Test bed part life evaluation method, training method, device and equipment

Abstract

The invention provides a test bed part life evaluation method, a training method, a training device and equipment, which can be applied to the technical field of machinery and the technical field of artificial intelligence. The test bench component life evaluation method comprises the following steps: obtaining maintenance data and test environment data of a to-be-evaluated component in a test bench, and vibration time sequence data, temperature time sequence data and current time sequence data about the to-be-evaluated component acquired by a plurality of sensors of different types; performing feature extraction on the maintenance data and the test environment data to obtain static features, and performing feature extraction on time sequence splicing data obtained by splicing the vibration time sequence data, the temperature time sequence data and the current time sequence data to obtain time sequence features; based on a first fusion weight and a second fusion weight determined by the static feature and the time sequence feature, performing feature fusion on the static feature and the time sequence feature to obtain a fusion feature; and carrying out life evaluation on the to-be-evaluated part based on the fusion features.

Description

Technical Field

[0001] This disclosure relates to the fields of mechanical technology and artificial intelligence technology, and more specifically, to a method, training method, apparatus, and equipment for assessing the lifespan of test bench components. Background Technology

[0002] The performance and stability of test bench components directly affect the efficiency and quality of testing. However, due to wear, fatigue, corrosion, and other reasons, components gradually age during operation, leading to performance degradation or even failure. Therefore, it is particularly important to conduct life assessments of test bench components and develop corresponding maintenance strategies.

[0003] In realizing the concept disclosed herein, the inventors discovered at least the following problems in the related technology: traditional life assessment methods rely on a single type of data, which does not make full use of information, resulting in inaccurate life assessment results obtained from life assessment of test bench components. Summary of the Invention

[0004] In view of this, the present disclosure provides a method, training method, apparatus, and equipment for evaluating the lifespan of test bench components.

[0005] One aspect of this disclosure provides a method for assessing the lifespan of a test bench component, comprising: acquiring maintenance data, test environment data, and multiple sets of time-series operating state data of the component under assessment in different modes collected by multiple sensors of different types in the test bench, wherein the multiple sets of time-series operating state data include vibration time-series data, temperature time-series data, and current time-series data; extracting features from the maintenance data and test environment data to obtain static features, and extracting features from the time-series spliced ​​data obtained by splicing vibration time-series data, temperature time-series data, and current time-series data to obtain time-series features; fusing the static features and time-series features based on a first fusion weight and a second fusion weight determined by the static features and time-series features to obtain fused features; assessing the lifespan of the component under assessment based on the fused features to obtain an initial lifespan assessment result for the component under assessment; determining the lifespan change rate based on the initial lifespan assessment result and the historical lifespan assessment results for the component under assessment; and determining the initial lifespan assessment result as the target lifespan assessment result if the lifespan change rate meets a predetermined lifespan change rate threshold.

[0006] According to embodiments of this disclosure, the test bench component life assessment method further includes: acquiring maintenance records, which include historical maintenance times and historical fault types for each maintenance of the component to be assessed; determining maintenance attenuation factors based on multiple historical maintenance times and the current time; using the mapping relationship between fault types and fault codes to perform feature encoding on multiple historical fault types to obtain multiple historical fault codes; and using the maintenance attenuation factors and multiple historical fault codes as maintenance data.

[0007] According to embodiments of this disclosure, the multiple sensors each have different acquisition frequencies; the test bench component life assessment method further includes: performing timestamp alignment processing on vibration time series data, temperature time series data and current time series data to obtain target vibration time series data, target temperature time series data and target current time series data; and splicing the target vibration time series data, target temperature time series data and target current time series data to obtain time series spliced ​​data.

[0008] According to embodiments of this disclosure, time-series vibration data, temperature data, and current data are timestamped to obtain target vibration time-series data, target temperature time-series data, and target current time-series data. This includes: fitting the vibration time-series data, temperature data, and current time-series data to obtain vibration curves, temperature curves, and current curves; verifying the fluctuation degree of each of the vibration time-series data, temperature data, and current time-series data based on their respective amplitudes to obtain verification results; and performing timestamping on the vibration time-series data, temperature data, and current time-series data when the verification results indicate that the amplitudes of the vibration time-series data, temperature data, and current time-series data are less than a predetermined amplitude threshold, to obtain target vibration time-series data, target temperature time-series data, and target current time-series data.

[0009] According to embodiments of this disclosure, based on a first fusion weight and a second fusion weight determined based on static features and temporal features, feature fusion is performed on static features and temporal features to obtain fused features, including: using the weight generation layer of the component life assessment model to determine the first fusion weight and the second fusion weight based on the splicing features of static features and temporal features; and using the feature fusion layer of the component life assessment model to perform feature fusion on static features and temporal features based on the first fusion weight and the second fusion weight to generate fused features.

[0010] According to embodiments of this disclosure, feature extraction is performed on maintenance data and test environment data to obtain static features, and feature extraction is performed on time-series spliced ​​data obtained by splicing vibration time-series data, temperature time-series data and current time-series data to obtain time-series features. This includes: using the first extraction layer of the component life assessment model, using a multilayer perceptron to extract features from maintenance data and test environment data to obtain static features; and using the second extraction layer of the component life assessment model, using a gated loop unit to extract features from time-series operating state data to obtain time-series features.

[0011] According to embodiments of this disclosure, the life assessment result includes the remaining life of the component to be assessed, and the method further includes: determining a remaining life range based on a preset confidence level and the remaining life; comparing the minimum value of the remaining life range with a preset remaining life threshold to obtain a comparison result; and generating a warning message for the component to be assessed if the comparison result indicates that the minimum value of the remaining life range is less than the preset remaining life threshold.

[0012] Another aspect of this disclosure provides a training method for a component life assessment model, comprising: acquiring sample maintenance data, sample test environment data, and multiple sets of sample time-series operating state data of different modes of the sample component collected by multiple sensors of different types in a sample test bench, wherein the multiple sets of sample time-series operating state data include sample vibration time-series data, sample temperature time-series data, and sample current time-series data; using an initial component life assessment model, performing feature extraction on the sample maintenance data and sample test environment data to obtain sample static features, and performing analysis on the sample time-series spliced ​​data obtained by splicing the sample vibration time-series data, sample temperature time-series data, and sample current time-series data. Feature extraction is performed to obtain sample time-series features. Using the initial component lifetime assessment model, the sample static features and sample time-series features are fused according to the first and second sample fusion weights determined based on the sample static features and sample time-series features to obtain sample fusion features. Based on the sample fusion features, the lifetime of the sample component is assessed to obtain the sample lifetime assessment result for the component to be assessed. Based on the actual remaining lifetime of the sample component, the physical degradation assessment result, and the sample lifetime assessment result, the initial component lifetime assessment model is trained to obtain the component lifetime assessment model. The physical degradation assessment result is obtained by assessing the lifetime of the sample component based on the physical state parameters of the sample component.

