phase data and operating parameter data of each node of the electric drive bridge, to simulate the operating status of the electric drive bridge in real time and generate simulation data.

[0102] The deep learning module is used to optimize the threshold set and the optimal zero-adjustment threshold calibration process using historical values ​​of temperature data, phase data and operating parameter data of each node of the electric drive bridge and simulation data generated by the digital twin model of the electric drive bridge, as well as to train and optimize the pre-built life prediction model.

[0103] The early warning module is used to provide real-time vehicle-side early warning of the electric drive axle temperature rise based on the temperature rise fluctuation characteristics, phase characteristics, and the operating status of the electric drive axle simulated by the digital twin model of the electric drive axle, and to apply an optimized threshold set and the optimal zero-adjustment threshold.

[0104] The life prediction module is used to predict the remaining lifespan of the electric drive bridge based on the temperature rise fluctuation characteristics, phase characteristics, the operating status of the electric drive bridge simulated by the digital twin model of the electric drive bridge, and the optimized life prediction model.

[0105] As can be seen from the above technical solutions, this application has the following advantages:

[0106] The electric drive axle temperature rise early warning and life prediction method and system provided in this application achieve accurate and real-time early warning of electric drive axle temperature rise and accurate prediction of remaining lifespan through multi-dimensional data acquisition, feature extraction, digital twin simulation, and deep learning optimization, effectively improving the operational safety and reliability of the electric drive axle. Based on historical data and real-time simulation data, the threshold set and optimal zero-adjustment threshold are dynamically adjusted to adapt to different operating conditions and electric drive axle states, avoiding the limitations of traditional fixed threshold methods and reducing false alarms and missed alarms. By extracting temperature rise fluctuation characteristics and phase characteristics, early weak abnormal characteristics in the temperature rise process can be identified, potential faults can be detected in a timely manner, and support for preventative maintenance. Through early warning and life prediction, maintenance plans can be rationally arranged, reducing downtime caused by sudden failures, improving vehicle operating efficiency, and ensuring driving safety. Attached Figure Description

[0107] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0108] Figure 1 is a flowchart illustrating the electric drive bridge temperature rise early warning and life prediction method of the present invention.

[0109] Figure 2 is a schematic diagram of the electric drive bridge temperature rise early warning and life prediction system of the present invention. Detailed Implementation

[0110] The various embodiments of this disclosure will be described more fully in the detailed steps of the electric drive bridge temperature rise early warning and life prediction method described below. This disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of this disclosure to the specific embodiments disclosed herein, but rather this disclosure should be understood to cover all adjustments, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of this disclosure.

[0111] This embodiment provides a method for early warning and life prediction of electric drive axle temperature rise. Through multi-dimensional data collection and analysis, it can accurately warn of abnormal temperature rise in electric drive axle, dynamically adjust thresholds to adapt to different operating conditions, identify fault characteristics in advance, extend service life, and improve vehicle operation safety and reliability.

[0112] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0113] Please refer to Figure 1, which is a flowchart of a method for early warning and life prediction of electric drive bridge temperature rise in a specific embodiment. The method includes the following steps:

[0114] S1. Set several monitoring positions as nodes on the electric drive bridge, periodically collect temperature data of each node of the electric drive bridge to obtain temperature time series data of each node, and at the same time collect phase data and operating parameter data of the electric drive bridge.

[0115] It should be noted that by setting monitoring nodes at key locations on the electric drive bridge, temperature data and operating parameter data are collected to provide a data source for subsequent analysis;

[0116] S2. Based on the temperature time series data of each node, calculate the temperature rise rate and the standard deviation of the temperature rise rate of each node, compare the standard deviation of the temperature rise rate with the threshold set generated based on historical fault data, and extract the temperature rise fluctuation characteristics based on the comparison results.

[0117] It should be noted that by calculating the temperature rise rate and its standard deviation and comparing it with the threshold set, abnormal fluctuations in the temperature rise process can be effectively identified, providing key features for temperature rise early warning; the use of moving average filtering to remove noise and detect and correct abnormal sampling values ​​improves the stability and reliability of the data and enhances the accuracy of feature extraction.

[0118] S3. Based on the temperature time series data of each node, analyze the correspondence between temperature fluctuation and phase change of each node, dynamically calibrate the optimal zero-adjustment threshold, and output the phase characteristics of temperature fluctuation between each node based on the optimal zero-adjustment threshold.

[0119] It should be noted that by analyzing the phase relationship of temperature fluctuations at each node and dynamically calibrating the optimal zero-adjustment threshold, phase deviation characteristics can be identified, expanding the dimensions of fault feature extraction. The introduction of phase characteristics makes the health status assessment of the electric drive bridge more comprehensive and improves the sensitivity and accuracy of fault identification.

[0120] S4. Based on the collected temperature data, phase data and operating parameter data of each node of the electric drive bridge, construct and drive the digital twin model of the electric drive bridge, simulate the operating status of the electric drive bridge in real time and generate simulation data.

[0121] It should be noted that by constructing a digital twin model consistent with the physical electric drive bridge, the operating status of the electric drive bridge can be simulated in real time, enabling virtual verification of temperature rise warning and lifespan prediction.

[0122] S5. Using the historical values ​​of temperature data, phase data, and operating parameter data of each node of the electric drive bridge, as well as the simulation data generated by the digital twin model of the electric drive bridge, optimize the threshold set, the optimal zero-adjustment threshold calibration process, and train and optimize the pre-built life prediction model.

[0123] It should be noted that deep learning optimization of the threshold set, the optimal zero-adjustment threshold calibration process, and the lifetime prediction model improved the model's performance and prediction accuracy. Through training the neural network model, intelligent dynamic threshold adjustment and lifetime prediction were achieved, improving the overall performance and reliability of the system.

[0124] S6. Based on the temperature rise fluctuation characteristics, phase characteristics, and the operating status of the electric drive axle simulated by the digital twin model of the electric drive axle, an optimized threshold set and the optimal zero-adjustment threshold are applied to provide real-time vehicle-side warning of the temperature rise of the electric drive axle.

[0125] It should be noted that, based on the characteristics of temperature rise fluctuations, phase characteristics, and the operating status of the digital twin model, multi-level early warning judgments can be made, and graded responses can be made according to the degree of anomaly.

[0126] S7. Based on the temperature rise fluctuation characteristics, phase characteristics, the operating status of the electric drive bridge simulated by the digital twin model of the electric drive bridge, and the optimized life prediction model, predict the remaining service life of the electric drive bridge.

