A vacuum environment simulation equipment fault diagnosis system based on function test

Abstract

The application discloses a vacuum environment simulation equipment fault diagnosis system based on function test, comprising: a data acquisition module for acquiring function test original data; a data processing module for pre-processing to obtain function test response data; a stage division module for extracting and dividing to obtain a function stage response data set; a normal response construction module for screening normal test records, aligning stages, extracting normal response data to obtain a normal function response set; a residual difference extraction module for extracting differences to obtain a stage residual response difference set; a residual drive inversion module for inverting a Gaussian potential model based on a vacuum residual drive to obtain a residual driving force feature set; a fault diagnosis module for generating a fault diagnosis result; and a maintenance output module for generating maintenance processing information. The application adopts a residual drive inversion model to realize vacuum equipment fault attribution, and has the advantages of precision, interpretability and closed-loop maintenance.

Description

Technical Field

[0001] This invention relates to the field of testing and diagnostic technology for vacuum environment simulation equipment, and in particular to a fault diagnosis system for vacuum environment simulation equipment based on functional testing. Background Technology

[0002] Current solutions typically monitor the operation of vacuum environment simulation equipment by measuring pressure, temperature, current, voltage, valve feedback, and alarm records. These are then combined with human experience or fixed judgment rules to identify anomalies in the vacuum pump, temperature control circuit, valve structure, and interlocking protection process, thus enabling the detection of some obvious faults.

[0003] However, the above methods usually judge electrical test data, drive feedback data and functional response data separately, lacking the correlation processing of boundary identification, normal response alignment and residual response difference extraction by functional stage, making it difficult to further attribute abnormal air extraction, abnormal pressure holding, abnormal temperature control and abnormal valve response to specific fault locations and fault degrees.

[0004] Therefore, how to provide a fault diagnosis system for vacuum environment simulation equipment based on functional testing is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a fault diagnosis system for vacuum environment simulation equipment based on functional testing. This invention uses a residual drive inversion model to achieve fault attribution for vacuum equipment, and has the advantages of accuracy, interpretability, and closed-loop maintenance.

[0006] A fault diagnosis system for a vacuum environment simulation device based on functional testing according to an embodiment of the present invention includes: The data acquisition module is used to collect raw functional test data of the vacuum environment simulation equipment during the functional testing process; The data processing module is used to preprocess the raw functional test data to obtain functional test response data; The phase segmentation module is used to extract phase identification data from the functional test response data and segment the functional test response data to obtain a functional phase response data set. The normal response construction module is used to read the historical test records that have been completed and stored by the vacuum environment simulation equipment, filter the normal test records and align them according to the functional stage boundaries, and extract the normal response data corresponding to each functional stage to obtain the normal functional response set. The residual difference extraction module is used to extract the differences between the functional stage response data set and the normal functional response set to obtain the stage residual response difference set. The residual drive inversion module is used to input the set of stage residual response differences into the vacuum residual drive inversion Gaussian potential model, perform Gaussian process potential force inversion on the set of stage residual response differences, and obtain the set of residual driving force characteristics. The fault diagnosis module is used to generate fault diagnosis results based on each residual driving force feature in the residual driving force feature set; The maintenance output module is used to generate maintenance processing information based on the fault diagnosis results and send the fault diagnosis results and maintenance processing information to the operation and maintenance terminal of the vacuum environment simulation equipment.

[0007] Optionally, the data acquisition module includes: The functional testing process includes vacuuming, pressure holding, temperature control, valve switching, pressure recovery, and interlock protection. The raw data for the functional tests include electrical test data, drive feedback data, and functional response data; The drive feedback data is used to characterize the execution feedback status of valves, relays, and control commands.

[0008] Optionally, the data processing module includes: Preprocessing includes time alignment, signal identification unification, invalid sample point removal, missing sample point completion, and data format conversion to obtain functional test response data; The functional test response data includes standard electrical test data, standard drive feedback data, and standard functional response data.

[0009] Optionally, the stage division module includes: The drive feedback status and function response values ​​are read from the phase identification data according to the acquisition time to form a phase identification sequence; State transition identification is performed on the driving feedback states in the stage identification sequence. The acquisition time corresponding to the state transition is determined as the driving trigger point. The driving trigger points are arranged according to the acquisition time to obtain the driving trigger point sequence. The functional response values ​​in the stage identification sequence are calculated by the adjacent sampling difference. The acquisition time when the sign of the adjacent sampling difference changes or the adjacent sampling difference changes from non-zero to zero is determined as the response inflection point. The response inflection points are arranged according to the acquisition time to obtain the response inflection point sequence. Based on the order of acquisition time, each drive trigger point in the drive trigger point sequence is paired with a response inflection point that is adjacent to the acquisition time after that drive trigger point to obtain stage boundary candidate pairs. The candidate pair of stage boundaries with the smallest acquisition time difference is selected from the candidate pairs of stage boundaries as the stage boundary pair. The acquisition time of the driving trigger point in the stage boundary pair is used as the functional stage boundary. The functional stage boundaries are arranged according to the acquisition time to obtain the functional stage boundary sequence. The functional test response data is continuously segmented according to the functional phase boundary sequence to obtain the functional phase response data set.

