Fault diagnosis method for electromechanical composite transmission system of special vehicle

Abstract

The invention belongs to the technical field of intelligent operation and maintenance of armored vehicles, and particularly relates to a fault diagnosis method for an electromechanical composite transmission system of a special vehicle. The method is suitable for fault positioning and failure mode identification of a system level, a component level (a driving motor, a driving motor controller, a coupling mechanism, a planetary speed change mechanism and the like) and a part level (a bearing, a gear, a permanent magnet, a winding and the like) under complex working conditions. The problems of low fault detection rate, fuzzy positioning and the like of the special vehicle transmission system are solved by constructing a system-component-part three-level diagnosis framework and combining a multi-source data fusion and digital twinning technology, full-level fault quick positioning of the armored vehicle transmission system is realized, and the reliability and battlefield viability under complex working conditions are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance technology for armored vehicles, specifically involving a fault diagnosis method for electromechanical composite transmission systems of special vehicles. It is a multi-level fault diagnosis method for electromechanical composite transmission systems based on multi-source information fusion and digital twins, which is applicable to fault location and failure mode identification at the system level, component level and part level under complex working conditions. Background Technology

[0002] The electromechanical composite transmission system of special vehicles consists of a drive motor, planetary gear set, track transmission mechanism and electronic control unit, etc., and must meet the requirements of high load, strong impact and extreme environment adaptability.

[0003] 1. System level: Relies on monitoring of a single parameter (such as vibration), lacking global diagnosis of abnormal power distribution and gear shifting logic failure; 2. Component level: Gearbox fault feature extraction is susceptible to spectral aliasing interference, and the identification rate of weak faults in motor bearings is low; 3. Insufficient data fusion: Information from multiple sensors (vibration, current, temperature) is not effectively correlated, and the intelligent algorithm still mainly relies on threshold judgment. Summary of the Invention

[0004] (a) Technical problems to be solved The technical problem this invention aims to solve is that existing fault diagnosis methods suffer from insufficient monitoring sensors, making it difficult to comprehensively and systematically diagnose multi-level faults in electromechanical composite transmission systems. These methods cannot quickly and accurately pinpoint the specific level and location of faults, resulting in low efficiency and poor accuracy, and failing to meet the high reliability and stability requirements of modern warfare for electromechanical composite transmission systems. Therefore, there is an urgent need for a method capable of multi-level fault diagnosis of electromechanical composite transmission systems, thereby improving the efficiency and accuracy of fault diagnosis.

[0005] The purpose of this invention is to provide a multi-level fault diagnosis method for electromechanical composite transmission systems. By performing hierarchical analysis of the system and combining various diagnostic techniques, this method enables accurate fault diagnosis of electromechanical composite transmission systems from the system level to the component level and part level, thereby improving the efficiency and accuracy of fault diagnosis.

[0006] (II) Technical Solution To address the aforementioned technical problems, this invention provides a fault diagnosis method for electromechanical composite transmission systems of special vehicles. This method addresses the issues of low fault detection rate and ambiguous fault location caused by insufficient monitoring sensors in the transmission systems of special vehicles by constructing a three-level diagnostic framework of "system-component-part".

[0007] The fault diagnosis method, at the transmission system level, collects key monitoring parameters of the electromechanical composite transmission system in real time, including output speed and drive motor input voltage / current, and sets dynamic thresholds based on operating conditions. If any key monitoring parameter exceeds the threshold or is inconsistent with the command issued by the controller, it first determines whether the sensor is working properly. If it is normal, it determines that an abnormal phenomenon has occurred in the electromechanical composite transmission system. Based on the fault tree, a fault propagation path is constructed, and DS evidence theory is used for decision-level fusion, before proceeding to the component level for anomaly determination. At the component level, different key parameters are collected for different components. Combined with the component's structure and working principle, as well as historical operating data and maintenance records, a digital twin model is called to obtain the mapping relationship between the component's input and output, and to determine whether the component has any abnormalities. If it is determined that the component has abnormalities, then the fault diagnosis is performed at the part level. In component-level fault diagnosis, for directly monitorable components, fault diagnosis is performed by checking whether the monitoring parameters exceed thresholds. For components that cannot be directly monitored, if external sensors are installed for non-destructive testing, fault features can be extracted from the monitoring parameters, and monitoring indicators can be constructed through feature layer fusion. Fault diagnosis is then performed by checking whether the monitoring indicators exceed thresholds. For components that cannot be monitored, component faults are injected through component digital twin models to generate virtual monitoring parameters. External monitoring parameters at the time of component failure are obtained, and cross-validation is performed by combining multi-source data fusion. Based on machine learning, a mapping relationship between external monitoring parameters and internal component faults is constructed, and component fault diagnosis is achieved through this mapping relationship.