[0127] It should be noted that, based on the optimized life prediction model, the remaining service life of the electric drive axle can be accurately predicted, providing an intuitive reference for vehicle maintenance; the prediction results can help to rationally arrange maintenance plans, reduce unexpected failures, and improve vehicle operating efficiency and safety.

[0128] This embodiment achieves accurate and real-time early warning of electric drive axle temperature rise and accurate prediction of remaining service life through multi-dimensional data collection, feature extraction, digital twin simulation and deep learning optimization; dynamically adjusts thresholds to adapt to different operating conditions, reduces false alarms and missed alarms, detects potential faults in advance, provides strong support for preventive maintenance, and improves vehicle operating efficiency and safety.

[0129] Furthermore, as a refinement and extension of the specific implementation methods of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, a certain new energy commercial vehicle electric drive axle is used as the application object. This electric drive axle has a rated power of 150kW and a rated speed of 3000rpm, and is suitable for urban delivery and short-to-medium distance transportation scenarios. The core components include a permanent magnet synchronous motor, a two-stage reduction gearbox, and key components of tapered roller bearings. Another method for temperature rise warning and life prediction of electric drive axles is provided, which includes the following steps:

[0130] S1. Set several monitoring locations as nodes on the electric drive bridge, periodically collect temperature data of each node to obtain the temperature time series data of each node, and simultaneously collect the phase data and operating parameter data of the electric drive bridge; the specific steps of step S1 are as follows:

[0131] S11. Use the positions of the motor windings, bearings, and gearbox housing of the electric drive bridge as nodes for monitoring, and install temperature sensors;

[0132] For example, six key locations are selected as temperature monitoring nodes: the motor windings of the electric drive bridge (one monitoring point for each of the three phase windings), the front bearing, the rear bearing, and the gearbox housing (one monitoring point each at the input shaft end and the output shaft end). Each node is equipped with a high-precision platinum resistance temperature sensor (model PT1000, measurement range -50℃~200℃, accuracy ±0.1℃). The sensor probe is fixed to the monitoring surface with thermally conductive adhesive, and an outer heat-insulating protective sleeve is wrapped to prevent airflow interference. Simultaneously, a speed sensor (Hall effect type, measurement range 0~5000rpm, accuracy ±1rpm) and a torque sensor (strain gauge type, measurement range 0~1000N) are also deployed. m, accuracy ±0.5%FS), voltage sensor (Hall type, measurement range 0~800V, accuracy ±0.2% FS), current sensor (closed-loop Hall type, measurement range 0~500A, accuracy ±0.3% FS), to collect the speed, load torque, working voltage and working current parameters of the electric drive bridge during operation;

[0133] S12. Use each temperature sensor to synchronously collect temperature data of each node, and synchronously collect the speed, load torque, working voltage and working current of the electric drive bridge as operating parameter data.

[0134] For example, a data acquisition card (model NI cDAQ-9178) is used to synchronously acquire data from various sensors. The acquisition frequency is set to 10Hz to ensure the consistency of timestamps between temperature time-series data and operating parameter data. The analog signal output by the temperature sensor is converted into a digital signal by the built-in A / D converter (16-bit resolution) of the acquisition card. Differential input is used during the conversion process to suppress common-mode interference. The digital signal from the operating parameter sensor is directly connected through the digital interface of the acquisition card to achieve parallel acquisition of multi-source data.

[0135] S13. The collected data is treated as an analog signal and converted into a digital signal using an A / D converter;

[0136] S14. The converted digital signal is transmitted to the central processing unit via the data bus to form temperature time series data and operating parameter dataset;

[0137] For example, the converted digital signal is transmitted to the vehicle central processing unit (model STM32H743) via CAN bus (transmission rate 250kbps). After the processor timestamps the data, it forms a temperature time series dataset (format: [timestamp, motor winding U-phase temperature, motor winding V-phase temperature, ..., gearbox output shaft end housing temperature]), a phase dataset (format: [timestamp, instantaneous phase value of each node, phase difference between nodes]), and an operating parameter dataset (format: [timestamp, speed, load torque, operating voltage, operating current]).

[0138] S2. Based on the temperature time-series data of each node, calculate the temperature rise rate and standard deviation of the temperature rise rate for each node, compare the standard deviation of the temperature rise rate with the threshold set generated based on historical fault data, and extract the temperature rise fluctuation characteristics based on the comparison results; the specific steps of step S2 are as follows:

[0139] S21. Preprocess the real-time temperature time series data, using moving average filtering to remove noise and detect and correct abnormal sampled values;

[0140] For example, a moving average filter is used with a window size of 5 sampling points to remove noise; when the temperature rise data of a certain sampling point deviates from the average value of the adjacent 5 sampling points by more than 3 times the standard deviation, it is judged as an outlier and corrected by linear interpolation.

[0141] For example, preprocessing is performed on the real-time acquired temperature time-series data:

[0142] High-frequency noise is removed using a moving average filter with a window size of 5 sampling points. The filtering formula is as follows: (i≥3, edge points are processed using symmetrical expansion) Outlier detection and correction adopt the 3σ criterion. When the temperature value of a certain sampling point... satisfy (in It is the average of 5 adjacent sampling points. When the value is less than the corresponding standard deviation, it is considered an outlier and corrected using linear interpolation. ;

[0143] S22. Based on the preprocessed temperature sequence and corresponding time series Calculate the instantaneous temperature rise rate :

[0144]

[0145] in, and These are the temperature values ​​of the i-th and (i-1)-th sampling points, respectively. and The sampling time is for the i-th and (i-1)-th sampling points;

[0146] For example, based on the preprocessed temperature sequence and the corresponding time sequence, the instantaneous temperature rise rate of each node is calculated; taking the U-phase temperature of the motor winding as an example, the formula for calculating the instantaneous temperature rise rate of the i-th sampling point is:

[0147]

[0148] in, and The values ​​are the preprocessed temperature values ​​of the i-th and (i-1)-th sampling points, respectively, and the sum is the corresponding sampling time (in seconds). The calculation result is rounded to two decimal places and is in °C / s.

[0149] S23. Calculate the real-time temperature rise rate sequence within a sliding window of length N. Standard deviation of the rate of temperature rise :

[0150]

[0151] in, This represents the average rate of temperature rise within the window;

[0152] For example, the sliding window length N=20 (corresponding to 2 seconds of data) is set, and the standard deviation of the temperature rise rate sequence within the window is calculated in real time; taking the U phase of the motor winding as an example, the formula for calculating the standard deviation is:

[0153]

[0154] in, The average temperature rise rate within the window is calculated at the same frequency as the data acquisition frequency (10Hz).

