Method and equipment for predicting fatigue life of compressor assembly and medium

Abstract

The invention provides a compressor assembly fatigue life prediction method and device and a medium, and relates to the technical field of data processing, in the method, at least one target sub-time period corresponding to each preset vehicle operation condition is determined from target time periods, and a target data set corresponding to the target sub-time period is obtained, the target data set comprises vibration data collected by vibration sensors arranged at multiple positions of the compressor assembly and operation data of target operation parameters of the compressor assembly; based on the target data set, acquiring data feature vectors corresponding to the preset vehicle operation conditions; fusing the data feature vectors corresponding to the preset vehicle operation conditions into a comprehensive feature vector, and inputting the comprehensive feature vector into a fatigue life prediction model to obtain a fatigue life value of the compressor assembly; the additional influence of external vibration and bumping generated by the vehicle under different preset vehicle operation conditions on the compressor assembly is fully considered, and the accuracy of fatigue life prediction is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, device and medium for predicting the fatigue life of compressor components. Background Technology

[0002] As a key moving component in a vehicle system, the fatigue life of the compressor assembly directly affects the reliability and maintenance cost of the entire vehicle. Therefore, it is necessary to predict the fatigue life of the compressor assembly in the vehicle to identify structural damage risks in advance, avoid sudden failures, and ensure the safe and reliable operation of the system. In the existing technology, the fatigue life prediction of the compressor assembly is usually based on the stress and vibration data collected by the compressor assembly when it is running alone in the laboratory or under ideal working conditions. For example, the compressor is started and stopped normally or continuously on a fixed test bench to measure its internal stress and vibration, and the fatigue life of the compressor assembly is estimated accordingly.

[0003] However, the above method also has the following technical problems: In practical applications, compressor assemblies are typically installed on mobile devices (such as vehicles). During actual operation, vehicles experience changes in operating conditions, such as starting, climbing hills, and driving on bumpy roads at high speeds, which leads to bumps and vibrations. These bumps and vibrations are transmitted to the compressor assembly, and when combined with the vibrations generated by the compressor assembly itself during operation, the internal critical components are subjected to more complex and severe repeated stresses. However, the above method does not take into account the additional impact of the bumps and vibrations generated during vehicle operation on the compressor assembly. Therefore, the accuracy of the fatigue life predicted based on the above method is low, and it is difficult to truly reflect the durability performance of the compressor assembly. Summary of the Invention

[0004] To address the aforementioned technical problems, the technical solution adopted by this invention is as follows: According to a first aspect of the present invention, a method for predicting the fatigue life of a compressor assembly is provided, the method comprising the following steps: S1. Determine at least one target sub-time period corresponding to each preset vehicle operating condition from the target time period; the end time of the target time period is the current time point, and the duration is the first preset duration t1; the target sub-time period corresponding to the preset vehicle operating condition is the time period during which the vehicle is in the preset operating condition and the operating status is stable within the target time period.

[0005] S2. For each target sub-time period corresponding to each preset vehicle operating condition, obtain the target dataset corresponding to the target sub-time period. The target dataset includes vibration data collected by various vibration sensors located at multiple positions on the compressor assembly and operating data of the target operating parameters of the compressor assembly within the target sub-time period. The compressor assembly is located inside the vehicle. The target operating parameters are operating parameters related to the preset vehicle operating conditions.

[0006] S3. Based on the target dataset corresponding to all target sub-time periods corresponding to each preset vehicle operating condition, obtain the data feature vector corresponding to each preset vehicle operating condition.

[0007] S4. Perform vector fusion on the data feature vectors corresponding to each preset vehicle operating condition to obtain a comprehensive feature vector.

[0008] S5. Input the comprehensive feature vector into the fatigue life prediction model to obtain the fatigue life value of the compressor component.

[0009] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, wherein a computer program is stored in the storage medium, and the computer program is loaded and executed by a processor to implement the aforementioned method.

[0010] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned method.