[0008] The key monitoring parameters of the electromechanical composite transmission system include: the output speed of the electromechanical composite transmission system, the input voltage, current, output torque and speed of the drive motor, as well as the internal winding temperature and the output bearing temperature; the input current, voltage and output speed of the oil pump motor of the hydraulic control system, the main pressure valve oil pressure, the first gear oil pressure, the second gear oil pressure, the pre-filter and post-filter pressure, the lubricating oil pressure, the return oil temperature and the oil tank level; and index parameters including transmission efficiency, the output speed difference of the transmission system under straight driving conditions and the vibration amplitude.

[0009] The method for determining whether a sensor is functioning properly is as follows: by comparing the feedback parameters of multiple sensors with the control input and by performing multi-sensor correlation analysis, it is determined whether the sensor has faults such as open circuit / short circuit, signal drift, nonlinear error, response delay, or decreased sensitivity. If the function is normal, it is further determined whether the key monitoring parameters collected exceed the upper and lower thresholds set under this working condition. If they exceed the thresholds, it is determined that an abnormal phenomenon has occurred in the electromechanical composite transmission system.

[0010] The abnormal phenomena of the electromechanical composite transmission system include: power interruption, abnormal power coupling, gear shifting failure, overall efficiency reduction, and discontinuous output speed, which are used to characterize the straight driving, steering and shifting functions of the transmission system.

[0011] The components include: an integrated management and control system, a drive motor, a drive motor controller, a coupling mechanism, a planetary gear transmission mechanism, and a hydraulic control mechanism. The component-level abnormal phenomena include: communication abnormalities in the integrated management and control system; no power output from the drive motor, insufficient output power, or unstable speed; communication abnormalities, excessively high temperature, three-phase power imbalance, phase loss, excessively high or low output current, or fluctuations in output current; uncoordinated speed or uneven load distribution in the coupling mechanism; fluctuating transmission ratio, localized overheating, or increased vibration in the planetary gear transmission mechanism; and no pressure, insufficient pressure, or pressure fluctuations in the hydraulic control mechanism.

[0012] The component fault diagnosis includes: abnormal CAN communication signals in the integrated management and control system; stator insulation failure, stator winding short circuit / open circuit, permanent magnet demagnetization, output bearing wear, and resolver sensor drift / open circuit in the drive motor; IGBT overheating, capacitor open circuit / leakage, and inductor open circuit / short circuit in the drive motor controller; broken sun gear teeth, broken planetary gear teeth, bearing wear, and seal leakage in the coupling mechanism; broken sun gear teeth, broken planetary gear teeth, bearing wear, shaft breakage, and seal leakage in the planetary transmission mechanism; stator insulation failure, stator winding short circuit / open circuit, permanent magnet demagnetization, output bearing wear, and resolver sensor drift / open circuit in the hydraulic control mechanism; as well as main pressure valve jamming, 1st gear switch valve jamming, 2nd gear switch valve jamming, pressure regulating valve jamming, and filter blockage.

[0013] The multi-source data fusion includes: Data layer fusion: Normalized vibration, current, and temperature signals are sampled synchronously using a sliding window; Feature layer fusion: Extract time, frequency domain statistical features and dynamic features such as entropy values ​​of monitoring parameters, extract principal components through methods such as principal component analysis, and input them into unsupervised machine learning such as deep belief networks to construct monitoring indicators; Decision-making level integration: Based on D-S evidence theory, it integrates multi-source data analysis results, expert knowledge and uncertainty reasoning to improve the accuracy of fault diagnosis.