[0155] S24. Standard deviation of temperature rise rate With a set of thresholds pre-generated based on historical fault data Compare the results and output the temperature rise fluctuation characteristics:

[0156] like If so, the output temperature rise fluctuation is normal;

[0157] like Then, the first-level early warning feature for temperature rise fluctuations will be output;

[0158] like Then, the output will show the characteristics of a level-two early warning for temperature rise fluctuations;

[0159] like Then, the output will show the three-level early warning characteristics of temperature rise fluctuation;

[0160] The threshold set Generate via the following offline steps:

[0161] Obtain a historical temperature rise dataset with fault labels;

[0162] Calculate the standard deviation of the temperature rise rate for each data point in the historical temperature rise dataset;

[0163] Divide the standard deviation of the temperature rise rate of all historical temperature rise data into continuous intervals;

[0164] For the Calculate the failure probability using a temperature rise rate standard deviation interval. :

[0165]

[0166] in, Let be the number of samples that failed within the k-th standard deviation interval of the temperature rise rate. is the total number of samples within the k-th standard deviation interval of the temperature rise rate;

[0167] Sort the standard deviation intervals of each temperature rise rate in ascending order of value, and calculate the cumulative probability of failure:

[0168] The standard deviation corresponding to the first time the cumulative probability of failure reaches or exceeds the first proportional threshold (e.g., 10%) is set as the first-level warning threshold. ;

[0169] The threshold for a secondary warning is set when the threshold is first reached or exceeded (e.g., 30%). ;

[0170] The threshold for a Level 3 warning will be set when the threshold is first reached or exceeded (e.g., 50%). ;

[0171] For example, firstly, 5 years of historical fault data for this model of electric drive axle were obtained (including 1200 sets of normal operation data, 300 sets of minor fault data, 150 sets of moderate fault data, and 80 sets of severe fault data). For each data point, the standard deviation of the temperature rise rate at each node was calculated, and all standard deviations were divided into 11 continuous intervals: 0~0.05, 0.05~0.10, ..., 0.50~0.55. The fault probability for each interval was calculated: taking the 0.10~0.15℃ / s interval as an example, the total number of samples in this interval... =180, the number of samples that failed. =18, then the probability of failure The cumulative probability of faults is calculated by sorting them by standard deviation from smallest to largest. When the cumulative probability first reaches 10%, the corresponding standard deviation of 0.12℃ / s is set as the first-level warning threshold. When the threshold first reaches 30%, the corresponding standard deviation of 0.25℃ / s is set as the level two warning threshold. When the threshold first reaches 50%, the corresponding standard deviation of 0.40℃ / s is set as the level three warning threshold. This forms a threshold set [0.12, 0.25, 0.40]; when extracting temperature rise fluctuation features in real time, the standard deviation of the temperature rise rate in the current window is used. Compared with the threshold set: If If <0.12, output "Temperature rise fluctuation is normal"; if Output "Level 1 Early Warning Characteristics of Temperature Fluctuation"; if Output "Level 2 Early Warning Characteristics of Temperature Fluctuation"; if Output "Level 3 Early Warning Characteristics of Temperature Fluctuation";

[0172] S3. Based on the temperature time-series data of each node, analyze the correspondence between temperature fluctuations and phase changes at each node, dynamically calibrate the optimal zero-adjustment threshold, and output the phase characteristics of temperature fluctuations between each node based on the optimal zero-adjustment threshold; the specific steps of step S3 are as follows:

[0173] S31. Use the electric drive bridge speed signal as a synchronization reference to perform time alignment of the temperature timing data of each node.

[0174] For example, the electric drive bridge speed signal is used as a synchronization reference to time-align the timing data of 6 temperature nodes; using the pulse signal output by the speed sensor, 60 pulses are generated per revolution, and the rising edge of the pulse is used as the time synchronization trigger point to map the sampling data of each temperature node onto a unified time axis. The time alignment accuracy is controlled within ±1ms to ensure the time correlation of temperature fluctuations at each node.

[0175] S32. Filter and detrend the aligned temperature time series data of each node to obtain the fluctuation signal. ;

[0176] For example, the aligned temperature time series data for each node is filtered and detrended. First, a Butterworth low-pass filter (cutoff frequency 0.5Hz) is used to remove low-frequency trend terms, and then a linear fitting method is used to remove the linear trend in the data, resulting in a signal containing only fluctuation components. (i=1~6, corresponding to 6 temperature nodes);

[0177] S33. Use Fourier transform to transform the wave signal. Transform to the frequency domain and extract the instantaneous phase. ;

[0178] S34. Taking the motor winding temperature node as reference node i, calculate the instantaneous phase difference between node j and reference node i. ;

[0179] For example, a Fast Fourier Transform (FFT) is used to transform the wave signal. The frequency domain is transformed, the main frequency components within the frequency range of 0.1~1Hz are extracted, and then the instantaneous phase of each main frequency component is calculated using Hilbert transform. Using the U-phase temperature node of the motor winding (j=1) as the reference node i, calculate the instantaneous phase difference between other nodes j (j=2~6) and the reference node: ;

[0180] S35. Calculate the average phase difference between nodes based on historical data of the electric drive bridge's health status. As an inherent phase bias ;

[0181] S36. Select to set or calibrate the optimal zeroing threshold;

[0182] If not set, proceed to step S37;

[0183] If it is for calibration, proceed to step S38;

[0184] S37. Calculate the standard deviation of the phase difference sequence of health data. Standard deviation based on health data phase difference sequence Set the optimal zeroing threshold ;

[0185] Where K is a constant selected based on the characteristics of the normal distribution, for example, a value of 3;

[0186] For example, based on historical health data of this type of electric drive axle (100 sets of fault-free operation data, each set lasting 2 hours), the average phase difference of each node to the reference node is calculated as 0.2 rad, which is used as the inherent phase offset; at the same time, the standard deviation of the health data phase difference sequence is calculated as 0.05 rad; if the optimal zero-adjustment threshold is selected, then the optimal zero-adjustment threshold is set. =3×0.05rad=0.15rad (K=3 is a constant selected based on the characteristics of normal distribution); if the optimal zero-adjustment threshold is selected, then based on the inherent phase offset =0.2rad and the preset initial proportional tolerance of 5%, the initial zero-adjustment threshold is determined to be 0.2rad×(1+5%)=0.21rad; during operation, the real-time deviation between the current instantaneous phase difference and the inherent phase offset is calculated in real time, and the preset period is set to 10 minutes. Every time the period is reached, the current zero-adjustment threshold is iteratively adjusted according to the real-time deviation; when the real-time deviation of 3 consecutive iterations is less than 0.001rad (satisfying the preset convergence condition), the current zero-adjustment threshold of 0.148rad is determined as the optimal zero-adjustment threshold;

[0187] S38. Based on the inherent phase offset Determine the initial zeroing threshold based on the preset initial proportional tolerance;

[0188] During operation, the current instantaneous phase difference and the inherent phase offset are calculated in real time. Real-time deviation;

[0189] When the preset period or event triggering condition is met, the current zeroing threshold is iteratively adjusted based on the real-time deviation.