[0013] According to embodiments of this disclosure, an initial component life assessment model is trained based on the actual remaining life of the sample component, the physical degradation assessment result, and the sample life assessment result to obtain a component life assessment model. This includes: determining a first loss value based on the actual remaining life and the sample life assessment result; determining a second loss value based on the physical degradation assessment result and the sample life assessment result; and adjusting the parameters of the initial component life assessment model based on the first loss value and the second loss value until a preset termination condition is met to obtain the component life assessment model.

[0014] According to embodiments of this disclosure, the physical degradation assessment result is determined as follows: based on the energy entropy of the vibration signal of the sample component, the current crack length of the sample component is determined by utilizing the mapping relationship between the energy entropy and the crack; based on the current crack length and a preset crack length threshold, the life of the sample component is assessed using the crack propagation equation to obtain the physical degradation assessment result.

[0015] Another aspect of this disclosure provides a test bench component life assessment device, comprising: an acquisition module for acquiring maintenance data, test environment data, and multiple sets of time-series operating state data of the component to be assessed in different modes collected by multiple sensors of different types, the multiple sets of time-series operating state data including vibration time-series data, temperature time-series data, and current time-series data; an extraction module for extracting features from the maintenance data and test environment data to obtain static features, and extracting features from the time-series spliced ​​data obtained by splicing vibration time-series data, temperature time-series data, and current time-series data to obtain time-series features; a fusion module for fusing the static features and time-series features based on a first fusion weight and a second fusion weight determined by the static features and time-series features to obtain fused features; an assessment module for performing a life assessment on the component to be assessed based on the fused features to obtain an initial life assessment result for the component to be assessed; a change determination module for determining the life change rate based on the initial life assessment result and historical life assessment results for the component to be assessed; and a life determination module for determining the initial life assessment result as the target life assessment result if the life change rate meets a predetermined life change rate threshold.

[0016] Another aspect of this disclosure provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the methods described above.

[0017] Another aspect of this disclosure provides a computer-readable storage medium storing computer-executable instructions that, when executed, are used to implement the methods described above.

[0018] Another aspect of this disclosure provides a computer program product including computer-executable instructions that, when executed, are used to implement the methods described above.

[0019] According to embodiments of this disclosure, multidimensional data is used for life assessment through feature extraction and feature fusion. During the fusion process, a dynamic weight fusion mechanism adaptively balances the importance of time-series operational status data with non-time-series maintenance and test environment data. Based on dynamic weights, time-series and non-time-series static features are fused, overcoming the limitations of traditional fixed-weight splicing or average fusion, improving the utilization rate of multidimensional data, and further enhancing the accuracy of component life assessment. Furthermore, by combining historical life assessment results with the initial life assessment results, the obtained target life assessment results are ensured to conform to life change patterns, further improving the accuracy of life assessment. Attached Figure Description

[0020] The above and other objects, features and advantages of this disclosure will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0021] Figure 1 The illustrations illustrate application scenarios of the test bench component life assessment method, training method, apparatus, and equipment according to embodiments of the present disclosure;

[0022] Figure 2 A flowchart illustrating a method for evaluating the lifespan of test bench components according to an embodiment of the present disclosure is shown schematically.

[0023] Figure 3 A schematic diagram illustrating the determination of fusion features according to an embodiment of the present disclosure is shown;

[0024] Figure 4 The illustration shows a schematic diagram of generating initial lifetime assessment results according to a specific embodiment of the present disclosure;

[0025] Figure 5 A flowchart illustrating a method for training a component life assessment model according to an embodiment of the present disclosure is shown schematically.

[0026] Figure 6 A block diagram of a test bench component life assessment apparatus according to an embodiment of the present disclosure is schematically shown; and

[0027] Figure 7 A block diagram of an electronic device suitable for implementing a test bench component life assessment method according to an embodiment of the present disclosure is shown schematically. Detailed Implementation

[0028] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0029] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0030] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0031] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0032] In the embodiments disclosed herein, the collection, updating, analysis, processing, use, transmission, provision, disclosure, and storage of data (e.g., including but not limited to user personal information) comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures have been taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0033] In the embodiments disclosed herein, user authorization or consent is obtained before acquiring or collecting user personal information.

[0034] Traditional life prediction methods rely on a single type of data (such as maintenance records) and cannot effectively integrate multi-source heterogeneous information such as time-series sensor data (vibration, temperature), non-time-series maintenance records (fault type, maintenance frequency) and operating loads. This results in insufficient information being used when assessing the life of test bench components, leading to inaccurate life assessment results.

[0035] This disclosure provides a method for assessing the lifespan of a test bench component, comprising: acquiring maintenance data, test environment data, and multiple sets of time-series operating state data of the component to be assessed in different modes, collected by multiple sensors of different types, including vibration time-series data, temperature time-series data, and current time-series data; extracting features from the maintenance data and test environment data to obtain static features, and extracting features from the time-series spliced ​​data obtained by splicing vibration time-series data, temperature time-series data, and current time-series data to obtain time-series features; fusing the static features and time-series features based on a first fusion weight and a second fusion weight determined by the static features and time-series features to obtain fused features; assessing the lifespan of the component to be assessed based on the fused features to obtain an initial lifespan assessment result for the component to be assessed; determining the lifespan change rate based on the initial lifespan assessment result and the historical lifespan assessment results for the component to be assessed; and determining the initial lifespan assessment result as the target lifespan assessment result if the lifespan change rate meets a predetermined lifespan change rate threshold.

[0036] The embodiments of this disclosure utilize multimodal data to assess the lifespan of the component under evaluation by employing methods such as feature extraction and feature fusion. During the fusion process, a dynamic weight fusion mechanism is used to adaptively balance the contributions of temporal sensor features and non-temporal maintenance / environmental features. Based on dynamic weights, temporal and non-temporal features are fused, overcoming the limitations of traditional fixed-weight splicing or average fusion, improving the utilization rate of multimodal data, and further enhancing the accuracy of component lifespan assessment.

[0037] Figure 1 The illustration schematically depicts application scenarios of the test bench component life assessment method, training method, apparatus, and equipment according to embodiments of this disclosure. It should be noted that... Figure 1 The examples shown are merely examples of system architectures that can be applied to the embodiments of this disclosure, in order to help those skilled in the art understand the technical content of this disclosure, but do not mean that the embodiments of this disclosure cannot be used in other devices, systems, environments or scenarios.

[0038] like Figure 1 As shown, the system architecture according to this embodiment may include a test bench 110 and a test bench control cabinet 120, wherein the test bench 110 includes a component 111 to be evaluated.

[0039] The test bench 110 can be a test bench for rail train-related products, such as a bogie fatigue test bench or a whole vehicle vibration test bench. The component to be evaluated 111 can be a core component of the test bench 110, such as a hydraulic cylinder, gearbox, bearing, etc.