[0011] The present invention has at least the following beneficial effects: This invention provides a method, device, and medium for predicting the fatigue life of a compressor assembly. The method involves determining at least one target sub-time period corresponding to each preset vehicle operating condition from a target time period, and obtaining a target dataset corresponding to each target sub-time period. The target dataset includes vibration data collected by vibration sensors located at multiple positions on the compressor assembly within the target sub-time period, and operating data of the compressor assembly's target operating parameters. Based on the target datasets corresponding to all target sub-time periods for each preset vehicle operating condition, a data feature vector corresponding to each preset vehicle operating condition is obtained. Vector fusion is performed on the data feature vectors corresponding to each preset vehicle operating condition to obtain a comprehensive feature vector. The comprehensive feature vector is input into a fatigue life prediction model to obtain the fatigue life value of the compressor assembly. As can be seen, this invention acquires vibration data and operating data of the compressor component's target operating parameters within the target sub-time period corresponding to each preset vehicle operating condition experienced by the vehicle in actual operation. It extracts vibration characteristics and operating characteristics under each preset vehicle operating condition and integrates them to form a comprehensive feature that can fully reflect the real usage environment. Then, fatigue life prediction is performed based on this comprehensive feature, which fully considers the additional impact of external vibration and bumps generated by the vehicle under different preset vehicle operating conditions on the compressor component, significantly improving the accuracy of fatigue life prediction and truly reflecting the durability performance of the compressor component. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating a method for predicting the fatigue life of a compressor component, provided in an embodiment of the present invention. Detailed Implementation

[0014] 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.

[0015] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar tasks and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0016] Embodiments of the present invention provide a method for predicting the fatigue life of a compressor component, the method comprising the following steps, such as... Figure 1 As shown: S1. Determine at least one target sub-time period corresponding to each preset vehicle operating condition from the target time period; the end time of the target time period is the current time point, and the duration is the first preset duration t1; the target sub-time period corresponding to the preset vehicle operating condition is the time period during which the vehicle is in the preset operating condition and the operating status is stable within the target time period; wherein, the first preset duration is the duration preset by those skilled in the art according to actual needs, such as 5 days, 1 week, which will not be elaborated here.

[0017] Specifically, the preset vehicle operating conditions are typical vehicle operating states that have been determined in advance, including cold start, hot start, driving on flat roads, cruise control, climbing hills, driving on high-speed bumpy roads, and driving on low-speed bumpy roads.

[0018] In one specific embodiment, steps S1-S5 are executed at preset fixed intervals to complete a fatigue life prediction of the compressor component; wherein, the preset fixed interval is a duration of not less than t1 that is pre-set by those skilled in the art according to actual needs, such as 2 weeks or 1 month, which will not be elaborated here.

[0019] Specifically, step S1 includes the following sub-steps S11-S14: S11. Obtain several specific time periods corresponding to each preset vehicle operating condition. The specific time periods corresponding to the preset vehicle operating conditions are the time periods during which the vehicle is in the preset vehicle operating condition within the target time period.

[0020] Specifically, during vehicle operation, the current operating condition is determined in real time based on the vehicle's operating parameters. When a change in the operating condition is detected, the moment of the change is used as the end time of the previous operating condition and the start time of the next operating condition. At the end of any operating condition, the duration of that operating condition is calculated. If the duration is not less than a second preset duration, the time period between the start and end times of that operating condition is marked as the duration segment corresponding to that operating condition and stored in the local database; otherwise, it is not stored. Here, the operating condition can be understood as the vehicle's operating state. The second preset duration is a duration preset by those skilled in the art according to actual needs, such as 1 minute or 3 minutes, which will not be elaborated here.

[0021] Furthermore, those skilled in the art will understand that any existing method for determining the current operating condition of a vehicle in real time based on its operating parameters during operation is within the scope of protection of this invention. For example, a logical judgment method based on preset rules and thresholds may include, but is not limited to: determining a cold start when the engine coolant temperature is below 40°C and a rising edge of the ignition signal is detected; determining a hot start when the engine coolant temperature is above 70°C and an ignition restart signal is detected; determining flat driving when the vehicle speed is stable within the range of 55–65 km / h and both longitudinal acceleration and road gradient are close to zero; and determining flat driving when... When the cruise control system is detected to be active (e.g., CAN signal Cruise_Active=1) and the actual vehicle speed deviates from the set speed by less than the preset tolerance (e.g., ±2km / h), it is determined to be cruise control. When the estimated gradient is not less than 3% (approximately 1.7°) and the duration is not less than 2 seconds, it is determined to be hill climbing. When the vehicle body vibration intensity (e.g., suspension displacement or vertical acceleration RMS) exceeds the preset threshold and the vehicle speed is greater than 40km / h, it is determined to be driving on a high-speed bumpy road. When the vehicle body vibration intensity exceeds the preset threshold and the vehicle speed is less than 20km / h, it is determined to be driving on a low-speed bumpy road. These details will not be elaborated further.