[0010] Optionally, the normal response construction module includes: Read the equipment test conditions corresponding to the current functional test, and combine the equipment model, cabin volume, target pressure, target temperature and functional test procedure into the current test condition identifier; Read historical normal test records and their historical test condition identifiers from the historical test records that have been completed and stored by the vacuum environment simulation equipment. Match the historical test condition identifiers with the current test condition identifiers. Retain historical normal test records that are consistent in terms of equipment model, chamber volume, target pressure, target temperature and functional test procedures to obtain the matched normal test records. Determine the start and end points of each functional phase in the current functional test according to the functional phase boundaries, and combine the phase start and end points into a phase alignment benchmark. The normal response data in the matching normal test record is stage-aligned according to the stage alignment benchmark to obtain stage-aligned normal response data. Extract the normal response data corresponding to each functional stage from the stage alignment normal response data, and combine the normal response data corresponding to each functional stage into a normal functional response set.

[0011] Optionally, the residual difference extraction module includes: The functional stage response data in the functional stage response data set is matched with the normal response data in the normal functional response set according to the functional stage to obtain the stage matching data set. The functional stage response data and normal response data in the stage matching data set are sampled and matched according to the collection time sequence to obtain the stage matching data set; Read the stage matching data corresponding to each functional stage from the stage matching data set, extract the functional stage pressure response and functional stage temperature response from the functional stage response data, extract the normal pressure response and normal temperature response from the normal response data, and perform sampling point difference calculation to obtain the pressure residual response sequence and temperature residual response sequence. The residual direction, continuous residual segment statistics, and residual amplitude accumulation are performed on the pressure residual response sequence to obtain the pressure residual direction, pressure residual duration, and pressure residual accumulation. The residual direction, continuous residual segment statistics, and residual amplitude accumulation are performed on the temperature residual response sequence to obtain the temperature residual direction, temperature residual duration, and temperature residual accumulation. The residual pressure direction, residual pressure duration, residual pressure accumulation, residual temperature direction, residual temperature duration, and residual temperature accumulation are combined according to the corresponding functional stages to obtain the stage residual response item set. According to the functional phase sequence corresponding to the functional testing process, the set of residual response items for each phase is collected to obtain the set of residual response differences for each phase.

[0012] Optionally, the residual drive inversion module includes: The set of stage residual response differences is input into the vacuum residual drive inversion Gaussian potential model, which includes a residual input construction layer, a stage Gaussian kernel construction layer, a potential force inversion layer, a residual driving force feature extraction layer, and a residual driving force convergence output layer. In the residual input construction layer, the residual response difference set of the stage is sorted according to the functional stage order to obtain the residual input sequence carrying the functional stage identifier, and the residual input sequence is sent to the stage Gaussian kernel construction layer and the potential force inversion layer. In the stage Gaussian kernel construction layer, based on the functional stages and residual response differences corresponding to the residual input sequence, stage Gaussian kernels corresponding to each functional stage are constructed, and the stage Gaussian kernels corresponding to each functional stage are combined into a stage Gaussian kernel set. In the latent force inversion layer, the Gaussian kernel set of the stages is called to generate the covariance matrix corresponding to each functional stage, and the Gaussian process latent force inversion is performed on the residual response difference in the residual input sequence based on the covariance matrix to obtain the residual driving force sequence. In the residual driving force feature extraction layer, the residual driving force sequence is marked with the occurrence stage, the driving force direction is identified, the driving force duration is statistically analyzed, and the cumulative driving force intensity is extracted to obtain the residual driving force stage feature set. In the residual driving force convergence output layer, the residual driving force stage feature set is aggregated according to the functional stage to obtain the residual driving force feature set.