[0040] The test bench control cabinet 120 can be used to control the test bench 110 to conduct tests and can perform life assessments on components in the test bench 110, such as the component 111 to be evaluated.

[0041] It should be noted that the test bench component life assessment method provided in this disclosure embodiment can generally be executed by the test bench control cabinet 120. Accordingly, the test bench component life assessment device provided in this disclosure embodiment can generally be disposed in the test bench control cabinet 120. In some embodiments, the life assessment device can be an embedded chip integrated into the test bench control cabinet 120.

[0042] The test bench component life assessment method provided in this disclosure can also be performed by a device different from the test bench control cabinet 120 but capable of communicating with the test bench 110 and / or the test bench control cabinet 120. Accordingly, the test bench component life assessment device provided in this disclosure can also be located in a device different from the test bench control cabinet 120 but capable of communicating with the test bench 110 and / or the test bench control cabinet 120.

[0043] It should be understood that Figure 1 The number of test benches, test bench control cabinets, and components to be evaluated shown in the diagram is merely illustrative. Any number of test benches, test bench control cabinets, and components to be evaluated can be included depending on implementation requirements.

[0044] Figure 2 A flowchart illustrating a method for evaluating the lifespan of test bench components according to an embodiment of the present disclosure is shown schematically.

[0045] like Figure 2 As shown, the method includes operations S210~S260.

[0046] During operation S210, maintenance data, test environment data, and multiple sets of time-series operating status data of the component to be evaluated in different modes are acquired from multiple sensors of different types in the test bench.

[0047] During operation of S220, feature extraction is performed on maintenance data and test environment data to obtain static features, and feature extraction is performed on time sequence spliced ​​data obtained by splicing vibration time sequence data, temperature time sequence data and current time sequence data to obtain time sequence features.

[0048] In operation S230, based on the first fusion weight and the second fusion weight determined by the static features and the temporal features, feature fusion is performed on the static features and the temporal features to obtain the fused features.

[0049] In operation S240, based on the fusion features, a lifetime assessment is performed on the component to be evaluated, and an initial lifetime assessment result is obtained for the component to be evaluated.

[0050] In operation S250, the life change rate is determined based on the initial life assessment results and the historical life assessment results for the component to be assessed.

[0051] In operation S260, if the rate of change of lifetime meets the predetermined rate of change of lifetime threshold, the initial lifetime assessment result is determined as the target lifetime assessment result.

[0052] According to embodiments of this disclosure, in order to improve the accuracy of life assessment, the life assessment of the component under assessment is performed using data from multiple dimensions and modes of the component under assessment. Specifically, the data used for life assessment of the component under assessment may include maintenance data of the component under assessment, test environment data, and multiple sets of time-series operating status data of the component under assessment in different modes collected by multiple sensors of different types.

[0053] According to embodiments of this disclosure, maintenance data can be obtained from the maintenance records of the test bench. For example, the maintenance records may include maintenance time and fault type of the test bench, and maintenance data related to the component to be evaluated can be obtained from the maintenance records.

[0054] According to embodiments of this disclosure, the test environment data can be relevant data about the environment in which the component to be evaluated is located, such as the load intensity of the test bench, ambient temperature and humidity, etc., and can also include the cumulative operating time of the component to be evaluated.

[0055] According to embodiments of this disclosure, the timing operation status data can be obtained by collecting the operating status of the component to be evaluated using sensors during the operation of the test bench and arranging it in chronological order, such as vibration timing data, temperature timing data and current timing data.

[0056] Since the acquired data is multi-dimensional and multi-modal heterogeneous data, it is necessary to perform feature extraction and feature fusion on this data in order to convert it into a unified, model-recognizable form, so as to more effectively and reliably utilize this multi-dimensional and multi-modal heterogeneous data to evaluate the life of the component under evaluation.

[0057] During feature extraction, features can be extracted from static data such as maintenance data and test environment data, as well as from time-series data obtained by splicing together vibration time-series data, temperature time-series data, and current time-series data. In some embodiments, different models can be used to extract features from static data and time-series data respectively.

[0058] When performing feature fusion, since static features and temporal features each have different importance, and the importance of static features and temporal features varies for different components to be evaluated, the weights for fusing static features and temporal features can be determined based on the degree of influence of the information represented by each static feature and temporal feature on the lifespan of the component to be evaluated. Specifically, the first fusion weight can be the fusion weight of the static features, and the second fusion weight can be the fusion weight of the temporal features.

[0059] In some embodiments, a specific activation function can be used to process static features and temporal features to obtain a first fusion weight and a second fusion weight. The first fusion weight and the second fusion weight are then used to perform weighted fusion of static features and temporal features to obtain fused features. The fused features containing multi-dimensional information can then be used to evaluate the lifespan of the component to be evaluated.

[0060] When performing lifetime assessment on a component based on fused features, the fused features can be input into a fully connected layer for nonlinear transformation to obtain an initial lifetime assessment result. Specifically, the initial lifetime assessment result can be the remaining lifetime of the component under assessment, such as 10 hours.

[0061] To further improve the accuracy of life assessment results, after obtaining the initial life assessment results, the historical life assessment results of the component to be assessed can be combined to determine the life change rate of the component to be assessed, so as to determine whether the current initial life assessment results conform to the life change pattern.

[0062] For example, if the current remaining life, as represented by the initial life assessment result, is greater than the historical remaining life obtained from the previous life assessment of the component under assessment, it can be determined that the initial life assessment result does not conform to the life change pattern, and a new life assessment can be performed. Similarly, if the difference between the current remaining life and the historical remaining life is much greater than the time interval between the two life assessments, it can also be determined that the initial life assessment result does not conform to the life change pattern.

[0063] If the initial life assessment result conforms to the life change pattern, the initial life assessment result can be determined as the target life assessment result, and the target life assessment result can be output so that the maintenance personnel of the test bench can perform timely maintenance on the component to be assessed based on the target life assessment result.

[0064] According to embodiments of this disclosure, multidimensional data is used for life assessment through feature extraction and feature fusion. During the fusion process, a dynamic weight fusion mechanism adaptively balances the importance of time-series operational status data with non-time-series maintenance and test environment data. Based on dynamic weights, time-series and non-time-series static features are fused, overcoming the limitations of traditional fixed-weight splicing or average fusion, improving the utilization rate of multidimensional data, and further enhancing the accuracy of component life assessment. Furthermore, by combining historical life assessment results with the initial life assessment results, the obtained target life assessment results are ensured to conform to life change patterns, further improving the accuracy of life assessment.

[0065] Figure 3 A schematic diagram illustrating the determination of fusion features according to an embodiment of the present disclosure is shown.

[0066] like Figure 3 As shown, a first fusion feature 21 and a second fusion feature 22 are determined based on static feature 11 and temporal feature 12. The static feature 11 and temporal feature 12 are then weighted and fused using the first fusion feature 21 and the second fusion feature 22 to obtain fusion feature 30.