[0190] When the real-time deviation meets the preset convergence condition, the current zeroing threshold is determined as the optimal zeroing threshold. ;

[0191] S39. Calculate the original instantaneous phase difference based on real-time data. With inherent phase bias Perform zeroing calibration:

[0192]

[0193] in, The corrected phase difference;

[0194] like If the condition is met, it is determined to be a valid phase deviation feature, and a warning message is output.

[0195] For example, based on real-time data, the original instantaneous phase difference is calculated to be 0.32 rad, and zero-adjustment correction is performed: = 0.32 rad - 0.2 rad = 0.12 rad; if If the value is 0.12 rad > the optimal zero-adjustment threshold of 0.15 rad (setting method) or 0.148 rad (calibration method), it is determined to be a valid phase deviation feature (format: [motor winding U phase - front bearing, 0.12 rad, moderate deviation]), and a warning message is output; otherwise, it is determined to be normal, and no feature is output.

[0196] S4. Based on the collected temperature data, phase data, and operating parameter data of each node of the electric drive bridge, construct and drive a digital twin model of the electric drive bridge to simulate the operating status of the electric drive bridge in real time and generate simulation data; the specific steps of step S4 are as follows:

[0197] S41. Construct a three-dimensional digital model that is consistent with the geometric parameters, materials and connection relationships of the physical electric drive bridge, and add temperature simulation logic and motion simulation logic to generate a digital twin model of the electric drive bridge;

[0198] For example, a three-dimensional digital model of the electric drive bridge is constructed using SolidWorks software. The geometric parameters of the model are completely consistent with the physical entity, including key parameters such as motor stator / rotor dimensions, gear module and number of teeth, bearing type, and housing thickness. The material properties are set to match the actual components (e.g., the motor windings are copper, the housing is aluminum alloy, and the gears are 20CrMnTi). Temperature simulation logic (using the transient thermal analysis module) and motion simulation logic (using the dynamic analysis module) are added to the model in ANSYS Workbench, and the two modules are called collaboratively through Python scripts to generate a complete digital twin model of the electric drive bridge.

[0199] S42. Receive the collected temperature, speed, and phase data of the electric drive bridge in real time through the data interface, and drive the digital twin model of the electric drive bridge to run synchronously;

[0200] For example, a data interface between the digital twin model and the vehicle central processing unit is established through the TCP / IP protocol to receive the collected temperature, speed, and phase data in real time (transmission delay ≤ 50ms). The temperature data is mapped to the temperature boundary conditions of the corresponding nodes in the model, the speed data is used as the input load for dynamic simulation, and the phase data is used to correct the relative motion relationship of the components in the model, so as to realize the synchronous operation of the digital twin model and the physical electric drive axle (synchronization error ≤ 0.1s).

[0201] S43. Perform temperature field simulation in the digital twin model of the electric drive bridge, and use the finite element analysis method to calculate the internal temperature distribution of the electric drive bridge based on real-time data;

[0202] S44. Perform kinematic simulation in the digital twin model of the electric drive bridge, and simulate the motion state of each component of the electric drive bridge based on phase and speed data;

[0203] For example, in the temperature field simulation, the finite element analysis method is used to divide the mesh (the mesh size of the motor winding is 2mm and the mesh size of the shell is 5mm). Based on real-time temperature data and the thermal conductivity of the material, the three-dimensional temperature distribution inside the electric drive bridge is calculated, and the instantaneous temperature cloud map and temperature gradient data of each component are output. In the kinematic simulation, based on the phase and speed data, the motion states such as gear meshing and bearing rotation are simulated, and the velocity, acceleration and contact stress distribution data of each component are output, providing multi-dimensional reference for fault diagnosis.

[0204] S45. By using the working condition simulation function, different combinations of load and speed parameters are set to simulate the operating state of the electric drive axle under different preset working conditions, including full-load climbing and high-speed driving, and generate simulation data. The generation frequency of the simulation data is consistent with the actual sampling frequency.

[0205] For example, five preset operating conditions are set (no-load idling, rated load constant speed, full-load climbing, high-speed driving, and frequent start-stop), and each operating condition includes 10 different combinations of load and speed parameters (such as a full-load climbing condition with a load torque of 800~1000N). m, speed 1000~1500rpm). The electric drive bridge operating status under various working conditions is simulated by a digital twin model to generate simulation data (including temperature time series data, phase data, and operating parameter data). The simulation data generation frequency is consistent with the actual sampling frequency (10Hz). A total of 100,000 sets of simulation data are generated, which together with historical operating data constitute the model training dataset.

[0206] S5. Using the historical values ​​of temperature data, phase data, and operating parameter data of each node of the electric drive bridge, as well as the simulation data generated by the digital twin model of the electric drive bridge, optimize the threshold set and the optimal zero-adjustment threshold calibration process, and train and optimize the pre-built life prediction model; the specific steps of step S5 are as follows:

[0207] S51. Optimize the threshold set:

[0208] Based on the acquired historical operating data and simulated operating data, the first neural network model is trained with the standard deviation sequence of temperature rise rate under different fault types as input and the verified optimal threshold corresponding to each fault type as the expected output, so as to generate a threshold set that can be dynamically adjusted according to real-time temperature rise characteristics.

[0209] The specific training process of the first neural network model used to optimize the threshold set:

[0210] Model Architecture: The first neural network model is a multilayer perceptron (MLP). The number of nodes in its input layer corresponds to the dimension of the input features, and the output layer has 3 nodes, which correspond to the optimized first-level, second-level, and third-level warning thresholds, respectively. The middle layer contains two hidden layers with 64 and 32 neurons respectively, using the ReLU activation function.

[0211] Training data and input / output: The training data comes from historical operating data and digital twin simulation data; the input features are the statistical features of the temperature rise rate standard deviation sequence extracted from historical data under different fault types, including mean, variance, skewness and peak value; the expected output labels are the calculated set of thresholds that have been verified as optimal by domain experts.