[0013] Optionally, the fault diagnosis module includes: The residual driving force features of air extraction, pressure holding, temperature control, valve hysteresis, pressure recovery, and interlock protection are read from the residual driving force feature set. The occurrence stage, change direction, and duration of each residual driving force feature are extracted to obtain residual driving force discrimination data. Based on the stage of occurrence, direction of change, and duration, the residual driving force characteristics in the residual driving force discrimination data are identified by cross-stage association, same-stage association, and duration overlap to obtain the combination relationship of residual driving forces. From the residual driving force combination relationship, extract the cross-stage combination of the residual driving force characteristics of air extraction and the residual driving force characteristics of pressure holding with the same change direction and overlapping duration to obtain the sealing leakage fault item; The pump set performance degradation fault item is obtained by extracting cross-stage combinations of residual driving force characteristics that have a duration but do not overlap with the duration of residual driving force characteristics of pressure holding. Extract the temperature control residual driving force characteristics of the same stage combination of the duration of existence to obtain the temperature control execution fault item; The valve execution failure item is obtained by extracting the valve hysteresis residual driving force characteristics and the pressure recovery residual driving force characteristics with a cross-stage combination of duration overlap. Extracting the interlocking protection residual driving force characteristics and the same-stage combination of any residual driving force characteristics with a duration overlap length yields the interlocking protection associated fault item; The fault diagnosis results are obtained by categorizing the fault items related to sealing leakage, pump performance degradation, temperature control, valve operation, and interlock protection.

[0014] Optionally, the maintenance output module includes: Read the fault type, fault location, and fault severity from the fault diagnosis results, and combine the fault type, fault location, and fault severity into fault description information; Read the fault maintenance correspondence stored in the system; Match the fault description information with the corresponding fault maintenance to obtain maintenance processing matching items; The maintenance object, maintenance action, and processing order in the maintenance processing match are combined into maintenance processing information; The fault diagnosis results and maintenance information are sent to the operation and maintenance terminal of the vacuum environment simulation equipment.

[0015] The beneficial effects of this invention are: This application collects raw functional test data from a vacuum environment simulation device during functional testing and processes electrical test data, drive feedback data, and functional response data into unified functional test response data. This allows for correlated analysis of the electrical operating status of the vacuum pump and temperature control circuit, the execution feedback status of valves and relays, pressure change status, and temperature change status on the same time reference. Functional stage boundaries are determined through stage identification data, and functional stage response data sets are obtained by segmenting according to these boundaries. This enables separate identification of response changes during the vacuuming process, pressure holding process, temperature control process, valve switching process, pressure recovery process, and interlock protection process, avoiding the mixing of data from different functional stages for judgment, thereby improving the specificity of fault diagnosis and the interpretability of the process.

[0016] This application further uses equipment testing conditions as a matching basis, selecting normal test records that match the current functional test from historical normal test records, and obtaining a normal functional response set through stage alignment. This allows the response data of the current functional test to be compared with normal response data under the same equipment model, chamber volume, target pressure, target temperature, and functional test procedure. The resulting stage residual response difference set can isolate normal differences caused by equipment specifications, test targets, and test procedures, highlighting the pressure residual response, temperature residual response, and their direction, duration, and accumulation that are truly related to the fault, thus improving the accuracy of identifying abnormal pumping, pressure holding, temperature control, and valve response.