[0067] According to embodiments of this disclosure, the test bench component life assessment method further includes: acquiring maintenance records; determining a maintenance attenuation factor based on multiple historical maintenance times and the current time; using the mapping relationship between fault types and fault codes to perform feature encoding on multiple historical fault types to obtain multiple historical fault codes; and using the maintenance attenuation factor and multiple historical fault codes as maintenance data.

[0068] In embodiments of this disclosure, maintenance data can be determined based on maintenance records. Since maintenance records are typically in text format, they need to be encoded or processed to obtain maintenance data, which facilitates the identification and processing of the maintenance data.

[0069] The maintenance record can include the historical fault type corresponding to each maintenance time of the component to be evaluated. To ensure the dynamic decay characteristic of maintenance effect, an exponential decay model can be used to determine the maintenance decay factor to quantify the decay of maintenance effect. The process of determining the maintenance decay factor can be shown in the following formula (1):

[0070] (1);

[0071] Among them, W repair This represents the maintenance attenuation factor, where n represents the total number of repairs to the component being evaluated, and t represents the maintenance attenuation factor. current Indicates the current time, t k This represents the historical maintenance time of the k-th maintenance, and λ is used to control the decay rate of the maintenance impact, which is usually taken as 0.01.

[0072] In embodiments of this disclosure, a mapping relationship between fault types and fault codes can be pre-defined, and the fault codes corresponding to historical fault types can be determined as historical fault codes, thereby converting textual fault types into machine-recognizable historical fault codes. In some embodiments, fault types can be encoded using one-hot encoding.

[0073] After obtaining the maintenance attenuation factor and historical fault codes, the maintenance attenuation factor and multiple historical fault codes can be used as maintenance data.

[0074] According to embodiments of this disclosure, by quantifying multiple historical maintenance moments into maintenance attenuation factors that characterize the attenuation of maintenance effectiveness, and encoding multiple historical maintenance types into historical fault codes, the obtained maintenance data can accurately characterize the attenuation of maintenance effectiveness, thereby further improving the accuracy of the life assessment process.

[0075] According to embodiments of this disclosure, the test bench component life assessment method further includes: performing timestamp alignment processing on vibration time series data, temperature time series data, and current time series data to obtain target vibration time series data, target temperature time series data, and target current time series data; and splicing the target vibration time series data, target temperature time series data, and target current time series data to obtain time series spliced ​​data.

[0076] In the embodiments of this disclosure, the acquisition frequencies of multiple sensors may differ, thus requiring timestamp alignment processing of the vibration time-series data, temperature time-series data, and current time-series data. For example, if the acquisition frequency corresponding to the vibration time-series data is higher than that of the temperature and current time-series data, interpolation processing can be performed on the temperature and current time-series data based on the acquisition frequency of the vibration time-series data to achieve timestamp alignment of the vibration, temperature, and current time-series data, thereby obtaining the target vibration time-series data, target temperature time-series data, and target current time-series data.

[0077] In some embodiments, a sliding window can be used to segment and align vibration time-series data, temperature time-series data, and current time-series data to obtain windowed target vibration time-series data, target temperature time-series data, and target current time-series data.

[0078] After timestamp alignment, the target vibration time series data, target temperature time series data, and target current time series data have the same number of columns, and can be vertically spliced ​​to obtain spliced ​​time series data.

[0079] According to embodiments of this disclosure, by performing timestamp alignment processing on vibration time-series data, temperature time-series data, and current time-series data, and then stitching together the aligned target vibration time-series data, target temperature time-series data, and target current time-series data, the resulting time-series stitched data can accurately and systematically represent the time-series operating characteristics of the component to be evaluated, thereby improving the accuracy of life assessment.

[0080] According to embodiments of this disclosure, time-series vibration data, temperature data, and current data are timestamped to obtain target vibration time-series data, target temperature time-series data, and target current time-series data. This includes: fitting the vibration time-series data, temperature data, and current time-series data to obtain vibration curves, temperature curves, and current curves; verifying the fluctuation degree of each of the vibration time-series data, temperature data, and current time-series data based on their respective amplitudes to obtain verification results; and performing timestamping on the vibration time-series data, temperature data, and current time-series data when the verification results indicate that the amplitudes of the vibration time-series data, temperature data, and current time-series data are less than a predetermined amplitude threshold, to obtain target vibration time-series data, target temperature time-series data, and target current time-series data.

[0081] Before performing timestamp alignment, vibration timing data, temperature timing data, and current timing data can be verified. Valid vibration timing data, temperature timing data, and current timing data can be used for lifetime assessment, improving the accuracy of lifetime assessment.

[0082] When verifying vibration time series data, temperature time series data, and current time series data, the vibration time series data, temperature time series data, and current time series data can be fitted separately to obtain vibration curves, temperature curves, and current curves. The amplitudes of the vibration curves, temperature curves, and current curves can be determined accordingly, so as to verify the fluctuation degree of the vibration time series data, temperature time series data, and current time series data based on the amplitudes.

[0083] In some embodiments, different predetermined amplitude thresholds can be set for different types of curves. The amplitude is compared with the predetermined amplitude threshold. If the amplitude is greater than the predetermined amplitude threshold, it can be determined that the time series data corresponding to the curve has failed the verification. Otherwise, it can be determined that the time series data has passed the verification, and then subsequent timestamp alignment processing can be performed.

[0084] According to embodiments of this disclosure, vibration time-series data, temperature time-series data, and current time-series data are fitted to obtain vibration curves, temperature curves, and current curves. Based on the amplitudes of the vibration curves, temperature curves, and current curves, the fluctuation levels of the vibration time-series data, temperature time-series data, and current time-series data are verified. This allows for the use of effective vibration time-series data, temperature time-series data, and current time-series data for life assessment, thereby improving the effectiveness of the life assessment process.

[0085] According to embodiments of this disclosure, feature extraction is performed on maintenance data and test environment data to obtain static features, and feature extraction is performed on time-series spliced ​​data obtained by splicing vibration time-series data, temperature time-series data and current time-series data to obtain time-series features. This includes: using the first extraction layer of the component life assessment model, using a multilayer perceptron to extract features from maintenance data and test environment data to obtain static features; and using the second extraction layer of the component life assessment model, using a gated loop unit to extract features from time-series operating state data to obtain time-series features.

[0086] In the embodiments of this disclosure, a component life assessment model can be used to assess the life of the component to be assessed. The component life assessment model can be trained using data from sample components.

[0087] According to embodiments of this disclosure, the component life assessment model may include a first extraction layer and a second extraction layer. The first extraction layer may be implemented based on a multilayer perceptron for extracting static features, and the second extraction layer may be implemented based on a gated recurrent unit for extracting temporal features.