[0022] Specifically, based on the duration segments corresponding to each operating condition stored in the local database corresponding to the vehicle, several specific time periods corresponding to each preset vehicle operating condition are obtained.

[0023] S12. When the preset vehicle operating condition is the first type of operating condition, for each specific time period corresponding to the preset vehicle operating condition, a time period with a length of the second preset duration t2 is extracted from the starting time point of the specific time period as the target sub-time period corresponding to the preset vehicle operating condition; when the preset vehicle operating condition is the second type of operating condition, for each specific time period corresponding to the preset vehicle operating condition, a sliding window with a window length of t2 and a sliding step size of the third preset duration t3 is used to divide the specific time period to obtain several specific sub-time periods corresponding to the specific time period; t3 < t2 < t1; the third preset duration is a duration less than the second preset duration set by those skilled in the art according to actual needs, such as 10 seconds or 30 seconds, which will not be elaborated here.

[0024] Specifically, the first type of operating condition is the start-stop transient operating condition; for example: cold start and hot start.

[0025] Specifically, the second category of operating conditions is dynamic driving conditions; for example: driving on flat roads, cruise control, climbing hills, driving on high-speed bumpy roads, and driving on low-speed bumpy roads.

[0026] Specifically, the duration of a particular time period shall not be less than t2.

[0027] Specifically, the number of target sub-time periods corresponding to the preset vehicle operating conditions is the same as the number of specific time periods corresponding to the preset vehicle operating conditions, and the target sub-time periods corresponding to the preset vehicle operating conditions correspond one-to-one with the specific time periods corresponding to the preset vehicle operating conditions.

[0028] Specifically, in several specific sub-time periods corresponding to a specific time period, the interval between the start times of any two adjacent specific sub-time periods is consistent with t3.

[0029] S13. Based on the vibration data collected by each vibration sensor, obtain the vibration signal energy assessment index corresponding to each specific sub-time period.

[0030] Specifically, vibration sensors are installed at multiple locations on the compressor assembly; the compressor assembly is located inside the vehicle; it can be understood that a vibration sensor is installed at each of the multiple locations on the compressor assembly; the locations of the vibration sensors in the compressor assembly are determined by those skilled in the art according to actual needs, such as: the rigid surface of the compressor housing near the motor / drive end, near the exhaust port, the compressor mounting base, which will not be elaborated here.

[0031] In one specific embodiment, the vibration sensor is an acceleration sensor, and the vibration data collected by the vibration sensor is acceleration.

[0032] Specifically, step S13 includes the following sub-steps S131-S133: S131. For each specific sub-time period, obtain the vibration dataset corresponding to the specific sub-time period, including the vibration data sequence corresponding to each vibration sensor. The vibration data sequence is composed of the vibration data collected by the corresponding vibration sensor within the specific sub-time period.

[0033] Specifically, all vibration data in the vibration data sequence are arranged in chronological order according to the time of acquisition.

[0034] S132. For each vibration data sequence, calculate the RMS value corresponding to the vibration data sequence based on all vibration data in the vibration data sequence.

[0035] Specifically, the RMS value corresponding to vibration data sequence A. A The following conditions must be met: Where Ai is the i-th vibration data in A, 1≤i≤n, and n is the number of vibration data in A.

[0036] S133. The average value of the RMS values ​​of all vibration data sequences in the vibration dataset corresponding to a specific sub-time period is used as the vibration signal energy assessment index for that specific sub-time period.

[0037] Through the above steps, a vibration dataset corresponding to a specific sub-time period is obtained, and the RMS value corresponding to each vibration data sequence in the vibration dataset is calculated to quantify the overall vibration intensity at the location of the vibration sensor corresponding to the vibration data sequence under the preset vehicle operating conditions corresponding to the specific sub-time period. The average value of the RMS values ​​corresponding to all vibration data sequences in the vibration dataset corresponding to the specific sub-time period is used as the vibration signal energy assessment index for that specific sub-time period. This index is used to characterize the overall vibration level of the compressor assembly under the preset vehicle operating conditions corresponding to the specific sub-time period. This approach retains the vibration response information of different installation locations while avoiding redundancy and noise interference caused by directly using the original time series data. At the same time, the obtained vibration signal energy assessment index has good stability and representativeness. Based on the vibration signal energy assessment index, the time period in which the vehicle is in the preset operating conditions and the operating state is stable can be effectively determined, which is beneficial to improving the accuracy of the determined target sub-time period.