[0017] Simultaneously, this application inputs the set of stage residual response differences into the vacuum residual drive inversion Gaussian potential model. Through stage Gaussian kernel construction, potential force inversion, and residual driving force feature extraction, it obtains residual driving force features for pumping, pressure holding, temperature control, valve hysteresis, pressure recovery, and interlocking protection. Fault diagnosis results are generated by analyzing the occurrence stage, direction of change, duration, and combination relationships of each residual driving force feature. This allows for the differentiation of sealing leakage faults, pump performance degradation faults, temperature control execution faults, valve execution faults, and interlocking protection-related faults. The fault location and severity are simultaneously determined, and maintenance information is then generated and sent to the operation and maintenance terminal, achieving a closed loop from functional testing and fault attribution to maintenance. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of a fault diagnosis system for a vacuum environment simulation device based on functional testing, as proposed in this invention. Figure 2 This is a flowchart illustrating the generation of a normal functional response set in a fault diagnosis system for a vacuum environment simulation device based on functional testing, as proposed in this invention. Figure 3 This is a flowchart illustrating the generation of fault diagnosis results based on the residual driving force combination relationship in a fault diagnosis system for a vacuum environment simulation device based on functional testing, as proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3A fault diagnosis system for a vacuum environment simulation device based on functional testing, comprising: The data acquisition module is used to acquire raw functional test data of the vacuum environment simulation equipment during the functional testing process. The raw functional test data includes electrical test data, drive feedback data and functional response data. The electrical test data is used to characterize the electrical operating status of the vacuum pump and temperature control circuit. The drive feedback data is used to characterize the execution feedback status of valves, relays and control commands. The functional response data is used to characterize the pressure change status and temperature change status. The data processing module is used to preprocess the raw functional test data to obtain functional test response data, which includes standard electrical test data, standard drive feedback data, and standard functional response data. The phase segmentation module is used to extract phase identification data from functional test response data. The phase identification data includes standard driver feedback data and standard functional response data. Based on the phase identification data, the functional phase boundaries are determined, and the functional test response data is segmented according to the functional phase boundaries to obtain a functional phase response data set. The normal response construction module is used to read the historical test records that have been completed and stored by the vacuum environment simulation equipment, and to filter the normal test records that match the current functional test from the historical normal test records based on the equipment test conditions. The normal test records are then aligned with the functional stage boundaries, and the normal response data corresponding to each functional stage is extracted to obtain the normal functional response set. The equipment test conditions include the equipment model, chamber volume, target pressure, target temperature and functional test process. The residual difference extraction module is used to extract the differences between the functional stage response data set and the data with the same functional stage in the normal functional response set to obtain the stage residual response difference set. The stage residual response difference set includes vacuuming residual response difference, pressure holding residual response difference, temperature control residual response difference, valve switching residual response difference, pressure recovery residual response difference, and interlock protection residual response difference. The residual drive inversion module is used to input the set of stage residual response differences into the vacuum residual drive inversion Gaussian potential model, and perform Gaussian process potential inversion on the set of stage residual response differences according to the functional stage to obtain the set of residual driving force characteristics. The set of residual driving force characteristics includes residual driving force characteristics of pumping, residual driving force characteristics of pressure holding, residual driving force characteristics of temperature control, residual driving force characteristics of valve hysteresis, residual driving force characteristics of pressure recovery, and residual driving force characteristics of interlock protection. The fault diagnosis module is used to generate fault diagnosis results based on the occurrence stage, change direction, duration and combination relationship of each residual driving force feature in the residual driving force feature set. The fault diagnosis results include fault type, fault location and fault degree. The maintenance output module is used to generate maintenance processing information based on the fault diagnosis results and send the fault diagnosis results and maintenance processing information to the operation and maintenance terminal of the vacuum environment simulation equipment.

[0021] In this embodiment, the data acquisition module includes: The functional testing process includes vacuuming, pressure holding, temperature control, valve switching, pressure recovery, and interlock protection. The raw data for the functional tests include electrical test data, drive feedback data, and functional response data. The electrical test data is used to characterize the electrical operating status of the vacuum pump and the temperature control circuit. The drive feedback data is used to characterize the execution feedback status of valves, relays, and control commands, while the functional response data is used to characterize pressure change status and temperature change status.

[0022] In this embodiment, the data processing module includes: Preprocessing includes time alignment, signal identification unification, invalid sample point removal, missing sample point completion, and data format conversion to obtain functional test response data; The functional test response data includes standard electrical test data, standard drive feedback data, and standard functional response data.

[0023] In this embodiment, the stage division module includes: The drive feedback status and functional response value are read from the stage identification data according to the acquisition time to form a stage identification sequence. The drive feedback status includes control command status, valve action status and relay trigger status. The functional response value includes pressure response value and temperature response value. State transition identification is performed on the driving feedback states in the stage identification sequence. The acquisition time corresponding to the state transition is determined as the driving trigger point. The driving trigger points are arranged according to the acquisition time to obtain the driving trigger point sequence. The functional response values ​​in the stage identification sequence are calculated by the adjacent sampling difference. The acquisition time when the sign of the adjacent sampling difference changes or the adjacent sampling difference changes from non-zero to zero is determined as the response inflection point. The response inflection points are arranged according to the acquisition time to obtain the response inflection point sequence. According to the order of acquisition time, each drive trigger point in the drive trigger point sequence is paired with the response inflection point that is adjacent to the acquisition time after the drive trigger point to obtain a stage boundary candidate pair. The stage boundary candidate pair includes the drive trigger point, the response inflection point and the acquisition time difference between the two. The candidate pair of stage boundaries with the smallest acquisition time difference is selected from the candidate pairs of stage boundaries as the stage boundary pair. The acquisition time of the driving trigger point in the stage boundary pair is used as the functional stage boundary. The functional stage boundaries are arranged according to the acquisition time to obtain the functional stage boundary sequence. Specifically, this includes: for each drive trigger point, firstly screening response inflection points whose acquisition time is later than that drive trigger point, then calculating the acquisition time difference between the drive trigger point and the response inflection point to obtain a candidate time difference set, and determining the stage boundary candidate pair corresponding to the acquisition time difference with the smallest value in the candidate time difference set as the stage boundary pair. If the same drive trigger point corresponds to multiple stage boundary candidate pairs with the same acquisition time difference, then retain the stage boundary candidate pair that appears for the first time in the acquisition time sequence of the response inflection point. The stage boundary pairs obtained through this process are used to generate the subsequent functional stage boundary sequence. The functional test response data is continuously segmented according to the functional phase boundary sequence to obtain the functional phase response data set.