[0088] When extracting features from maintenance data and test environment data using a multilayer perceptron, the maintenance data and test environment data sequentially pass through the input layer, hidden layer, and output layer of the multilayer perceptron to obtain the final static features. When extracting features from time-series operational state data using a gated recurrent unit, the hidden state of the last time step can be used as the time-series feature.

[0089] According to embodiments of this disclosure, by utilizing a multilayer perceptron suitable for extracting static features to extract features from maintenance data and test environment data, and by utilizing a gated loop unit suitable for extracting temporal features to extract features from temporal spliced ​​data, the accuracy of the feature extraction process is improved, thereby improving the accuracy of the life assessment process.

[0090] According to embodiments of this disclosure, based on a first fusion weight and a second fusion weight determined based on static features and temporal features, feature fusion is performed on static features and temporal features to obtain fused features, including: using the weight generation layer of the component life assessment model to determine the first fusion weight and the second fusion weight based on the splicing features of static features and temporal features; and using the feature fusion layer of the component life assessment model to perform feature fusion on static features and temporal features based on the first fusion weight and the second fusion weight to generate fused features.

[0091] In the embodiments of this disclosure, the component life assessment model further includes a weight generation layer and a feature fusion layer. The weight generation layer can be implemented based on an activation function such as the sigmoid function. Using the sigmoid function, the first fusion weight and the second fusion weight are determined based on the concatenated features. The concatenated features can be obtained by horizontally concatenating static features and temporal features. The process of determining the fusion weights can be shown in the following formula (2):

[0092] (2);

[0093] Where α represents the second fusion weight, h seq Characterizing temporal features, h static Characterizing static features, W α and b α The hyperparameters are characterized and determined during the training of the component life assessment model. The first fusion weight can be 1-α.

[0094] According to embodiments of this disclosure, the feature fusion layer can utilize a first fusion feature and a second fusion feature to perform feature fusion on static features and temporal features. The feature fusion process can be illustrated by the following formula (3):

[0095] (3);

[0096] Among them, h fused Characterize fusion features.

[0097] According to embodiments of this disclosure, a first fusion weight and a second fusion weight are determined based on the spliced ​​features obtained by horizontally splicing static features and temporal features. This enables the first fusion weight and the second fusion weight to accurately characterize the importance of the static features and temporal features respectively, thereby improving the accuracy of the fusion features and thus improving the accuracy of lifetime assessment.

[0098] Figure 4 The diagram illustrates the generation of initial lifetime assessment results according to a specific embodiment of the present disclosure.

[0099] like Figure 4As shown, the component life assessment model includes a first extraction layer 401, a second extraction layer 402, a weight generation layer 403, and a feature fusion layer 404.

[0100] like Figure 4 As shown, maintenance data 41 and test environment data 42 are input into the first extraction layer 401 to obtain static feature 11. The time-series operating status data is input into the second extraction layer 402 to obtain time-series feature 12. Static feature 11 and time-series feature 12 are input into the weight generation layer 403 to obtain the first fusion weight 21 and the second fusion weight 22. The first fusion weight 21, the second fusion weight 22, static feature 11 and time-series feature 12 are input into the feature fusion layer 404 to obtain fusion feature 30. Based on fusion feature 30, the initial life assessment result 50 is obtained.

[0101] According to embodiments of this disclosure, the test bench component life assessment method further includes: determining a remaining life interval based on a preset confidence level and remaining life; comparing the minimum value of the remaining life interval with a preset remaining life threshold to obtain a comparison result; and generating a warning message for the component to be assessed when the comparison result indicates that the minimum value of the remaining life interval is less than the preset remaining life threshold.

[0102] After obtaining the target life assessment results, warnings can be issued for the components to be assessed based on the remaining lifespan of the components included in the target life assessment results.

[0103] In the embodiments of this disclosure, since the remaining lifetime output by the component lifetime assessment model is not completely accurate, the remaining lifetime range can be determined based on a preset confidence level corresponding to the component lifetime assessment model. For example, the preset confidence level can be 95%, the remaining lifetime can be 200 hours, and the remaining lifetime range can be 185 hours to 215 hours.

[0104] After obtaining the remaining lifespan range, the minimum value of the remaining lifespan range can be compared with a preset remaining lifespan threshold. If the comparison result indicates that the minimum value of the remaining lifespan range is less than the preset remaining lifespan threshold, a warning message is generated for the component to be evaluated. For example, the preset remaining lifespan threshold can be 50 hours; if the minimum value is less than 50 hours, a warning message can be generated.

[0105] When generating early warning information, the remaining lifespan, maintenance data, test environment data, and time-series operating status data of the component to be evaluated can be added to the early warning information template to obtain early warning information for the component to be evaluated.

[0106] According to embodiments of this disclosure, by determining the remaining lifetime interval based on a preset confidence level and comparing the minimum value of the remaining lifetime interval with a preset remaining lifetime threshold, the accuracy of the timing of generating early warning information can be improved.

[0107] Figure 5 A flowchart illustrating a training method for a component life assessment model according to an embodiment of the present disclosure is shown.

[0108] like Figure 5 As shown, the training method for the component life assessment model includes operations S510 to S550.

[0109] During operation of S510, sample maintenance data, sample test environment data, and multiple sets of sample time-series operating status data of different modes of the sample components are acquired from multiple sensors of different types in the sample test bench.

[0110] When operating S520, the initial component life assessment model is used to extract features from the sample maintenance data and sample test environment data to obtain the sample static features. In addition, feature extraction is performed on the sample time series spliced ​​data obtained by splicing the sample vibration time series data, sample temperature time series data and sample current time series data to obtain the sample time series features.

[0111] When operating S530, the initial component life assessment model is used to perform feature fusion on the static features and temporal features of the samples based on the first sample fusion weight and the second sample fusion weight determined by the static features and temporal features of the samples, so as to obtain the sample fusion features.

[0112] During operation of S540, based on sample fusion features, the lifetime of the sample components is evaluated to obtain the sample lifetime evaluation results for the component to be evaluated.

[0113] When operating S550, the initial component life assessment model is trained based on the actual remaining life of the sample component, the physical degradation assessment results, and the sample life assessment results to obtain the component life assessment model. The physical degradation assessment results are obtained by assessing the life of the sample component based on the physical state parameters of the sample component.

[0114] In embodiments of this disclosure, an initial component life assessment model can be trained to obtain a component life assessment model. In some embodiments, data from retired sample components can be used to train the initial component life assessment model. Since the sample component has been retired, sample maintenance data, sample test environment data, and multiple sets of sample time-series operating status data corresponding to any moment in the sample component's life cycle can be selected. Based on this moment and the lifespan of the sample component, the actual remaining lifespan of the sample component is determined. The multiple sets of sample time-series operating status data include sample vibration time-series data, sample temperature time-series data, and sample current time-series data.

[0115] When training the initial component life assessment model using sample maintenance data, sample test environment data, and sample time-series operating status data, the processing procedure for these data is similar to that in the test bench component life assessment method, and will not be repeated here.