[0212] Training parameters: Mean squared error (MSE) was used as the loss function, Adam optimizer was used, the initial learning rate was set to 0.001, the batch size was 32, and the number of training iterations was 500.

[0213] For example, a multilayer perceptron (MLP) is constructed as the first neural network model. The input layer has 4 nodes (corresponding to the mean, variance, skewness, and peak value of the temperature rise rate standard deviation sequence), the hidden layer has 2 layers (64 and 32 neurons, ReLU activation function), and the output layer has 3 nodes (corresponding to the optimized first, second, and third level warning thresholds). The training data consists of 5000 sets of historical fault data and 3000 sets of simulated working condition data. The input features are the statistical features of the temperature rise rate standard deviation sequence under different fault types, and the expected output is the optimal threshold verified by domain experts. The training process uses mean squared error (MSE) as the loss function, Adam optimizer (initial learning rate 0.001, batch size 32), and 500 training iterations to finally obtain the optimized threshold set (e.g., [0.11, 0.23, 0.38]), which can be dynamically adjusted according to real-time temperature rise characteristics.

[0214] S52. Optimize the optimal zero-adjustment threshold calibration process:

[0215] Based on the phase characteristics of each node obtained by analysis and the historical data of the health status of the electric drive bridge, the standard deviation of the phase difference sequence under the health status is calculated offline.

[0216] Based on the standard deviation of the phase difference sequence under healthy conditions, the optimal zeroing threshold is calculated and determined by setting a fixed multiple.

[0217] For example, when optimizing the optimal zero-adjustment threshold calibration process, based on the phase characteristics of each node output by the multi-node phase analysis module (including instantaneous phase, inter-node phase difference, etc.) and the historical data of the electric drive bridge's health status (a complete phase dataset of 50 sets of fault-free operation), the standard deviation of the phase difference sequence under healthy conditions is calculated offline to be 0.045 rad; a fixed multiple K=3 is set, and the formula is used... =3 × 0.045 rad = 0.135 rad, thus completing the optimization of the calibration process;

[0218] S53. Constructing and optimizing the lifetime prediction model:

[0219] Based on the acquired historical operating data and simulated operating condition data, the second long short-term memory network model is trained with the operating time series, temperature rise data, phase data and maintenance records of the electric drive bridge as input and the actual lifespan as output, in order to build and optimize the lifespan prediction model and control the prediction error within a set error range (e.g., 5%).

[0220] The specific training process of the second long short-term memory network model used for lifespan prediction:

[0221] Model architecture: The second long short-term memory network model is a stacked LSTM network; it consists of two stacked LSTM layers, with the first LSTM layer having 100 units and the second layer having 50 units, followed by a fully connected layer with 50 neurons, and finally an output layer to predict the remaining lifetime.

[0222] Training data and input / output: The input is a multi-dimensional time series within a time window, specifically including: the running time of the electric drive bridge, temperature rise fluctuation characteristics, phase characteristics, and maintenance records encoded as values; the expected output label is the actual remaining service life of the electric drive bridge after the end of the time window.

[0223] Training parameters: Mean squared error (MSE) was used as the loss function, the Adam optimizer was employed, and the learning rate was 0.001. The training objective was to keep the model's prediction error within 5%. To prevent overfitting, an early stopping strategy was adopted during training, and a dropout method was used with a dropout rate of 0.2.

[0224] For example, a stacked LSTM network is constructed as the second long short-term memory network model. The input is multi-dimensional data within the time window (running time, temperature rise fluctuation characteristics of each node, phase characteristics, maintenance record encoding value), the first layer LSTM (100 units), the second layer LSTM (50 units), the fully connected layer (50 neurons), and the output layer (1 node, corresponding to the remaining lifetime).

[0225] S6. Based on the temperature rise fluctuation characteristics, phase characteristics, and the operating status of the electric drive axle simulated by the digital twin model of the electric drive axle, and applying an optimized threshold set and the optimal zero-adjustment threshold, a real-time vehicle-side warning is issued for the temperature rise of the electric drive axle; the specific steps of step S6 are as follows:

[0226] S61. Based on the operating status simulated by the temperature rise fluctuation characteristics, phase characteristics, and digital twin model of the electric drive bridge, and applying the optimized threshold set and the optimal zero-adjustment threshold, multi-level early warning judgment is performed.

[0227] For example, three levels of early warning conditions are set:

[0228] Level 1 warning conditions (minor anomalies): Temperature rise fluctuation characteristics exceed the Level 1 warning threshold (0.11℃ / s); or phase characteristics deviate from the optimal zeroing threshold by 1%~3% (e.g., front bearing phase deviation 0.15×(1%~3%)=0.0015~0.0045rad); or the operating state parameters simulated by the digital twin model (e.g., gear contact stress) exceed the first normal range threshold (150~200MPa).

[0229] Level II warning conditions (moderate anomaly): Temperature rise fluctuation characteristics exceed the Level II warning threshold (0.23℃ / s); or phase characteristics deviate from the optimal zeroing threshold by 3%~5% (0.0045~0.0075rad); or digital twin model simulation parameters exceed the second normal range threshold (200~250MPa).

[0230] Level 3 warning conditions (severe anomaly): Temperature rise fluctuation characteristics exceed the Level 3 warning threshold (0.38℃ / s); or phase characteristics deviate from the optimal zeroing threshold by 5% or more (≥0.0075rad); or digital twin model simulation parameters exceed the third normal range threshold (≥250MPa).

[0231] S62. When the conditions for a first-level warning are met, a first-level warning is triggered, and an intermittent alert sound at the first frequency is emitted through the audible and visual alarm device;

[0232] S63. When the conditions for a second-level warning are met, a second-level warning is triggered, and an intermittent warning tone at a second frequency is emitted through an audible and visual alarm device, supplemented by a first-level voice prompt (e.g., a voice prompt "The electric drive bridge is malfunctioning, please pay attention"); the second frequency is higher than the first frequency;

[0233] S64. When the conditions for a Level 3 warning are met, a Level 3 warning is triggered, and a continuous alert is issued through the audible and visual alarm device, supplemented by Level 2 voice prompts (e.g., a voice prompt saying "High risk of electric drive bridge failure, please stop the car immediately for inspection").

[0234] The first-level warning condition is any of the following:

[0235] Temperature fluctuations exceeded the Level 1 warning threshold.