[0024] In this embodiment, the normal response construction module includes: Read the equipment test conditions corresponding to the current functional test, and combine the equipment model, cabin volume, target pressure, target temperature and functional test procedure into the current test condition identifier; Read historical normal test records and their historical test condition identifiers from the historical test records that have been completed and stored by the vacuum environment simulation equipment. Match the historical test condition identifiers with the current test condition identifiers. Retain historical normal test records that are consistent in terms of equipment model, chamber volume, target pressure, target temperature and functional test procedures to obtain the matched normal test records. Determine the start and end points of each functional phase in the current functional test according to the functional phase boundaries, and combine the phase start and end points into a phase alignment benchmark. The normal response data in the matching normal test record is stage-aligned according to the stage alignment benchmark to obtain stage-aligned normal response data. When performing stage alignment on the normal response data in the matching normal test record, the stage start point in the stage alignment benchmark is first used as the alignment start position, and the stage end point is used as the alignment end position. Normal response segments of the corresponding functional stage are extracted. The extracted normal response segments are resampled according to the number of sampling points of the same functional stage in the current functional test to obtain stage-aligned normal response data with the same number of sampling points. During resampling, the corresponding sampling position is determined according to the acquisition time ratio, and adjacent two sampling values ​​are linearly interpolated according to the time ratio. The stage-aligned normal response data is used to extract the normal response data corresponding to each functional stage in the subsequent process. Extract the normal response data corresponding to each functional stage from the stage alignment normal response data, and combine the normal response data corresponding to each functional stage into a normal functional response set.

[0025] In this embodiment, the residual difference extraction module includes: The functional stage response data in the functional stage response data set is matched with the normal response data in the normal functional response set according to the functional stage to obtain the stage matching data set. The functional stage response data and normal response data in the stage matching data set are sampled and matched according to the collection time sequence to obtain the stage matching data set; Read the stage matching data corresponding to each functional stage from the stage matching data set. The stage matching data includes functional stage response data and normal response data. Extract the functional stage pressure response and functional stage temperature response from the functional stage response data. Extract the normal pressure response and normal temperature response from the normal response data. Perform sample point difference calculation to obtain the pressure residual response sequence and temperature residual response sequence. Specifically, after reading the stage matching data corresponding to each functional stage from the stage matching data set, the functional stage response data and normal response data are first distinguished according to the functional stage identifier. The pressure sampling values ​​and temperature sampling values ​​in the functional stage response data are arranged according to the acquisition time to generate the functional stage pressure response and functional stage temperature response. The pressure sampling values ​​and temperature sampling values ​​in the normal response data are arranged according to the acquisition time to generate the normal pressure response and normal temperature response. According to the same sampling sequence, the pressure sampling values ​​in the functional stage pressure response are subtracted one by one from the pressure sampling values ​​in the normal pressure response to obtain the pressure residual response sequence. The temperature sampling values ​​in the functional stage temperature response are subtracted one by one from the temperature sampling values ​​in the normal temperature response to obtain the temperature residual response sequence. The residual direction, continuous residual segment statistics, and residual amplitude accumulation are performed on the pressure residual response sequence to obtain the pressure residual direction, pressure residual duration, and pressure residual accumulation. The residual direction, continuous residual segment statistics, and residual amplitude accumulation are performed on the temperature residual response sequence to obtain the temperature residual direction, temperature residual duration, and temperature residual accumulation. When identifying residual direction, the number of positive and negative sampling points in the pressure residual response sequence is counted. If the number of positive sampling points is greater than the number of negative sampling points, the pressure residual direction is determined to be positive; if the number of negative sampling points is greater than the number of positive sampling points, the pressure residual direction is determined to be negative; if the two numbers are equal, the pressure residual direction is determined to be bidirectional. When counting continuous residual segments, adjacent sampling points with non-zero residual response values ​​are grouped into the same continuous residual segment, and the sampling time length corresponding to the continuous residual segment is taken as the pressure residual duration. When accumulating residual amplitude, the absolute values ​​of each pressure residual response value within the continuous residual segment are summed to obtain the pressure residual accumulation. The temperature residual direction, temperature residual duration, and temperature residual accumulation are obtained using the same processing method. The residual pressure direction, residual pressure duration, residual pressure accumulation, residual temperature direction, residual temperature duration, and residual temperature accumulation are combined according to the corresponding functional stages to obtain the stage residual response item set. According to the functional phase sequence corresponding to the functional testing process, the set of residual response items for each phase is collected to obtain the set of residual response differences for each phase.