[0038] S14. Among several specific sub-time periods corresponding to a specific time period, the specific sub-time period with the largest corresponding vibration signal energy assessment index shall be taken as the target sub-time period corresponding to the preset vehicle operating condition of that specific time period.

[0039] Through the above steps, for the first type of working condition, a fixed interception strategy is adopted to determine the target sub-time period. For the second type of working condition, a sliding window division strategy is adopted to obtain the specific sub-time period corresponding to the specific time period. The specific sub-time period with the largest vibration signal energy assessment index is taken as the target sub-time period. This ensures that the target dataset corresponding to the selected target sub-time period can truly reflect the vibration state and operating state of the compressor component under the corresponding preset vehicle operating conditions.

[0040] S2. For each target sub-time period corresponding to each preset vehicle operating condition, obtain the target dataset corresponding to the target sub-time period. The target dataset includes vibration data collected by each vibration sensor set at multiple locations on the compressor assembly and operating data of the target operating parameters of the compressor assembly within the target sub-time period. The target operating parameters are operating parameters related to the preset vehicle operating conditions.

[0041] Specifically, the target operating parameters include power consumption.

[0042] Furthermore, the target operating parameters also include intake pressure, exhaust pressure, and engine speed.

[0043] S3. Based on the target dataset corresponding to all target sub-time periods corresponding to each preset vehicle operating condition, obtain the data feature vector corresponding to each preset vehicle operating condition.

[0044] Specifically, step S3 includes the following sub-steps S31-S34: S31. For each preset vehicle operating condition, based on the target dataset corresponding to all target sub-time periods corresponding to the preset vehicle operating condition, obtain the total set of data sequence sets corresponding to the preset vehicle operating condition, including the data sequence set corresponding to each vibration sensor and the data sequence set corresponding to each target operating parameter; the data sequence set includes several data sequences, and the data sequences in the data sequence set correspond one-to-one with the target sub-time periods corresponding to the preset vehicle operating condition corresponding to the data sequence set.

[0045] Specifically, the data sequence in the data sequence set corresponding to the vibration sensor consists of the vibration data collected by the vibration sensor within the target sub-time period corresponding to the data sequence.

[0046] Furthermore, all vibration data in the data sequence corresponding to the vibration sensor are arranged in chronological order according to the time of acquisition.

[0047] Furthermore, the data sequence in the data sequence set corresponding to the target operating parameter consists of the operating data of the target operating parameter within the target sub-time period corresponding to the data sequence.

[0048] Furthermore, all the operational data in the data sequence corresponding to the target operational parameters are arranged in chronological order according to their acquisition time.

[0049] S32. For each set of data sequences, feature extraction is performed on each data sequence in the set of data sequences to obtain several first feature vectors corresponding to the set of data sequences. Those skilled in the art will know that any method in the prior art for feature extraction of data sequences to obtain corresponding feature vectors is within the protection scope of this invention, such as: time-domain statistical feature extraction method; frequency-domain feature extraction method; time-frequency domain feature extraction method; feature encoding method based on deep learning (using LSTM encoder, 1D-CNN or autoencoder to embed the data sequence), which will not be elaborated here.

[0050] S33. Cluster all the first feature vectors corresponding to the data sequence set to obtain several clusters, and take the cluster center vector of the cluster containing the most first feature vectors as the second feature vector corresponding to the data sequence set.

[0051] Specifically, the clustering algorithm used in step S33 is a clustering algorithm that does not require presetting the number of clusters, such as DBSCAN.

[0052] Specifically, each cluster contains several first feature vectors.

[0053] Furthermore, the cluster center vector of a cluster is the mean vector of all the first feature vectors contained in that cluster.

[0054] In a specific embodiment, if the number of first feature vectors corresponding to the data sequence set is less than a preset number threshold, then the mean vector of all first feature vectors corresponding to the data sequence set is used as the second feature vector corresponding to the data sequence set; if the number of first feature vectors corresponding to the data sequence set is not less than the preset number threshold, then step S33 is executed to obtain the second feature vector corresponding to the data sequence set; the preset number threshold is a number threshold preset by those skilled in the art according to actual needs, for example: 50, 60, which will not be elaborated here.