[0236] The phase characteristics deviate from the optimal zeroing threshold to the first preset range (e.g., 1%-3%).

[0237] The operating parameters simulated by the digital twin model exceed the first normal range threshold.

[0238] The conditions for a Level 2 warning are any of the following:

[0239] Temperature fluctuations exceeded the level-two warning threshold.

[0240] The phase characteristics deviate from the optimal zeroing threshold to the second preset range (e.g., 3%-5%).

[0241] The operating parameters simulated by the digital twin model exceed the second normal range threshold.

[0242] The conditions for a Level 3 warning are any of the following:

[0243] Temperature fluctuations exceed the level-three warning threshold;

[0244] The phase characteristic deviates from the optimal zeroing threshold to the third preset range (e.g., 5%).

[0245] The operating parameters simulated by the digital twin model exceed the third normal range threshold.

[0246] Furthermore, the deviation of the first preset range, the second preset range, and the third preset range increases sequentially, and the strictness of the first normal range threshold, the second normal range threshold, and the third normal range threshold increases sequentially.

[0247] For example, the warning module integrates an audible and visual alarm device (model LTE-1101J) and an in-vehicle display screen (10.2-inch TFT LCD). When the first-level warning condition is met, the display screen shows a yellow warning icon and the text "Slight abnormal temperature rise," and the audible and visual alarm device emits an intermittent warning sound at a frequency of 1Hz (0.5s on, 0.5s off). When the second-level warning condition is met, the display screen shows an orange warning icon and the text "Electric drive axle malfunction, please pay attention," and the audible and visual alarm device emits an intermittent warning sound at a frequency of 2Hz. At the same time, the corresponding voice prompt is played through the voice module (model SYN6288). When the third-level warning condition is met, the display screen shows a red warning icon and flashes, the audible and visual alarm device emits a continuous warning sound (volume ≥85dB), and the voice module loops the message "High risk of electric drive axle failure, please stop immediately for inspection." All warning information is synchronously uploaded to the cloud server for real-time monitoring by the Fleet management backend.

[0248] S7. Based on the temperature rise fluctuation characteristics, phase characteristics, the operating status of the electric drive bridge simulated by the digital twin model of the electric drive bridge, and the optimized life prediction model, predict the remaining service life of the electric drive bridge; the specific steps of step S7 are as follows:

[0249] S71. Obtain simulation data of temperature rise fluctuation characteristics, phase characteristics, and digital twin model of electric drive bridge, and input them into the optimized life prediction model;

[0250] For example, the temperature rise fluctuation features output by the weak feature extraction module, the phase features output by the multi-node phase analysis module, and the simulation data (temperature distribution, contact stress, etc.) generated by the digital twin model are acquired in real time. After the data is standardized (mean is zeroed and variance is normalized), it is input into the optimized LSTM lifetime prediction model.

[0251] S72. The optimized life prediction model uses deep learning to fit the degradation curve, predicts the remaining life of the electric drive bridge, and displays the predicted remaining life according to a preset time unit.

[0252] For example, the model uses a deep learning algorithm to curve fit the degradation process of the electric drive axle, and combines historical degradation data and real-time status data to predict the remaining service life (in hours). For instance, when the electric drive axle has been running for 3000 hours, the current temperature rise rate standard deviation is 0.18℃ / s, the phase deviation is 0.003rad, and the digital twin model simulates a gear contact stress of 180MPa, the model predicts a remaining service life of 2200 hours. The prediction result is rounded to the nearest integer and displayed on the "Maintenance Reminder" interface of the vehicle display screen.

[0253] S73. When the predicted remaining service life is less than a preset time threshold (e.g., 100 hours), highlight it on the vehicle display screen and provide reference information on the impact of usage intensity on service life;

[0254] For example, a preset time threshold is set to 100 hours; when the predicted remaining service life is less than 100 hours, the "Maintenance Reminder" interface on the vehicle display turns red and flashes, while displaying the message "Less than 100 hours remaining, it is recommended to stop maintenance as soon as possible"; in addition, based on the impact data of different usage intensities on service life simulated by the digital twin model, reference suggestions are given (such as "If the current full load condition is maintained, the remaining service life will be shortened by 30%; switching to the rated load condition can extend the remaining service life by 15%), providing a basis for users to formulate maintenance plans.

[0255] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0256] As shown in Figure 2, the following is an embodiment of the electric drive bridge temperature rise warning and life prediction system provided in this disclosure. This system and the electric drive bridge temperature rise warning and life prediction methods in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the electric drive bridge temperature rise warning and life prediction system, please refer to the embodiments of the above electric drive bridge temperature rise warning and life prediction methods.

[0257] The system includes:

[0258] The data acquisition module is used to set several monitoring positions as nodes on the electric drive bridge, periodically collect temperature data of each node of the electric drive bridge, obtain temperature time series data of each node, and at the same time collect phase data and operating parameter data of the electric drive bridge.

[0259] The weak feature extraction module is used to calculate the temperature rise rate and temperature rise rate standard deviation of each node based on the temperature time series data of each node, compare the temperature rise rate standard deviation with the threshold set generated based on historical fault data, and extract temperature rise fluctuation features based on the comparison results.

[0260] The multi-node phase analysis module is used to analyze the correspondence between temperature fluctuations and phase changes of each node based on the temperature time series data of each node, dynamically calibrate the optimal zero-adjustment threshold, and output the phase characteristics of temperature fluctuations between each node based on the optimal zero-adjustment threshold.

[0261] The digital twin module is used to build and drive a digital twin model of the electric drive bridge based on the collected temperature data, phase data and operating parameter data of each node of the electric drive bridge, to simulate the operating status of the electric drive bridge in real time and generate simulation data.

[0262] The deep learning module is used to optimize the threshold set and the optimal zero-adjustment threshold calibration process using historical values ​​of temperature data, phase data and operating parameter data of each node of the electric drive bridge and simulation data generated by the digital twin model of the electric drive bridge, as well as to train and optimize the pre-built life prediction model.

[0263] The early warning module is used to provide real-time vehicle-side early warning of the temperature rise of the electric drive axle based on the temperature rise fluctuation characteristics, phase characteristics, and the operating status of the electric drive axle simulated by the digital twin model of the electric drive axle, and to apply an optimized set of thresholds and the optimal zero-adjustment threshold.

[0264] The life prediction module is used to predict the remaining lifespan of the electric drive bridge based on the temperature rise fluctuation characteristics, phase characteristics, the operating status of the electric drive bridge simulated by the digital twin model of the electric drive bridge, and the optimized life prediction model.