[0014] The digital twin model includes: Model building: Construct a dynamic model of the electromechanical composite transmission system, import actual operating data to correct stiffness and damping parameters, and simulate the fault propagation path of the tracked vehicle under conditions such as straight driving, steering, and braking. Fault Injection and Simulation: Simulate typical faults in the mechanical, electrical, and hydraulic subsystems of the electromechanical composite transmission system, generate a fault feature library, and use it to train diagnostic algorithms and dynamically optimize alarm thresholds.

[0015] (III) Beneficial Effects Compared with the prior art, the beneficial effects of the present invention are as follows: (1) By adopting a multi-level fault diagnosis method, the analysis is gradually deepened from the system level to the component level and part level. This method can comprehensively and systematically diagnose faults in electromechanical composite transmission systems, greatly improving the accuracy and efficiency of fault diagnosis and enabling rapid location of the specific level and location of the fault.

[0016] (2) Combining multiple diagnostic technologies, including sensor data acquisition, signal processing, machine learning algorithms, digital twin technology, and fault feature extraction, fully utilizes the advantages of different technologies, effectively detects various types of faults, and improves the reliability and comprehensiveness of fault diagnosis.

[0017] (3) By establishing a system-level fault diagnosis model and a component-level fault diagnosis rule base, and using historical data for training and learning, it can adapt to the fault diagnosis needs under different working conditions and operating environments, and has strong versatility and adaptability. Attached Figure Description

[0018] Figure 1 This invention relates to a fault diagnosis method for electromechanical composite transmission systems of special vehicles. Figure 2 This is a diagram showing the structural composition of the electromechanical composite transmission system for special vehicles in this invention. Figure 3 This invention relates to a component fault diagnosis method based on digital twin fault injection and machine learning. Detailed Implementation

[0019] To make the objectives, contents, and advantages of the present invention clearer, the specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples.

[0020] To address the aforementioned technical problems, this invention provides a fault diagnosis method for electromechanical composite transmission systems of special vehicles. This method addresses the issues of low fault detection rate and ambiguous fault location caused by insufficient monitoring sensors in the transmission systems of special vehicles by constructing a three-level diagnostic framework of "system-component-part".

[0021] The fault diagnosis method, at the transmission system level, collects key monitoring parameters of the electromechanical composite transmission system in real time, including output speed and drive motor input voltage / current, and sets dynamic thresholds based on operating conditions. If any key monitoring parameter exceeds the threshold or is inconsistent with the command issued by the controller, it first determines whether the sensor is working properly. If it is normal, it determines that an abnormal phenomenon has occurred in the electromechanical composite transmission system. Based on the fault tree, a fault propagation path is constructed, and DS evidence theory is used for decision-level fusion, before proceeding to the component level for anomaly determination. At the component level, different key parameters are collected for different components. Combined with the component's structure and working principle, as well as historical operating data and maintenance records, a digital twin model is called to obtain the mapping relationship between the component's input and output, and to determine whether the component has any abnormalities. If it is determined that the component has abnormalities, then the fault diagnosis is performed at the part level. In component-level fault diagnosis, for directly monitorable components, fault diagnosis is performed by checking whether the monitoring parameters exceed thresholds. For components that cannot be directly monitored, if external sensors are installed for non-destructive testing, fault features can be extracted from the monitoring parameters, and monitoring indicators can be constructed through feature layer fusion. Fault diagnosis is then performed by checking whether the monitoring indicators exceed thresholds. For components that cannot be monitored, component faults are injected through component digital twin models to generate virtual monitoring parameters. External monitoring parameters at the time of component failure are obtained, and cross-validation is performed by combining multi-source data fusion. Based on machine learning, a mapping relationship between external monitoring parameters and internal component faults is constructed, and component fault diagnosis is achieved through this mapping relationship.