[0055] Through the above steps, when the number of first feature vectors corresponding to the data sequence set is less than the preset threshold, the mean vector of all first feature vectors corresponding to the data sequence set is used as the second feature vector corresponding to the data sequence set; this balances the stability of small samples with the pattern recognition ability of large samples, and avoids clustering when the number of samples is small, which would lead to unreasonable clustering results.

[0056] S34. Perform vector fusion on the second feature vector corresponding to all data sequence sets corresponding to the preset vehicle operating condition to obtain the data feature vector corresponding to the preset vehicle operating condition.

[0057] Specifically, the vector fusion in step S34 is achieved by concatenating vectors according to the first preset concatenation order corresponding to the data sequence set.

[0058] Through the above steps, based on the target dataset corresponding to all target sub-time periods corresponding to the preset vehicle operating conditions, a total set of data sequence sets corresponding to the preset vehicle operating conditions is obtained; features are extracted from each data sequence in the data sequence set to obtain several first feature vectors corresponding to the data sequence set; all first feature vectors corresponding to the data sequence set are clustered to obtain several clusters, and the cluster center vector of the cluster containing the most first feature vectors in the several clusters is used as the second feature vector corresponding to the data sequence set; the second feature vectors corresponding to all data sequence sets corresponding to the preset vehicle operating conditions are fused to obtain the data feature vector corresponding to the preset vehicle operating conditions; this not only preserves the operating condition response information of multiple locations and parameters, but also improves the representativeness of the second feature vector by obtaining the second feature vector through clustering. Based on the second feature vector, a data feature vector is obtained, and further, a comprehensive feature vector is obtained based on the data feature vector, providing high-quality and highly representative input for the subsequent fatigue life prediction model, thereby improving the accuracy of fatigue life prediction.

[0059] S4. Perform vector fusion on the data feature vectors corresponding to each preset vehicle operating condition to obtain a comprehensive feature vector.

[0060] Specifically, the vector fusion in step S4 is achieved by concatenating vectors according to the second preset concatenation order corresponding to the preset vehicle operating conditions or by averaging element by element.

[0061] Specifically, after step S3, the method further includes constructing a data feature vector set based on the data feature vectors corresponding to each preset vehicle operating condition, and using the current time point as the associated time point corresponding to the data feature vector set; and synchronously storing the data feature vector set together with its corresponding associated time point in the cloud and local database.

[0062] In one specific embodiment, prior to step S3, the method further includes: If at least one preset vehicle operating condition exists, and the corresponding target sub-time period is not detected within the target time period, then the preset vehicle operating condition for which the corresponding target sub-time period is not detected within the target time period will be marked as a missing condition, and the preset vehicle operating condition for which the corresponding target sub-time period is detected within the target time period will be marked as an identified condition. The data feature vectors S01-S03 corresponding to the missing conditions will be obtained through the following steps: S01. Query the local database to see if there is a set of data feature vectors with related time points within the key time period; the end time of the key time period is the current time point, and the duration is the fourth preset duration; the fourth preset duration is a duration longer than the first preset duration that is preset by those skilled in the art according to actual needs, such as 1 month or 2 months, which will not be elaborated here.

[0063] S02. If it exists, select the data feature vector corresponding to the missing working condition from the set of data feature vectors with the closest associated time point to the current time point in the local database, and use it as the data feature vector corresponding to the missing working condition. If it does not exist, generate a data feature vector completion request. The data feature vector completion request includes the data feature vectors corresponding to the identified working conditions and all missing working condition identifiers.

[0064] Specifically, the missing condition identifier is a unique identifier for the missing condition.

[0065] S03. Send the data feature vector completion request to the cloud to obtain the data feature vector corresponding to the missing working condition returned by the cloud.

[0066] Through the above steps, the system can proactively identify preset vehicle operating conditions that do not occur within the target time period, avoiding incomplete input to the fatigue life prediction model due to the absence of certain preset vehicle operating conditions within the target time period, thus preventing a decrease in the accuracy of fatigue life prediction. Specifically, it prioritizes retrieving recent data feature vector sets from the local historical database, achieving efficient and low-latency local completion, ensuring data relevance while reducing reliance on the network and cloud. When no local data is available, a data feature vector completion request is generated and sent to the cloud, facilitating the cloud's accurate inference of reasonable feature vectors for missing operating conditions based on a wider range of historical data. This effectively solves the problem of incomplete data coverage, ensuring that a comprehensive feature vector can be obtained regardless of whether the target time period includes all target sub-time periods corresponding to preset vehicle operating conditions. This provides the fatigue life prediction model with complete, reasonable, and representative input features, significantly improving the reliability and applicability of fatigue life prediction in real-world, complex usage scenarios.