[0116] After obtaining the sample lifetime assessment results, the parameters of the initial component lifetime assessment model can be adjusted based on the loss between the sample lifetime assessment results and the actual remaining lifetime, and the loss between the sample lifetime assessment results and the physical degradation assessment results, to obtain the component lifetime assessment model.

[0117] According to embodiments of this disclosure, by training an initial component life assessment model based on physical degradation assessment results and actual remaining life, the trained component life assessment model is made to satisfy the physical degradation law of mechanical components, thereby improving the accuracy of life assessment.

[0118] According to embodiments of this disclosure, an initial component life assessment model is trained based on the actual remaining life of the sample component, the physical degradation assessment result, and the sample life assessment result to obtain a component life assessment model. This includes: determining a first loss value based on the actual remaining life and the sample life assessment result; determining a second loss value based on the physical degradation assessment result and the sample life assessment result; and adjusting the parameters of the initial component life assessment model based on the first loss value and the second loss value until a preset termination condition is met to obtain the component life assessment model.

[0119] In the embodiments of this disclosure, a first loss value and a second loss value can be determined using a loss function. The process of determining the first loss value can be shown in the following formula (4):

[0120] (4);

[0121] Among them, L data This represents the first loss value. This represents the sample life assessment result of the i-th sample component. Let N represent the actual remaining lifespan of the i-th sample component, and N be the total number of sample components.

[0122] The process of determining the second loss value can be shown in the following formula (5):

[0123] (5);

[0124] Among them, L physics This represents the second loss value. This represents the physical degradation assessment result of the i-th sample component.

[0125] According to embodiments of this disclosure, after determining the first loss value and the second loss value, the first loss value and the second loss value can be weighted and summed to obtain the total loss value, and then the parameters of the initial lifetime assessment model can be adjusted based on the total loss value. The process of determining the total loss value can be shown in the following formula (6):

[0126] L total = L data + L physics (6);

[0127] Among them, L total This represents the total loss value.

[0128] In some embodiments, an optimizer can be used to adjust the parameters of the initial lifetime assessment model and an early stopping mechanism can be set to terminate training if the validation set loss does not decrease for 5 consecutive rounds.

[0129] According to embodiments of this disclosure, by adjusting the parameters of the initial lifetime assessment model using a first loss value and a second loss value, the physical degradation model can be embedded as a regularization term into the loss function, constraining the data-driven lifetime assessment results. This enhances the interpretability and generalization ability of the model in small sample scenarios, thereby improving the accuracy of lifetime assessment.

[0130] According to embodiments of this disclosure, the physical degradation assessment result is determined as follows: based on the energy entropy of the vibration signal of the sample component, the current crack length of the sample component is determined by utilizing the mapping relationship between the energy entropy and the crack; based on the current crack length and a preset crack length threshold, the life of the sample component is assessed using the crack propagation equation to obtain the physical degradation assessment result.

[0131] In the embodiments of this disclosure, the crack propagation equation can be used to assess the lifespan of a sample component and obtain a physical degradation assessment result. Specifically, the current crack length can be determined based on the energy entropy of the vibration signal of the sample component, thereby obtaining the crack growth rate, and the physical degradation assessment result can be obtained by integrating the crack growth rate. The crack propagation equation can be represented by the following formula (7):

[0132] (7);

[0133] in, X represents the current crack length, and X represents the number of cycles. is the stress intensity factor, and C and M are material constants.

[0134] According to embodiments of this disclosure, determining the physical degradation assessment results of sample components based on the crack propagation equation can improve the accuracy of the physical degradation assessment results, thereby improving the interpretability and generalization ability of the life assessment model, and further improving the accuracy of the life assessment process.

[0135] Figure 6 A block diagram of a test bench component life assessment apparatus according to an embodiment of the present disclosure is shown schematically.

[0136] like Figure 6 As shown, the test bench component life assessment device 600 includes an acquisition module 610, an extraction module 620, a fusion module 630, an assessment module 660, a change determination module 650, and a life determination module 660.

[0137] The acquisition module 610 is used to acquire maintenance data, test environment data, and multiple sets of time-series operating status data of the component to be evaluated in different modes, collected by multiple sensors of different types. The multiple sets of time-series operating status data include vibration time-series data, temperature time-series data, and current time-series data. In some embodiments, the acquisition module 610 can be used to perform the operation S210 described above, which will not be repeated here.

[0138] The extraction module 620 is used to extract features from maintenance data and test environment data to obtain static features, and to extract features from time-series spliced ​​data obtained by splicing vibration time-series data, temperature time-series data and current time-series data to obtain time-series features. In some embodiments, the extraction module 620 can be used to perform the operation S220 described above, which will not be repeated here.

[0139] The fusion module 630 is used to perform feature fusion on the static features and the time-series features based on a first fusion weight and a second fusion weight determined by the static features and the time-series features, to obtain fused features. In some embodiments, the fusion module 630 can be used to perform the operation S230 described above, which will not be repeated here.

[0140] The evaluation module 660 is used to perform a lifetime assessment on the component to be evaluated based on the fusion features, and obtain an initial lifetime assessment result for the component to be evaluated. In some embodiments, the evaluation module 660 can be used to perform the operation S260 described above, which will not be repeated here.

[0141] The change determination module 650 is used to determine the life change rate based on the initial life assessment results and the historical life assessment results for the component to be assessed. In some embodiments, the change determination module 650 may be used to perform the operation S250 described above, which will not be repeated here.

[0142] The lifetime determination module 660 is used to determine the initial lifetime assessment result as the target lifetime assessment result when the lifetime change rate meets a predetermined lifetime change rate threshold. In some embodiments, the lifetime determination module 660 can be used to perform the operation S260 described above, which will not be repeated here.

[0143] According to embodiments of this disclosure, the test bench component life assessment device 600 further includes a recording acquisition module, an attenuation determination module, a fault coding module, and a maintenance determination module.

[0144] The record acquisition module is used to acquire maintenance records, which include the historical maintenance time and historical fault type for each component to be evaluated during maintenance.

[0145] The attenuation determination module is used to determine the maintenance attenuation factor based on multiple historical maintenance times and the current time.

[0146] The fault coding module is used to perform feature coding on multiple historical fault types by utilizing the mapping relationship between fault types and fault codes, thereby obtaining multiple historical fault codes.

[0147] The maintenance determination module is used to use maintenance attenuation factors and multiple historical fault codes as maintenance data.

[0148] According to embodiments of this disclosure, the multiple sensors each have different acquisition frequencies; the test bench component life assessment device 600 also includes an alignment module and a splicing module.

[0149] The alignment module is used to perform timestamp alignment processing on vibration time series data, temperature time series data, and current time series data to obtain target vibration time series data, target temperature time series data, and target current time series data.

[0150] The stitching module is used to stitch together the target vibration time series data, target temperature time series data, and target current time series data to obtain time series stitched data.