[0265] This embodiment improves the efficiency and reliability of electric drive bridge monitoring through the interactive collaboration of the data acquisition module, weak feature extraction module, multi-node phase analysis module, digital twin module, deep learning module, early warning module, and life prediction module, thus ensuring the safe operation of the electric drive bridge.

[0266] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for early warning of temperature rise and prediction of lifespan of an electric drive bridge, characterized in that, The steps include: S1. Set several monitoring positions as nodes on the electric drive bridge, periodically collect temperature data of each node of the electric drive bridge to obtain temperature time series data of each node, and at the same time collect phase data and operating parameter data of the electric drive bridge; S2. Based on the temperature time series data of each node, calculate the temperature rise rate and standard deviation of the temperature rise rate of each node, compare the standard deviation of the temperature rise rate with the threshold set generated based on historical fault data, and extract the temperature rise fluctuation characteristics based on the comparison results; S3. Based on the temperature time series data of each node, analyze the correspondence between temperature fluctuation and phase change of each node, dynamically calibrate the optimal zero-adjustment threshold, and output the phase characteristics of temperature fluctuation between each node based on the optimal zero-adjustment threshold; Step S3 is as follows: S31. Using the electric drive bridge speed signal as a synchronization reference, time-align the temperature timing data of each node; S32. Filter and detrend the aligned temperature timing data of each node to obtain the fluctuation signal. S33. Use Fourier transform to convert the wave signal. Transform to the frequency domain and extract the instantaneous phase. S34. Using the motor winding temperature node as reference node i, calculate the instantaneous phase difference between node j and reference node i. S35. Based on historical data of the health status of the electric drive bridge, calculate the average phase difference between nodes. As an inherent phase bias S36. Select to set or calibrate the optimal zeroing threshold; if setting, proceed to step S37; if calibrating, proceed to step S38; S37. Calculate the standard deviation of the health data phase difference sequence. Standard deviation based on health data phase difference sequence Set the optimal zeroing threshold Wherein, K is a constant selected based on the characteristics of the normal distribution; S38. Based on the inherent phase offset Determine the initial zeroing threshold based on the preset initial proportional tolerance; During operation, the current instantaneous phase difference and the inherent phase offset are calculated in real time. The real-time deviation is calculated; when the preset period or event triggering condition is met, the current zeroing threshold is iteratively adjusted based on the real-time deviation; when the real-time deviation meets the preset convergence condition, the current zeroing threshold is determined as the optimal zeroing threshold. ； S39. Calculate the original instantaneous phase difference based on real-time data. With inherent phase bias Perform zero adjustment and calibration: in, The phase difference is the corrected value; if If the temperature rise fluctuation feature is detected, it is determined to be a valid phase deviation feature, and an early warning message is output; S4. Based on the collected temperature data, phase data, and operating parameter data of each node of the electric drive axle, a digital twin model of the electric drive axle is constructed and driven to simulate the operating state of the electric drive axle in real time and generate simulation data; S5. Using the historical values ​​of the collected temperature data, phase data, and operating parameter data of each node of the electric drive axle, and the simulation data generated by the digital twin model of the electric drive axle, the threshold set and the optimal zero-adjustment threshold calibration process are optimized, and the pre-constructed life prediction model is trained and optimized; S6. Based on the temperature rise fluctuation feature, phase feature, and the operating state of the electric drive axle simulated by the digital twin model of the electric drive axle, the optimized threshold set and the optimal zero-adjustment threshold are applied to provide real-time vehicle-side early warning for the temperature rise of the electric drive axle; S7. Based on the temperature rise fluctuation feature, phase feature, the operating state of the electric drive axle simulated by the digital twin model of the electric drive axle, and the optimized life prediction model, the remaining service life of the electric drive axle is predicted.

2. The method for early warning and life prediction of electric drive bridge temperature rise according to claim 1, characterized in that, Step S1 is as follows: S11. Use the positions of the motor windings, bearings, and gearbox housing of the electric drive bridge as monitoring nodes and set temperature sensors; S12. Use each temperature sensor to synchronously collect temperature data from each node, and synchronously collect the speed, load torque, operating voltage, and operating current of the electric drive bridge as operating parameter data; S13. Use the collected data as analog signals and convert them into digital signals through an A / D converter; S14. Transmit the converted digital signals to the central processing unit through the data bus to form temperature timing data and operating parameter dataset.

3. The method for early warning and life prediction of electric drive bridge temperature rise according to claim 1, characterized in that, Step S2 is as follows: S21. Preprocess the real-time temperature time series data, using a moving average filter to remove noise and detect and correct abnormal sampled values; S22. Based on the preprocessed temperature sequence... and corresponding time series Calculate the instantaneous temperature rise rate ： in, and These are the temperature values ​​of the i-th and (i-1)-th sampling points, respectively. and S23. Within a sliding window of length N, calculate the real-time temperature rise rate sequence. Standard deviation of temperature rise rate ： in, S24. The average rate of temperature rise within the window; The standard deviation of the rate of temperature rise. With a set of thresholds pre-generated based on historical fault data Compare the results and output the temperature rise fluctuation characteristics: If If the output temperature rise fluctuation is normal, then the output temperature rise fluctuation is normal; if Then, output the first-level warning feature for temperature rise fluctuation; if Then, output the secondary warning characteristics of temperature rise fluctuation; if Then, the output will show the three-level early warning characteristics of temperature rise fluctuation.

4. The method for early warning and life prediction of electric drive bridge temperature rise according to claim 3, characterized in that, The threshold set The following offline steps are used to generate the dataset: Obtain a historical temperature rise dataset with fault labels; calculate the standard deviation of the temperature rise rate for each data point in the historical temperature rise dataset; divide the standard deviation of the temperature rise rate of all historical temperature rise data into continuous intervals; and then... Calculate the failure probability using a temperature rise rate standard deviation interval. ： in, Let be the number of samples that failed within the k-th standard deviation interval of the temperature rise rate. Let be the total number of samples within the k-th standard deviation interval of the temperature rise rate; sort each standard deviation interval of the temperature rise rate in ascending order of value, and calculate the cumulative probability of failure: set the standard deviation corresponding to when the cumulative probability of failure first reaches or exceeds the first proportion threshold as the first warning threshold. The threshold for the second proportional threshold will be set as the level 2 warning threshold when it is first reached or exceeded. The threshold for a Level 3 warning will be set when the threshold is first reached or exceeded. 。 5. The method for early warning and life prediction of electric drive bridge temperature rise according to claim 1, characterized in that, Step S4 is as follows: S41. Construct a three-dimensional digital model consistent with the physical electric drive bridge's geometric parameters, materials, and connection relationships, and add temperature simulation logic and motion simulation logic to generate a digital twin model of the electric drive bridge; S42. Receive the collected temperature, speed, and phase data of the electric drive bridge in real time through the data interface, and drive the digital twin model of the electric drive bridge to run synchronously; S43. Perform temperature field simulation in the digital twin model of the electric drive bridge, and use the finite element analysis method to calculate the internal temperature distribution of the electric drive bridge based on real-time data; S44. Perform kinematic simulation in the digital twin model of the electric drive bridge, and simulate the motion state of each component of the electric drive bridge based on phase and speed data; S45. Through the working condition simulation function, set different combinations of load and speed parameters to simulate the operating state of the electric drive bridge under different preset working conditions, including full-load climbing and high-speed driving, and generate simulation data. The generation frequency of the simulation data is consistent with the actual sampling frequency.