[0067] Specifically, after the cloud receives a data feature vector completion request, it executes the following steps S001-S005: S001. According to the preset vector concatenation order, concatenate all the data feature vectors corresponding to the identified working conditions included in the data feature vector completion request into the first vector.

[0068] S002. Take the set of data feature vectors corresponding to other vehicles of the same model as the vehicle corresponding to the data feature vector completion request as the associated feature vector set.

[0069] S003. For each set of associated feature vectors, according to the preset vector concatenation order, extract the data feature vectors corresponding to each identified working condition in the data feature vector completion request from the set of associated feature vectors, and concatenate the extracted data feature vectors into a second vector.

[0070] S004. Obtain the vector similarity between the first vector and each second vector.

[0071] Specifically, the vector similarity is no less than 0 and no greater than 1. The greater the vector similarity, the more similar the first vector is to its corresponding second vector.

[0072] S005. If the maximum value among all vector similarities corresponding to the first vector exceeds a preset similarity threshold, then the data feature vector corresponding to the missing working condition in the associated feature vector set corresponding to the maximum value is taken as the data feature vector corresponding to the missing working condition, so as to generate a completion result and send it to the vehicle corresponding to the data feature vector completion request.

[0073] Specifically, the preset similarity threshold ranges from 0.7 to 0.9.

[0074] Through the above steps, the data feature vectors corresponding to all identified operating conditions included in the data feature vector completion request are concatenated into a first vector according to a preset vector concatenation order; the corresponding preset vehicle operating conditions data feature vectors are extracted from the data feature vector set of other vehicles of the same model, and these data feature vectors are concatenated into a second vector; the matching degree between the first and second vectors is determined based on vector similarity, and after finding a highly similar second vector, the feature vectors corresponding to the missing operating conditions in the data feature vector set corresponding to the second vector are directly reused for completion. This achieves a reasonable, efficient, and interpretable estimation of the missing operating condition features, avoids prediction bias caused by random filling or zero-value substitution, and ensures that the comprehensive feature vector input to the fatigue life prediction model is complete and has a clear physical meaning. This allows the fatigue life prediction model to be provided with complete, reasonable, and representative input features regardless of whether the target time period includes all preset vehicle operating conditions corresponding to the target sub-time period, significantly improving the reliability and applicability of fatigue life prediction in real and complex usage scenarios.

[0075] Specifically, if the maximum value among all similarities of the first vector does not exceed the preset similarity threshold, an abnormal prompt message is sent to the vehicle corresponding to the data feature vector completion request, indicating that completion cannot be performed.

[0076] Specifically, if at least one data feature vector obtained through steps S01-S03 exists in the data feature vector set, then the data feature vector set will not be stored in the local database or the cloud.

[0077] S5. Input the comprehensive feature vector into the fatigue life prediction model to obtain the fatigue life value of the compressor component.

[0078] Specifically, the fatigue life value is the remaining fatigue life of the compressor assembly in the current state.

[0079] Through the above steps, at least one target sub-time period corresponding to each preset vehicle operating condition is determined from the target time period, and the target dataset corresponding to each target sub-time period is obtained. The target dataset includes vibration data collected by vibration sensors located at multiple positions on the compressor assembly and operating data of the target operating parameters of the compressor assembly within the target sub-time period. Based on the target datasets corresponding to all target sub-time periods for each preset vehicle operating condition, the data feature vectors corresponding to each preset vehicle operating condition are obtained. Vector fusion is performed on the data feature vectors corresponding to each preset vehicle operating condition to obtain a comprehensive feature vector. The comprehensive feature vector is input into the fatigue life prediction model to obtain the fatigue life value of the compressor assembly. This fully considers the additional impact of external vibrations and bumps generated by the vehicle under different preset vehicle operating conditions on the compressor assembly, significantly improving the accuracy of fatigue life prediction and truly reflecting the durability performance of the compressor assembly.

[0080] Specifically, the fatigue life prediction model is obtained by supervising the linear regression model using a pre-set training sample set.