[0026] In this embodiment, the residual drive inversion module includes: The set of stage residual response differences is input into the vacuum residual drive inversion Gaussian potential model, which includes a residual input construction layer, a stage Gaussian kernel construction layer, a potential force inversion layer, a residual driving force feature extraction layer, and a residual driving force convergence output layer. In the residual input construction layer, the residual response difference set of the stage is sorted according to the functional stage order to obtain the residual input sequence carrying the functional stage identifier, and the residual input sequence is sent to the stage Gaussian kernel construction layer and the potential force inversion layer. In the stage Gaussian kernel construction layer, based on the functional stages and residual response differences corresponding to the residual input sequence, stage Gaussian kernels corresponding to each functional stage are constructed, and the stage Gaussian kernels corresponding to each functional stage are combined into a stage Gaussian kernel set. Specifically, in the stage Gaussian kernel construction layer, the residual input sequence is first divided into vacuum residual input segment, pressure holding residual input segment, temperature control residual input segment, valve switching residual input segment, pressure recovery residual input segment, and interlock protection residual input segment according to the functional stage corresponding to the residual input sequence. For each residual input segment, the acquisition time interval, the distribution of the direction of change of residual response difference, and the amplitude distribution of residual response difference are statistically analyzed to form the kernel construction parameters corresponding to the functional stage. Based on the kernel construction parameters, the stage Gaussian kernel corresponding to the functional stage is constructed so that the stage Gaussian kernel simultaneously represents the proximity of acquisition time and the proximity of residual response difference. The stage Gaussian kernels corresponding to each functional stage are combined according to the functional stage order to obtain the stage Gaussian kernel set. In the latent force inversion layer, the Gaussian kernel set of the stages is called to generate the covariance matrix corresponding to each functional stage, and the Gaussian process latent force inversion is performed on the residual response difference in the residual input sequence based on the covariance matrix to obtain the residual driving force sequence. Specifically, after calling the stage Gaussian kernel set, any two sampling points belonging to the same functional stage in the residual input sequence are first input into the corresponding stage Gaussian kernel to obtain the correlation between the two sampling points. The correlation between each sampling point is arranged according to the acquisition time order to generate the covariance matrix corresponding to the functional stage. The covariance matrix and the residual response difference corresponding to the functional stage are input into the Gaussian process latent force inversion process to obtain the residual driving force value corresponding to each sampling point in the functional stage. The residual driving force value corresponding to each functional stage is arranged according to the functional stage order to obtain the residual driving force sequence. In the residual driving force feature extraction layer, the residual driving force sequence is marked with the occurrence stage, the driving force direction is identified, the driving force duration is statistically analyzed, and the cumulative driving force intensity is extracted to obtain the residual driving force stage feature set. When marking the occurrence stage of the residual driving force sequence, the functional stage corresponding to each sampling point in the residual driving force sequence is written into the stage label of that sampling point to obtain the residual driving force sequence with stage label. The driving force direction is identified by comparing the number of positive sampling points and the number of negative sampling points in the residual driving force sequence. The driving force duration is determined by the acquisition time length corresponding to the continuous non-zero residual driving force sampling segment. The cumulative driving force intensity is obtained by summing the absolute values ​​of each residual driving force value in the continuous non-zero residual driving force sampling segment, and together with the occurrence stage, driving force direction and driving force duration, they form the residual driving force stage feature set. In the residual driving force convergence output layer, the residual driving force stage feature set is collected according to the functional stage to obtain the residual driving force feature set. The residual driving force feature set includes the residual driving force feature of air extraction, the residual driving force feature of pressure holding, the residual driving force feature of temperature control, the residual driving force feature of valve hysteresis, the residual driving force feature of pressure recovery, and the residual driving force feature of interlock protection. The vacuum residual drive inversion