[0022] The key monitoring parameters of the electromechanical composite transmission system include: the output speed of the electromechanical composite transmission system, the input voltage, current, output torque and speed of the drive motor, as well as the internal winding temperature and the output bearing temperature; the input current, voltage and output speed of the oil pump motor of the hydraulic control system, the main pressure valve oil pressure, the first gear oil pressure, the second gear oil pressure, the pre-filter and post-filter pressure, the lubricating oil pressure, the return oil temperature and the oil tank level; and index parameters including transmission efficiency, the output speed difference of the transmission system under straight driving conditions and the vibration amplitude.

[0023] The method for determining whether a sensor is functioning properly is as follows: by comparing the feedback parameters of multiple sensors with the control input and by performing multi-sensor correlation analysis, it is determined whether the sensor has faults such as open circuit / short circuit, signal drift, nonlinear error, response delay, or decreased sensitivity. If the function is normal, it is further determined whether the key monitoring parameters collected exceed the upper and lower thresholds set under this working condition. If they exceed the thresholds, it is determined that an abnormal phenomenon has occurred in the electromechanical composite transmission system.

[0024] The abnormal phenomena of the electromechanical composite transmission system include: power interruption, abnormal power coupling, gear shifting failure, overall efficiency reduction, and discontinuous output speed, which are used to characterize the straight driving, steering and shifting functions of the transmission system.

[0025] The components include: an integrated management and control system, a drive motor, a drive motor controller, a coupling mechanism, a planetary gear transmission mechanism, and a hydraulic control mechanism. The component-level abnormal phenomena include: communication abnormalities in the integrated management and control system; no power output from the drive motor, insufficient output power, or unstable speed; communication abnormalities, excessively high temperature, three-phase power imbalance, phase loss, excessively high or low output current, or fluctuations in output current; uncoordinated speed or uneven load distribution in the coupling mechanism; fluctuating transmission ratio, localized overheating, or increased vibration in the planetary gear transmission mechanism; and no pressure, insufficient pressure, or pressure fluctuations in the hydraulic control mechanism.

[0026] The component fault diagnosis includes: abnormal CAN communication signals in the integrated management and control system; stator insulation failure, stator winding short circuit / open circuit, permanent magnet demagnetization, output bearing wear, and resolver sensor drift / open circuit in the drive motor; IGBT overheating, capacitor open circuit / leakage, and inductor open circuit / short circuit in the drive motor controller; broken sun gear teeth, broken planetary gear teeth, bearing wear, and seal leakage in the coupling mechanism; broken sun gear teeth, broken planetary gear teeth, bearing wear, shaft breakage, and seal leakage in the planetary transmission mechanism; stator insulation failure, stator winding short circuit / open circuit, permanent magnet demagnetization, output bearing wear, and resolver sensor drift / open circuit in the hydraulic control mechanism; as well as main pressure valve jamming, 1st gear switch valve jamming, 2nd gear switch valve jamming, pressure regulating valve jamming, and filter blockage.

[0027] The multi-source data fusion includes: Data layer fusion: Normalized vibration, current, and temperature signals are sampled synchronously using a sliding window; Feature layer fusion: Extract time, frequency domain statistical features and dynamic features such as entropy values ​​of monitoring parameters, extract principal components through methods such as principal component analysis, and input them into unsupervised machine learning such as deep belief networks to construct monitoring indicators; Decision-making level integration: Based on D-S evidence theory, it integrates multi-source data analysis results, expert knowledge and uncertainty reasoning to improve the accuracy of fault diagnosis.

[0028] The digital twin model includes: Model building: Construct a dynamic model of the electromechanical composite transmission system, import actual operating data to correct stiffness and damping parameters, and simulate the fault propagation path of the tracked vehicle under conditions such as straight driving, steering, and braking. Fault Injection and Simulation: Simulate typical faults in the mechanical, electrical, and hydraulic subsystems of the electromechanical composite transmission system, generate a fault feature library, and use it to train diagnostic algorithms and dynamically optimize alarm thresholds.

[0029] Example 1 As shown in Figure 1, this embodiment takes a certain type of electromechanical composite transmission system as an example. The components of this system include an integrated management and control system, a drive motor, a drive motor controller, a coupling mechanism, a planetary transmission mechanism, and a hydraulic control mechanism. The structural relationship of each component is shown in Figure 2.