[0081] Specifically, the preset training sample set includes several preset training samples, and each preset training sample includes an input feature and a manually labeled output label corresponding to that input feature.

[0082] Further, the input feature is a historical comprehensive feature vector, which is obtained based on the operating data of the target operating parameters collected before the failure of the compressor component that has already experienced fatigue failure in the test vehicle, and the vibration data collected by various vibration sensors set at multiple locations of the compressor component; the output label is a manually labeled fatigue life value, which represents the actual remaining fatigue life of the compressor component at the generation time point corresponding to its historical comprehensive feature vector; wherein, the historical comprehensive feature vector is obtained in the same way as the comprehensive feature vector obtained in steps S1-S4.

[0083] Specifically, based on a compressor component that has experienced fatigue failure in a test vehicle, multiple preset training samples can be obtained; the preset training sample set includes several preset training samples derived from compressor components that have experienced fatigue failure in multiple test vehicles.

[0084] Through the above steps, a pre-set training sample set is constructed based on the operating data of the target operating parameters collected before the failure of the compressor component that has already experienced fatigue failure in the test vehicle, and the vibration data collected by various vibration sensors set at multiple locations on the compressor component. The pre-set training sample set is used to conduct supervised training on the linear regression model to obtain the fatigue life prediction model. This ensures that the model learns a real and reliable "comprehensive feature-fatigue life" mapping relationship, significantly improving the model's generalization ability and engineering applicability, thereby providing high-precision and interpretable fatigue life prediction results in practical applications.

[0085] Embodiments of the present invention also provide a non-transitory computer-readable storage medium that can be disposed in an electronic device to store a computer program related to implementing a method in the method embodiments, the computer program being loaded and executed by the processor to implement the method provided in the above embodiments.

[0086] Embodiments of the present invention also provide an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method provided in the above embodiments.

[0087] Embodiments of the present invention also provide a computer program product including program code, which, when the program product is run on an electronic device, causes the electronic device to perform the steps of the methods described above in various exemplary embodiments of the present invention.

[0088] This invention provides a method, device, and medium for predicting the fatigue life of a compressor assembly. The method involves determining at least one target sub-time period corresponding to each preset vehicle operating condition from a target time period, and obtaining a target dataset corresponding to each target sub-time period. The target dataset includes vibration data collected by vibration sensors located at multiple positions on the compressor assembly within the target sub-time period, and operating data of the compressor assembly's target operating parameters. Based on the target datasets corresponding to all target sub-time periods for each preset vehicle operating condition, a data feature vector corresponding to each preset vehicle operating condition is obtained. Vector fusion is performed on the data feature vectors corresponding to each preset vehicle operating condition to obtain a comprehensive feature vector. The comprehensive feature vector is input into a fatigue life prediction model to obtain the fatigue life value of the compressor assembly. As can be seen, this invention acquires vibration data and operating data of the compressor component's target operating parameters within the target sub-time period corresponding to each preset vehicle operating condition experienced by the vehicle in actual operation. It extracts vibration characteristics and operating characteristics under each preset vehicle operating condition and integrates them to form a comprehensive feature that can fully reflect the real usage environment. Then, fatigue life prediction is performed based on this comprehensive feature, which fully considers the additional impact of external vibration and bumps generated by the vehicle under different preset vehicle operating conditions on the compressor component, significantly improving the accuracy of fatigue life prediction and truly reflecting the durability performance of the compressor component.

[0089] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the invention.

Claims

1. A method for predicting the fatigue life of compressor components, characterized in that, The method includes the following steps: S1. Determine at least one target sub-time period corresponding to each preset vehicle operating condition from the target time period; the end time of the target time period is the current time point, and the duration is the first preset duration t1; The target sub-time period corresponding to the preset vehicle operating condition is the time period during which the vehicle is in the preset operating condition and the operating status is stable within the target time period. S2. For each target sub-time period corresponding to each preset vehicle operating condition, obtain the target dataset corresponding to the target sub-time period. The target dataset includes vibration data collected by each vibration sensor set at multiple locations of the compressor assembly and operating data of the target operating parameters of the compressor assembly within the target sub-time period. The compressor assembly is located inside the vehicle; the target operating parameters are those related to preset vehicle operating conditions. S3. Based on the target dataset corresponding to all target sub-time periods corresponding to each preset vehicle operating condition, obtain the data feature vector corresponding to each preset vehicle operating condition. S4. Perform vector fusion on the data feature vectors corresponding to each preset vehicle operating condition to obtain a comprehensive feature vector; S5. Input the comprehensive feature vector into the fatigue life prediction model to obtain the fatigue life value of the compressor component.