6. The method for early warning and life prediction of electric drive bridge temperature rise according to claim 1, characterized in that, Step S5 is as follows: S51. Optimize the threshold set: Based on the acquired historical operating data and simulated operating condition data, using the standard deviation sequence of temperature rise rate under different fault types as input and the verified optimal threshold corresponding to each fault type as the expected output, train the first neural network model to generate a threshold set that can be dynamically adjusted according to real-time temperature rise characteristics; S52. Optimize the optimal zero-adjustment threshold calibration process: Based on the phase characteristics of each node obtained by analysis and the historical data of the health status of the electric drive bridge, calculate the standard deviation of the phase difference sequence under the health status offline; based on the standard deviation of the phase difference sequence under the health status, calculate and determine the optimal zero-adjustment threshold by setting a fixed multiple; S53. Construct and optimize the life prediction model: Based on the acquired historical operating data and simulated operating condition data, using the operating time series, temperature rise data, phase data and maintenance records of the electric drive bridge as input and the actual life as output, train the second long short-term memory network model to construct and optimize the life prediction model and control the prediction error within the set error range.

7. The method for early warning and life prediction of electric drive bridge temperature rise according to claim 1, characterized in that, The specific steps of step S6 are as follows: S61. Based on the temperature rise fluctuation characteristics, phase characteristics and the operating status simulated by the digital twin model of the electric drive bridge, and applying the optimized threshold set and the optimal zeroing threshold, perform multi-level early warning judgment; S62. When the first-level warning condition is met, the first-level warning is triggered, and an intermittent prompt sound at a first frequency is emitted through the audible and visual alarm device; S63. When the second-level warning condition is met, the second-level warning is triggered, and an intermittent prompt sound at a second frequency is emitted through the audible and visual alarm device, supplemented by first-level voice prompt information; the second frequency is higher than the first frequency; S64. When the third-level warning condition is met, the third-level warning is triggered, and a continuous prompt sound is emitted through the audible and visual alarm device, supplemented by second-level voice prompt information; wherein, the first-level warning condition is any of the following: the temperature rise fluctuation characteristic exceeds the first-level warning threshold; the phase characteristic deviates from the optimal zeroing threshold to reach a first preset range; digital twin model The simulated operating status parameters exceed the first normal range threshold; the second-level warning condition is any of the following: the temperature rise fluctuation characteristics exceed the second-level warning threshold; the phase characteristics deviate from the optimal zeroing threshold to reach the second preset range; the operating status parameters simulated by the digital twin model exceed the second normal range threshold; the third-level warning condition is any of the following: the temperature rise fluctuation characteristics exceed the third-level warning threshold; the phase characteristics deviate from the optimal zeroing threshold to reach the third preset range; the operating status parameters simulated by the digital twin model exceed the third normal range threshold; and the degree of deviation of the first preset range, the second preset range, and the third preset range increases sequentially, and the strictness of the first normal range threshold, the second normal range threshold, and the third normal range threshold increases sequentially.

8. The method for early warning and life prediction of electric drive bridge temperature rise according to claim 1, characterized in that, Step S7 is as follows: S71. Obtain the simulation data of temperature rise fluctuation characteristics, phase characteristics, and digital twin model of electric drive axle, and input them into the optimized life prediction model; S72. The optimized life prediction model uses deep learning to fit the degradation curve, predicts the remaining life of electric drive axle, and displays the predicted remaining life according to a preset time unit; S73. When the predicted remaining life is less than the preset time threshold, it is highlighted on the vehicle display screen, and reference information on the impact of usage intensity on life is given.

9. A system for early warning and life prediction of electric drive bridge temperature rise, using the method for early warning and life prediction of electric drive bridge according to any one of claims 1-8, characterized in that, include: The data acquisition module is used to set several monitoring positions as nodes on the electric drive bridge, periodically collect temperature data of each node of the electric drive bridge, obtain temperature time series data of each node, and at the same time collect phase data and operating parameter data of the electric drive bridge. The weak feature extraction module is used to calculate the temperature rise rate and standard deviation of each node based on the temperature time series data of each node, compare the standard deviation of the temperature rise rate with the threshold set generated based on historical fault data, and extract the temperature rise fluctuation features based on the comparison results; the multi-node phase analysis module is used to analyze the correspondence between temperature fluctuation and phase change of each node based on the temperature time series data of each node, dynamically calibrate the optimal zero-adjustment threshold, and output the phase features of temperature fluctuation between each node based on the optimal zero-adjustment threshold. The digital twin module is used to build and drive a digital twin model of the electric drive bridge based on the collected temperature data, phase data and operating parameter data of each node of the electric drive bridge, to simulate the operating status of the electric drive bridge in real time and generate simulation data. The deep learning module is used to optimize the threshold set and the optimal zero-adjustment threshold calibration process using historical values ​​of temperature data, phase data, and operating parameter data of each node of the electric drive axle, as well as simulation data generated by the digital twin model of the electric drive axle. It also trains and optimizes the pre-built life prediction model. The early warning module is used to provide real-time vehicle-side early warning of electric drive axle temperature rise based on the temperature rise fluctuation characteristics, phase characteristics, and the operating status of the electric drive axle simulated by the digital twin model of the electric drive axle, applying the optimized threshold set and the optimal zero-adjustment threshold. The life prediction module is used to predict the remaining lifespan of the electric drive bridge based on the temperature rise fluctuation characteristics, phase characteristics, the operating status of the electric drive bridge simulated by the digital twin model of the electric drive bridge, and the optimized life prediction model.

Images