[0030] 1. Acquisition of key monitoring parameters for electromechanical composite transmission systems Key monitoring parameters for the electromechanical composite transmission system include: the output speed (r / min) of the electromechanical composite transmission system, the input voltage (V), current (A), output torque (N·m), and speed of the drive motor, as well as the internal winding temperature (°C) and output bearing temperature (°C); the input current (A), voltage (V), and output speed (r / min) of the oil pump motor of the hydraulic control system, the main pressure valve oil pressure (MPa), the first-gear oil pressure (MPa), the second-gear oil pressure (MPa), the filter pressure before and after filtration (MPa), the lubricating oil pressure (MPa), the return oil temperature (°C), and the oil tank level (mm); and index parameters including transmission efficiency, the output speed difference of the transmission system under straight-line driving conditions (r / min), and vibration amplitude (m / s). 2 To reduce noise interference in the monitoring parameters, the collected parameters can be filtered and normalized.

[0031] 2. Determination of Abnormal Phenomena in Electromechanical Composite Transmission Systems Power interruption determination: When the vehicle is driving straight or turning, there is no speed or torque output. The reasons may be: no power output from the drive motor; abnormal communication or excessive temperature of the drive motor controller; broken drive shaft or failure of friction plates in the planetary gear transmission mechanism; insufficient pressure of the hydraulic control mechanism resulting in insufficient clamping force of the friction plates in the planetary gear transmission mechanism.

[0032] Power coupling anomaly diagnosis: When the vehicle is driving straight, the output speed of the left and right ends is inconsistent, or the torque output difference between the left and right ends is large. The cause may be damage to the gears or bearings inside the coupling mechanism. Gear shifting failure diagnosis: Inability to engage gear, inability to shift from neutral to 1st gear, inability to shift from neutral to 2nd gear, inability to shift from neutral to reverse gear, inability to shift from 1st gear to 2nd gear, etc., may be caused by wear or leakage of the planetary transmission mechanism's transmission friction plates; insufficient pressure of the hydraulic control mechanism may result in insufficient clamping force of the planetary transmission mechanism's friction plates; Overall efficiency decline determination: Under the same working conditions, the transmission efficiency of the transmission system shows a significant downward trend over time. The reasons may be that the aging of the drive motor components leads to increased heat generation; wear of the planetary gear transmission mechanism, etc. Discontinuous output speed of the transmission system: The speed is intermittent or there is large fluctuation in the transmission ratio. The reasons may be: large fluctuation in the speed of the drive motor; abnormal communication of the drive motor controller; insufficient pressure of the hydraulic control mechanism or insufficient clamping force of the friction plates in the planetary gear transmission mechanism due to excessive wear of the friction plates.

[0033] 3. Sensor Fault Diagnosis By comparing multi-sensor parameter feedback with control input and performing multi-sensor correlation analysis, it is determined whether the sensors have faults such as open circuit / short circuit, signal drift, nonlinear error, response delay, or decreased sensitivity. If the function is normal, it is further determined whether the collected key monitoring parameters exceed the upper and lower thresholds set for this operating condition. If they do, it is determined that an abnormal phenomenon has occurred in the electromechanical composite transmission system. If an abnormal phenomenon exists at the system level, characteristic parameters are collected for components such as the drive motor, drive motor controller, coupling mechanism, planetary transmission mechanism, and hydraulic control mechanism. Through fault transmission path and multi-source information fusion decision-making, the abnormal component is located.