2. The method for predicting the fatigue life of compressor components according to claim 1, characterized in that, Step S1 includes the following sub-steps: S11. Obtain several specific time periods corresponding to each preset vehicle operating condition. The specific time periods corresponding to the preset vehicle operating conditions are the time periods during which the vehicle is in the preset vehicle operating condition within the target time period. S12. When the preset vehicle operating condition is the first type of operating condition, for each specific time period corresponding to the preset vehicle operating condition, a time period of length t2 is extracted from the start time of the specific time period as the target sub-time period corresponding to the preset vehicle operating condition; when the preset vehicle operating condition is the second type of operating condition, for each specific time period corresponding to the preset vehicle operating condition, a sliding window with a window length of t2 and a sliding step size of a third preset time period t3 is used to divide the specific time period to obtain several specific sub-time periods corresponding to the specific time period; wherein, the first type of operating condition is a start-stop transient operating condition; the second type of operating condition is a driving dynamic operating condition; t3 < t2 < t1. S13. Based on the vibration data collected by each vibration sensor, obtain the vibration signal energy assessment index corresponding to each specific sub-time period. S14. Among several specific sub-time periods corresponding to a specific time period, the specific sub-time period with the largest corresponding vibration signal energy assessment index shall be taken as the target sub-time period corresponding to the preset vehicle operating condition of that specific time period.

3. The method for predicting the fatigue life of compressor components according to claim 2, characterized in that, Step S13 includes the following sub-steps: S131. For each specific sub-time period, obtain the vibration dataset corresponding to the specific sub-time period, including the vibration data sequence corresponding to each vibration sensor, which is composed of the vibration data collected by its corresponding vibration sensor within the specific sub-time period. S132. For each vibration data sequence, calculate the RMS value corresponding to the vibration data sequence based on all vibration data in the vibration data sequence. S133. The average value of the RMS values ​​of all vibration data sequences in the vibration dataset corresponding to a specific sub-time period is used as the vibration signal energy assessment index for that specific sub-time period.

4. The method for predicting the fatigue life of compressor components according to claim 1, characterized in that, Step S3 includes the following sub-steps: S31. For each preset vehicle operating condition, based on the target dataset corresponding to all target sub-time periods corresponding to the preset vehicle operating condition, obtain the total set of data sequence sets corresponding to the preset vehicle operating condition, including the data sequence set corresponding to each vibration sensor and the data sequence set corresponding to each target operating parameter. The data sequence set includes several data sequences, and the data sequences in the data sequence set correspond one-to-one with the target sub-time periods corresponding to the preset vehicle operating conditions of the data sequence set; S32. For each set of data sequences, perform feature extraction on each data sequence in the set of data sequences to obtain several first feature vectors corresponding to the set of data sequences. S33. Cluster all the first feature vectors corresponding to the data sequence set to obtain several clusters, and take the cluster center vector of the cluster containing the most first feature vectors as the second feature vector corresponding to the data sequence set. S34. Perform vector fusion on the second feature vector corresponding to all data sequence sets corresponding to the preset vehicle operating condition to obtain the data feature vector corresponding to the preset vehicle operating condition.

5. The method for predicting the fatigue life of compressor components according to claim 4, characterized in that, The data sequence in the data sequence set corresponding to the vibration sensor consists of vibration data collected by the vibration sensor within the target sub-time period corresponding to the data sequence.

6. The method for predicting the fatigue life of compressor components according to claim 4, characterized in that, The data sequence in the data sequence set corresponding to the target operating parameter consists of the operating data of the target operating parameter within the target sub-time period corresponding to the data sequence.

7. The method for predicting the fatigue life of compressor components according to claim 1, characterized in that, The fatigue life prediction model is obtained by supervising the linear regression model using a pre-set training sample set.

8. The method for predicting the fatigue life of compressor components according to claim 2, characterized in that, The duration of a specific time period is not less than t2.

9. A non-transitory computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is loaded and executed by a processor to implement the compressor component fatigue life prediction method as described in any one of claims 1-8.

10. An electronic device, comprising: A processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the compressor component fatigue life prediction method as described in any one of claims 1-8.

Images