Gaussian potential model is established by offline training followed by online invocation. The model consists of a residual input construction layer, a stage Gaussian kernel construction layer, a latent force inversion layer, a residual driving force feature extraction layer, and a residual driving force convergence output layer connected in sequence. The output of the residual input construction layer is input to the stage Gaussian kernel construction layer on one hand and to the latent force inversion layer on the other hand. The output of the stage Gaussian kernel construction layer is then input to the latent force inversion layer. The latent force inversion layer, the residual driving force feature extraction layer, and the residual driving force convergence output layer are connected in series. The training data comes from the historical test records completed and stored by the vacuum environment simulation equipment. The training samples adopt a two-dimensional time series data format. Each row corresponds to a sampling time, and each column corresponds to the pressure residual response sequence, temperature residual response sequence, pressure residual direction, pressure residual duration, pressure residual accumulation, temperature residual direction, temperature residual duration, temperature residual accumulation, and functional stage identifier. The annotation results of the training samples come from historical normal test records, equipment maintenance records and fault review records. The annotation content includes the characteristics of residual driving force for air extraction, residual driving force for pressure holding, residual driving force for temperature control, residual driving force for valve hysteresis, residual driving force for pressure recovery, residual driving force for interlock protection, as well as the corresponding fault type, fault location and fault degree. During training, the set of residual response differences between stages is used as input, and the set of labeled residual driving force features is used as the supervision target. The loss function is composed of residual driving force inversion error, functional stage consistency error, and fault diagnosis result matching error. The residual driving force inversion error is used to constrain the output residual driving force features to be consistent with the labeled results. The functional stage consistency error is used to constrain each residual driving force feature to be consistent with its functional stage. The fault diagnosis result matching error is used to constrain the fault type, fault location, and fault degree obtained from the set of residual driving force features to be consistent with the labeled results. The training parameters include the number of training rounds, the number of batch samples, the learning rate, the initial values ​​of the stage Gaussian kernel parameters and the initial values ​​of the noise term. During the training process, the stage Gaussian kernel parameters, the potential force inversion parameters and the residual driving force feature extraction parameters are updated sequentially according to the batch samples. When the total loss decreases less than the set convergence range for multiple consecutive training rounds, and the matching rate of fault type, fault location and fault degree in the verification samples no longer improves, the vacuum residual driving inversion Gaussian potential model is determined to have reached the convergence condition and training is stopped. The general Gaussian process potential force model is transformed into a residual driving force inversion model for functional testing of vacuum environment simulation equipment. Specifically, instead of directly judging faults based on pressure, temperature, and electrical data, the residual response difference set of the stage is first extracted by comparing the functional stage response data with the normal functional response set, and then the residual response difference set of the stage is input into the vacuum residual driving inversion Gaussian potential model. The core of this model is to construct stage Gaussian kernels according to the functional stages of vacuuming, pressure holding, temperature control, valve switching, pressure recovery, and interlock protection. This makes the Gaussian process potential force inversion no longer a general inversion, but rather an inversion of the corresponding residual driving force characteristics based on the differences in residual response at different functional stages. Furthermore, by combining the residual driving force characteristics of vacuuming, pressure holding, temperature control, valve hysteresis, pressure recovery, and interlock protection, it can distinguish between sealing leakage, pump performance degradation, temperature control execution failure, valve execution failure, and interlock protection associated failure. The training uses historical test records, historical normal test records, equipment maintenance records and fault review records to form training samples and labeled results. The model is constrained by residual driving force inversion error, functional stage consistency error and fault diagnosis result matching error, so that the residual driving force characteristics output by the model can not only conform to the functional stage, but also correspond to the fault type, fault location and fault degree.