[0034] 4. Component Anomaly Determination Drive motor anomaly detection: The drive motor's electromechanical-thermal-magnetic multiphysics coupled digital twin model is invoked. If the deviation between the real-time calculated value and the actual value of the output speed / torque exceeds a threshold, an anomaly is detected. Anomalies include no power output, insufficient output power, or unstable speed. The drive motor controller's electromechanical-thermal multiphysics coupled digital twin model is invoked. If the deviation between the real-time calculated value and the actual value exceeds a threshold, an anomaly is detected. Anomalies include communication anomalies, excessively high temperature, three-phase imbalance, phase loss, excessively high or low output current, or fluctuations. The coupling mechanism's digital twin model is invoked. If... If the deviation between the real-time calculated output speed and the actual value exceeds a threshold, an anomaly is determined. Anomalies include: uncoordinated speeds or uneven load distribution. If the deviation between the real-time calculated output speed / torque and the actual value exceeds a threshold when the digital twin model of the planetary transmission mechanism is invoked, an anomaly is determined. Anomalies include transmission ratio fluctuations, uneven torque distribution, localized overheating, or increased vibration. If the output lubricating oil pressure and 1st / 2nd gear pressure exceed a threshold when the digital twin model of the hydraulic control mechanism is invoked, an anomaly is determined. Anomalies include no pressure, insufficient pressure, and pressure fluctuations.

[0035] 5. Component Fault Diagnosis For drive motors, the stator temperature rise gradient is directly monitored by internal winding temperature sensors. Combined with current harmonic characteristics extracted from current sensors at the drive motor input, insulation faults are determined. An adaptive mode decomposition method is used to preprocess and perform spectral analysis on the drive motor input current and output speed, calculating the theoretical fault frequency of the output bearing. If the speed spectrum exceeds a specified threshold at the theoretical bearing fault frequency, the bearing is considered faulty. Fault injection into components such as windings and permanent magnets within a digital twin model of the drive motor's electromechanical-thermal-magnetic multi-physics coupling is used to obtain fault type characterization parameters, such as current changes during winding short circuits and flux linkage changes during permanent magnet demagnetization. Machine learning methods, such as support vector machines or convolutional neural networks, are used to construct a mapping relationship between external monitoring parameters and internal component fault type characterization parameters, determining whether windings or permanent magnets have faults during drive motor operation.

[0036] For the drive motor controller, the system directly monitors the drive motor controller's heartbeat to determine if communication is abnormal. It also directly monitors the IGBT temperature rise gradient based on the drive motor controller's temperature sensor, and combines this with the current harmonics calculated from the drive motor controller's output current sensor to determine if the IGBT is faulty. Furthermore, it injects faults into the drive motor controller's mechanical-electrical-thermal multiphysics digital twin model of components such as capacitors, diodes, and connectors to obtain component fault type characterization parameters. Finally, it constructs a mapping relationship between external monitoring parameters and internal component fault type characterization parameters based on machine learning methods such as support vector machines or convolutional neural networks to determine whether internal components of the drive motor controller have failed during operation.

[0037] For the coupling mechanism, the input rotational speed and vibration signals are preprocessed and analyzed by the adaptive mode decomposition method, and the theoretical fault frequencies of the sun gear, planet gears and bearings inside the coupling mechanism are calculated. If the rotational speed spectrum exceeds the specified threshold at the theoretical fault frequency point, the fault of the sun gear, planet gears and bearings is judged.

[0038] For planetary transmission mechanisms, an adaptive mode decomposition method is used to perform envelope demodulation and spectrum analysis on the input speed, output speed, and vibration signals. The theoretical fault frequencies of the sun gear, planet gears, and bearings within the planetary transmission mechanism are calculated. If the speed spectrum exceeds a specified threshold at the theoretical fault frequency point, a fault in the sun gear, planet gears, or bearings is identified. Friction plate faults are injected based on a digital twin model of the planetary transmission mechanism to obtain friction plate fault type characterization parameters. A mapping relationship between external monitoring parameters and friction plate fault characterization parameters is constructed using machine learning methods such as support vector machines or convolutional neural networks to determine whether internal friction plates have failed. Loosening or breakage of connecting parts in the planetary transmission mechanism is identified by analyzing whether the system's natural frequencies have shifted.

[0039] For the hydraulic control mechanism, the filter element is checked for blockage by checking if the pressure difference before and after the filter exceeds a set threshold. Based on the digital twin model of the hydraulic control mechanism, valve body fault injection is performed to obtain valve body fault type characterization parameters. Then, based on machine learning such as support vector machine or convolutional neural network, a mapping relationship between the external monitoring parameters of the component and the valve body fault characterization parameters is constructed to determine whether the valve body has malfunctioned. Valve body faults are determined by the main pressure valve oil pressure, the first gear oil pressure, and the second gear oil pressure.