[0151] According to embodiments of this disclosure, the alignment module includes a fitting submodule, a verification submodule, and an alignment submodule.

[0152] The fitting submodule is used to fit vibration time series data, temperature time series data and current time series data respectively to obtain vibration curves, temperature curves and current curves.

[0153] The verification submodule is used to verify the fluctuation degree of vibration time series data, temperature time series data and current time series data based on the amplitude of vibration curve, temperature curve and current curve respectively, and obtain the verification result.

[0154] The alignment submodule is used to perform timestamp alignment on the vibration time series data, temperature time series data, and current time series data when the amplitude of the vibration time series data, temperature time series data, and current time series data represented by the verification result is less than a predetermined amplitude threshold, so as to obtain the target vibration time series data, target temperature time series data, and target current time series data.

[0155] According to embodiments of this disclosure, the fusion module 630 includes a weight determination submodule and a feature fusion submodule.

[0156] The weight determination submodule is used to determine the first fusion weight and the second fusion weight based on the splicing features of static features and temporal features by utilizing the weight generation layer of the component life assessment model.

[0157] The feature fusion submodule is used to generate fused features by fusing static features and time-series features based on the feature fusion layer of the component life assessment model, using the first fusion weight and the second fusion weight.

[0158] According to embodiments of this disclosure, the extraction module 620 includes a first extraction submodule and a second extraction submodule.

[0159] The first extraction submodule is used to extract features from maintenance data and test environment data using the first extraction layer of the component life assessment model and a multilayer perceptron to obtain static features.

[0160] The second extraction submodule is used to extract features from the time-series spliced ​​data using the second extraction layer of the component life assessment model and a gated loop unit to obtain time-series features.

[0161] According to embodiments of this disclosure, the life assessment result includes the remaining life of the component to be assessed; the test bench component life assessment device 600 further includes an interval determination module, a life comparison module, and an early warning generation module.

[0162] The interval determination module is used to determine the remaining lifetime interval based on preset confidence level and remaining lifetime.

[0163] The lifespan comparison module is used to compare the minimum value of the remaining lifespan range with the preset remaining lifespan threshold to obtain the comparison result.

[0164] The early warning generation module is used to generate early warning information for the component to be evaluated when the minimum value of the remaining life interval represented by the comparison results is less than the preset remaining life threshold.

[0165] Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure, or at least part of the functions of any one or more of them, can be implemented in one module. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be implemented by dividing them into multiple modules. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as hardware circuitry, such as a Field-Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a System-on-Chip, a System-on-a-Substrate, a System-on-Package, an Application-Specific Integrated Circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as computer program modules, which, when run, can perform corresponding functions.

[0166] For example, any and more of the acquisition module 610, extraction module 620, fusion module 630, evaluation module 660, change determination module 650, and lifetime determination module 660 can be combined into one module / unit / subunit, or any one of these modules / units / subunits can be split into multiple modules / units / subunits. Alternatively, at least some of the functionality of one or more of these modules / units / subunits can be combined with at least some of the functionality of other modules / units / subunits and implemented in one module / unit / subunit. According to embodiments of this disclosure, at least one of the acquisition module 610, extraction module 620, fusion module 630, evaluation module 660, change determination module 650, and lifetime determination module 660 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or any other reasonable method of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three methods. Alternatively, at least one of the acquisition module 610, extraction module 620, fusion module 630, evaluation module 660, change determination module 650, and lifetime determination module 660 can be at least partially implemented as a computer program module, which, when run, can perform the corresponding function.

[0167] It should be noted that the test bench component life assessment device part in the embodiments of this disclosure corresponds to the test bench component life assessment method part in the embodiments of this disclosure. For a detailed description of the test bench component life assessment device, please refer to the test bench component life assessment method part, which will not be repeated here.

[0168] Figure 7 A block diagram of an electronic device suitable for implementing the methods described above, according to embodiments of the present disclosure, is illustrated schematically. Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0169] like Figure 7 As shown, an electronic device 700 according to an embodiment of the present disclosure includes a processor 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage portion 708 into a random access memory (RAM) 703. The processor 701 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 701 may also include onboard memory for caching purposes. The processor 701 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0170] RAM 703 stores various programs and data required for the operation of electronic device 700. Processor 701, ROM 702, and RAM 703 are interconnected via bus 704. Processor 701 performs various operations of the method flow according to embodiments of the present disclosure by executing programs in ROM 702 and / or RAM 703. It should be noted that programs may also be stored in one or more memories other than ROM 702 and RAM 703. Processor 701 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in one or more memories.

[0171] According to embodiments of this disclosure, the electronic device 700 may further include an input / output (I / O) interface 705, which is also connected to a bus 704. The electronic device 700 may also include one or more of the following components connected to the input / output (I / O) interface 705: an input section 706 including a keyboard, mouse, etc.; an output section 707 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the input / output (I / O) interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed into the storage section 708 as needed.

[0172] According to embodiments of this disclosure, the method flow according to embodiments of this disclosure can be implemented as a computer software program. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable storage medium, the computer program containing program code for performing the methods shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via communication section 709, and / or installed from removable medium 711. When the computer program is executed by processor 701, it performs the functions defined in the system of embodiments of this disclosure. According to embodiments of this disclosure, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0173] This disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0174] According to embodiments of this disclosure, the computer-readable storage medium can be a non-volatile computer-readable storage medium. Examples include, but are not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0175] For example, according to embodiments of this disclosure, a computer-readable storage medium may include the ROM 702 and / or RAM 703 described above and / or one or more memories other than ROM 702 and RAM 703.

[0176] Embodiments of this disclosure also include a computer program product comprising a computer program containing program code for performing the methods provided in the embodiments of this disclosure. When the computer program product is run on an electronic device, the program code is used to enable the electronic device to implement the test bench life assessment method provided in the embodiments of this disclosure.

[0177] When the computer program is executed by the processor 701, it performs the functions defined in the system / apparatus of this disclosure embodiments. According to embodiments of this disclosure, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0178] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 709, and / or installed from a removable medium 711. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0179] According to embodiments of this disclosure, program code for executing the computer programs provided in embodiments of this disclosure can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can execute entirely on a user's computing device, partially on a user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0180] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions. Those skilled in the art will understand that the features described in the various embodiments of the present disclosure can be combined and / or combined in various ways, even if such combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments of this disclosure may be combined and / or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0181] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A test bench component life assessment method, comprising: obtaining maintenance data of a component to be assessed in a test bench, test environment data, and a plurality of sets of time-series operating state data of different modalities about the component to be assessed collected by a plurality of sensors of different types, the plurality of sets of time-series operating state data comprising vibration time-series data, temperature time-series data, and current time-series data; extracting features from the maintenance data and the test environment data to obtain static features, and extracting features from time-series splicing data obtained by splicing the vibration time-series data, the temperature time-series data, and the current time-series data to obtain time-series features; performing feature fusion on the static features and the time-series features based on first fusion weights and second fusion weights determined based on the static features and the time-series features to obtain fused features; performing life assessment on the component to be assessed based on the fused features to obtain an initial life assessment result for the component to be assessed; determining a life change rate based on the initial life assessment result and a historical life assessment result for the component to be assessed; determining the initial life assessment result as a target life assessment result in a case where the life change rate meets a predetermined life change rate threshold.