[0040] The flowchart for fault diagnosis of undetectable internal components based on digital twin fault injection and machine learning is attached. Figure 3 As shown.

[0041] Example 2 To address the aforementioned technical problems, this embodiment provides a fault diagnosis method for a special vehicle electromechanical composite transmission system, constructing a three-level diagnostic framework of "system-component-part"; the solution includes the following steps: At the transmission system level, key monitoring parameters such as the output speed of the electromechanical composite transmission system and the input voltage / current of the drive motor are collected in real time, and dynamic thresholds are set based on the operating conditions. If any key monitoring parameter exceeds the threshold or is inconsistent with the command issued by the controller, the sensor is first checked for normal operation. If it is normal, it is determined that an abnormal phenomenon has occurred in the electromechanical composite transmission system. The fault propagation path is constructed based on the fault tree, and the decision-level fusion is performed using DS evidence theory. The result is then transferred to the component level for anomaly determination. At the component level, different key parameters are collected for different components. Combined with the component's structure and working principle, as well as historical operating data and maintenance records, a digital twin model is called to obtain the mapping relationship between the component's input and output, and to determine whether the component has any abnormalities. If it is determined that the component has abnormalities, then the fault diagnosis is performed at the part level. In component-level fault diagnosis, for components that can be directly monitored, fault diagnosis is performed by using monitoring parameter thresholds. For components that cannot be directly monitored, if conditions permit, external sensors can be installed for non-destructive testing, fault features can be extracted from the monitoring parameters, and a component-level fault diagnosis rule base can be established. Machine learning is used to compare and match the extracted component parameter features with the fault features in the rule base to achieve fault diagnosis. For components that cannot be monitored, virtual monitoring parameters are generated first through a digital twin model, and then cross-validation is performed by combining multi-source data fusion to achieve component fault diagnosis.

[0042] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A fault diagnosis method for a special vehicle's electromechanical composite transmission system, characterized in that, The fault diagnosis method, at the transmission system level, collects key monitoring parameters of the electromechanical composite transmission system in real time, including output speed and drive motor input voltage / current, and sets dynamic thresholds based on operating conditions. If any key monitoring parameter exceeds the threshold or is inconsistent with the command issued by the controller, it first determines whether the sensor is working properly. If it is normal, it determines that an abnormal phenomenon has occurred in the electromechanical composite transmission system. Based on the fault tree, a fault propagation path is constructed, and DS evidence theory is used for decision-level fusion, before proceeding to the component level for anomaly determination. At the component level, different key parameters are collected for different components. Combined with the component's structure and working principle, as well as historical operating data and maintenance records, a digital twin model is called to obtain the mapping relationship between the component's input and output, and to determine whether the component has any abnormalities. If an abnormality is found in a component, then proceed to the component level for fault diagnosis. In component-level fault diagnosis, for directly monitorable components, fault diagnosis is performed by checking whether the monitoring parameters exceed thresholds. For components that cannot be directly monitored, if external sensors are installed for non-destructive testing, fault features can be extracted from the monitoring parameters, and monitoring indicators can be constructed through feature layer fusion. Fault diagnosis is then performed by checking whether the monitoring indicators exceed thresholds. For components that cannot be monitored, component faults are injected through component digital twin models to generate virtual monitoring parameters. External monitoring parameters at the time of component failure are obtained, and cross-validation is performed by combining multi-source data fusion. Based on machine learning, a mapping relationship between external monitoring parameters and internal component faults is constructed, and component fault diagnosis is achieved through this mapping relationship.

2. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The key monitoring parameters of the electromechanical composite transmission system include: the output speed of the electromechanical composite transmission system, the input voltage, current, output torque and speed of the drive motor, as well as the internal winding temperature and the output bearing temperature; the input current, voltage and output speed of the oil pump motor of the hydraulic control system, the main pressure valve oil pressure, the first gear oil pressure, the second gear oil pressure, the pre-filter and post-filter pressure, the lubricating oil pressure, the return oil temperature and the oil tank level; and index parameters including transmission efficiency, the output speed difference of the transmission system under straight driving conditions and the vibration amplitude.

3. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The method for determining whether a sensor is functioning properly is as follows: by comparing the feedback parameters of multiple sensors with the control input and performing multi-sensor correlation analysis, it is determined whether the sensor has faults such as open circuit / short circuit, signal drift, nonlinear error, response delay, or decreased sensitivity. If the function is normal, it is further determined whether the key monitoring parameters collected exceed the upper and lower thresholds set under this working condition. If they exceed the thresholds, it is determined that an abnormal phenomenon has occurred in the electromechanical composite transmission system.

4. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The abnormal phenomena of the electromechanical composite transmission system include: power interruption, abnormal power coupling, gear shifting failure, overall efficiency reduction, and discontinuous output speed, which are used to characterize the straight driving, steering and shifting functions of the transmission system.

5. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The components include: an integrated management and control system, a drive motor, a drive motor controller, a coupling mechanism, a planetary gear transmission mechanism, and a hydraulic control mechanism; the component-level abnormal phenomena include: communication abnormalities in the integrated management and control system, no power output from the drive motor, insufficient output power or unstable speed, communication abnormalities in the drive motor controller, excessively high temperature, three-phase power imbalance, phase loss, excessively high or low output current or fluctuations, uncoordinated speed or uneven load distribution in the coupling mechanism, fluctuations in the transmission ratio of the planetary gear transmission mechanism, local overheating or increased vibration, and no pressure, insufficient pressure, or pressure fluctuations in the hydraulic control mechanism.

6. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The component fault diagnosis includes: abnormal CAN communication signal in the integrated management and control system; stator insulation failure, stator winding short circuit / open circuit, permanent magnet demagnetization, output bearing wear, and resolver sensor drift / open circuit in the drive motor; IGBT overheating, capacitor open circuit / leakage, and inductor open circuit / short circuit in the drive motor controller; broken sun gear teeth, broken planetary gear teeth, bearing wear, and seal leakage in the coupling mechanism; broken sun gear teeth, broken planetary gear teeth, bearing wear, shaft breakage, and seal leakage in the planetary transmission mechanism; stator insulation failure, stator winding short circuit / open circuit, permanent magnet demagnetization, output bearing wear, and resolver sensor drift / open circuit in the hydraulic control mechanism; as well as main pressure valve jamming, 1st gear switch valve jamming, 2nd gear switch valve jamming, pressure regulating valve jamming, and filter blockage.

7. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The multi-source data fusion includes: Data layer fusion: Normalized vibration, current, and temperature signals are sampled synchronously using a sliding window; Feature layer fusion: Extract time, frequency domain statistical features and dynamic features such as entropy values ​​of monitoring parameters, extract principal components through methods such as principal component analysis, and input them into unsupervised machine learning such as deep belief networks to construct monitoring indicators; Decision-making level integration: Based on D-S evidence theory, it integrates multi-source data analysis results, expert knowledge and uncertainty reasoning to improve the accuracy of fault diagnosis.

8. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The digital twin model includes: Model building: Construct a dynamic model of the electromechanical composite transmission system, import actual operating data to correct stiffness and damping parameters, and simulate the fault propagation path of the tracked vehicle under conditions such as straight driving, steering, and braking. Fault Injection and Simulation: Simulate typical faults in the mechanical, electrical, and hydraulic subsystems of the electromechanical composite transmission system, generate a fault feature library, and use it to train diagnostic algorithms and dynamically optimize alarm thresholds.

9. The fault diagnosis method for the electromechanical composite transmission system of special vehicles as described in claim 1, characterized in that, The method addresses the problem of low fault detection rate and ambiguous fault location in the transmission system of special vehicles due to insufficient number of monitoring sensors by constructing a three-level diagnostic framework of "system-component-part".

10. The fault diagnosis method for a special vehicle electromechanical composite transmission system as described in claim 1, characterized in that, The method described belongs to the field of intelligent operation and maintenance technology for armored vehicles.

Images