2. The method of claim 1, wherein, The method further comprises: obtaining maintenance records, the maintenance records comprising historical maintenance time instants and historical failure types of each maintenance of the component to be assessed; determining a maintenance attenuation factor based on a plurality of the historical maintenance time instants and a current time instant; encoding a plurality of the historical failure types based on a mapping relationship between failure types and failure codes to obtain a plurality of historical failure codes; using the maintenance attenuation factor and the plurality of the historical failure codes as the maintenance data.

3. The method of claim 1, wherein, The plurality of sensors each have different collection frequencies; the method further comprises: performing timestamp alignment processing on the vibration time-series data, the temperature time-series data, and the current time-series data to obtain target vibration time-series data, target temperature time-series data, and target current time-series data; splicing the target vibration time-series data, the target temperature time-series data, and the target current time-series data to obtain the time-series splicing data.

4. The method of claim 3, wherein, The timestamp alignment processing on the vibration time-series data, the temperature time-series data, and the current time-series data to obtain target vibration time-series data, target temperature time-series data, and target current time-series data comprises: performing fitting processing on the vibration time-series data, the temperature time-series data, and the current time-series data respectively to obtain vibration curves, temperature curves, and current curves; verifying fluctuation degrees of the vibration time-series data, the temperature time-series data, and the current time-series data based on amplitudes of the vibration curves, the temperature curves, and the current curves respectively to obtain verification results; in a case where the verification results represent that amplitudes of the vibration time-series data, the temperature time-series data, and the current time-series data are less than a predetermined amplitude threshold, performing timestamp alignment processing on the vibration time-series data, the temperature time-series data, and the current time-series data to obtain target vibration time-series data, target temperature time-series data, and target current time-series data.

5. The method of claim 1, wherein, The first fusion weight and the second fusion weight determined based on the static feature and the time sequence feature are used for feature fusion of the static feature and the time sequence feature to obtain a fusion feature, and the method comprises the following steps of: A weight generation layer of the component life assessment model determines the first fusion weight and the second fusion weight based on the spliced features of the static feature and the time sequence feature; A feature fusion layer of the component life assessment model generates the fusion feature by performing feature fusion on the static feature and the time sequence feature based on the first fusion weight and the second fusion weight.

6. The method of claim 1, wherein, The method further comprises the following steps of: A first extraction layer of the component life assessment model extracts the static feature from the maintenance data and the test environment data by using a multilayer perception machine; A second extraction layer of the component life assessment model extracts the time sequence feature from the time sequence spliced data by using a gated recurrent unit.

7. The method of claim 1, wherein, The target life assessment result comprises a remaining life of the component to be assessed, and the method further comprises the following steps of: Based on a preset reliability and the remaining life, a remaining life interval is determined; A minimum value of the remaining life interval is compared with a preset remaining life threshold to obtain a comparison result; In a case where the comparison result indicates that the minimum value of the remaining life interval is less than the preset remaining life threshold, warning information for the component to be assessed is generated.

8. A training method of a component life assessment model, comprising the following steps of: Obtaining sample maintenance data, sample test environment data and a plurality of groups of sample time sequence running state data of different modalities about a sample component collected by a plurality of sensors of different types in a sample test bench, wherein the plurality of groups of sample time sequence running state data comprise sample vibration time sequence data, sample temperature time sequence data and sample current time sequence data; Extracting a sample static feature from the sample maintenance data and the sample test environment data by using an initial component life assessment model, and extracting a sample time sequence feature from sample time sequence spliced data obtained by splicing the sample vibration time sequence data, the sample temperature time sequence data and the sample current time sequence data; Using the initial component life assessment model, a sample fusion feature is obtained by performing feature fusion on the sample static feature and the sample time sequence feature based on a first sample fusion weight and a second sample fusion weight determined based on the sample static feature and the sample time sequence feature; Based on the sample fusion feature, a sample life assessment result for the sample component is obtained by performing life assessment on the sample component; The initial component life assessment model is trained based on the actual remaining life of the sample component, a physical degradation assessment result, and the sample life assessment result, to obtain a component life assessment model, wherein the physical degradation assessment result is obtained by assessing the life of the sample component based on a physical state parameter of the sample component.

9. The training method of claim 8, wherein, The initial component life assessment model is trained based on the actual remaining life of the sample component, a physical degradation assessment result, and the sample life assessment result, to obtain a component life assessment model, wherein the physical degradation assessment result is obtained by assessing the life of the sample component based on a physical state parameter of the sample component. A first loss value is determined based on the actual remaining life and the sample life assessment result. A second loss value is determined based on the physical degradation assessment result and the sample life assessment result. The parameters of the initial component life assessment model are adjusted based on the first loss value and the second loss value until a preset end condition is met, to obtain the component life assessment model.

10. The training method of claim 8, wherein, The physical degradation assessment result is determined by: Based on the energy entropy of the vibration signal of the sample component, the current crack length of the sample component is determined by using the mapping relationship between the energy entropy and the crack. Based on the current crack length and a preset crack length threshold, the life of the sample component is assessed by using a crack propagation equation to obtain the physical degradation assessment result.

11. A test bench component life assessment device, comprising: an acquisition module configured to acquire maintenance data of a component to be assessed in a test bench, test environment data, and a plurality of groups of time-series operation state data about different modalities of the component to be assessed collected by a plurality of sensors of different types, wherein the plurality of groups of time-series operation state data include vibration time-series data, temperature time-series data, and current time-series data; an extraction module configured to extract features from the maintenance data and the test environment data to obtain static features, and extract features from time-series splicing data obtained by splicing the vibration time-series data, the temperature time-series data, and the current time-series data to obtain time-series features; a fusion module configured to perform feature fusion on the static features and the time-series features based on first fusion weights and second fusion weights determined based on the static features and the time-series features to obtain fusion features; an assessment module configured to assess the life of the component to be assessed based on the fusion features to obtain an initial life assessment result for the component to be assessed; a change determination module configured to determine a life change rate based on the initial life assessment result and a historical life assessment result for the component to be assessed; a life determination module configured to determine the initial life assessment result as a target life assessment result when the life change rate meets a predetermined life change rate threshold.

12. An electronic device, comprising: one or more processors; memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1-7.

13. A computer-readable storage medium having stored thereon executable instructions that, as a result of being executed by a processor, cause the processor to perform the method as claimed in any one of claims 1 to 7.

Images