[0027] In this embodiment, the fault diagnosis module includes: The residual driving force features of air extraction, pressure holding, temperature control, valve hysteresis, pressure recovery, and interlock protection are read from the residual driving force feature set. The occurrence stage, change direction, and duration of each residual driving force feature are extracted to obtain residual driving force discrimination data. Based on the stage of occurrence, direction of change, and duration, the residual driving force features in the residual driving force discrimination data are identified through cross-stage association, same-stage association, and duration overlap to obtain the residual driving force combination relationship. The residual driving force combination relationship includes cross-stage combination, same-stage combination, and duration overlap length. When performing cross-stage association on residual driving force discrimination data, two residual driving force features that appear in different stages are paired, and the order of their appearance stages, the consistency of their change direction, and the overlap of their durations are recorded to form a cross-stage combination. When performing same-stage association on residual driving force discrimination data, two residual driving force features that appear in the same stage are paired, and the overlap of their durations within the same appearance stage is recorded to form a same-stage combination. When identifying duration overlap, the start and end times of two residual driving force features are compared, and the time interval in which they coexist is taken as the duration overlap length. If there is no time interval in which they coexist, the duration overlap length is recorded as zero. The cross-stage combination, the same-stage combination, and the duration overlap length are combined to form a residual driving force combination relationship. Extracting the cross-stage combination of the residual driving force characteristics of pumping and the residual driving force characteristics of pressure holding with the same direction of change and the duration of overlap from the residual driving force combination relationship, the sealing leakage fault item is obtained. The sealing leakage fault item includes the sealing leakage fault type, the fault location of the vacuum chamber sealing area, and the degree of sealing leakage fault determined by the duration of overlap. When a sealing leakage fault item is generated, the residual driving force characteristics of the pumping and the residual driving force characteristics of the pressure holding are read from the cross-stage combination, and the direction of change of the two is compared. If the direction of change of the two is consistent and the duration overlap length is positive, the cross-stage combination is written into the sealing leakage fault item. The degree of sealing leakage fault is obtained by dividing the duration overlap length by the shorter duration of the pumping residual driving force characteristic and the pressure holding residual driving force characteristic. Extracting the pumping residual driving force feature from the residual driving force combination relationship, and finding cross-stage combinations where the pumping residual driving force feature has a duration that does not overlap with the pressure holding residual driving force feature, yields the pump set performance degradation fault item. The pump set performance degradation fault item includes the pump set performance degradation fault type, the vacuum pump set fault location, and the degree of pump set performance degradation fault determined by the duration of the pumping residual driving force feature. When a pump set performance degradation fault item is generated, the residual driving force characteristic of the pumping is read from the cross-stage combination, and it is determined whether the residual driving force characteristic of the pumping has a duration. If the residual driving force characteristic of the pumping has a duration and the residual driving force characteristic of the pumping does not overlap with the duration of the pressure holding residual driving force characteristic, then the cross-stage combination is written into the pump set performance degradation fault item. The degree of pump set performance degradation fault is obtained by dividing the duration of the residual driving force characteristic of the pumping by the duration of the stage in which the residual driving force characteristic of the pumping is located. Extracting the same-stage combination of the duration of the residual driving force characteristics of temperature control from the residual driving force combination relationship yields temperature control execution fault items. Temperature control execution fault items include temperature control execution fault type, temperature control circuit fault location, and temperature control execution fault degree determined by the duration of the residual driving force characteristics of temperature control. When generating a temperature control execution fault item, the residual driving force feature of temperature control is read from the same stage combination, and it is determined whether the residual driving force feature of temperature control has a duration during the temperature control process. If the residual driving force feature of temperature control has a duration, the same stage combination is written into the temperature control execution fault item. The degree of temperature control execution fault is obtained by dividing the duration of the residual driving force feature of temperature control by the duration of the temperature control process. The valve execution failure item is obtained by extracting the valve hysteresis residual driving force characteristics and pressure recovery residual driving force characteristics with a duration overlap length from the residual driving force combination relationship. The valve execution failure item includes the valve execution failure type, the valve drive structure failure location and the valve execution failure degree determined by the duration overlap length. When generating a valve execution failure item, the valve hysteresis residual driving force characteristics and pressure recovery residual driving force characteristics are read from the cross-stage combination, and the duration overlap length of the two is calculated. When the duration overlap length is positive, the cross-stage combination is written into the valve execution failure item. The degree of valve execution failure is obtained by dividing the duration overlap length by the duration of the valve hysteresis residual driving force characteristics. Extracting the interlocking protection residual driving force feature from the residual driving force combination relationship and the same stage combination with any residual driving force feature having a duration overlap length, the interlocking protection associated fault item is obtained. The interlocking protection associated fault item includes the interlocking protection associated fault type, the interlocking protection control circuit fault location, and the interlocking protection associated fault degree determined by the duration overlap length. When generating interlocking protection associated fault items, the residual driving force characteristics of interlocking protection are combined with the residual driving force characteristics of air extraction, pressure holding, temperature control, valve hysteresis, and pressure recovery at the same stage for identification. If any combination at the same stage has a duration overlap length, the combination at the same stage is written into the interlocking protection associated fault item. The degree of interlocking protection associated fault is obtained by dividing the duration overlap length by the duration of the interlocking protection residual driving force characteristics. The fault diagnosis results are obtained by categorizing the fault items related to sealing leakage, pump performance degradation, temperature control, valve operation, and interlock protection.

[0028] In this embodiment, the maintenance output module includes: Read the fault type, fault location, and fault severity from the fault diagnosis results, and combine the fault type, fault location, and fault severity into fault description information; Read the fault maintenance correspondence stored in the system. The fault maintenance correspondence includes the correspondence between fault type, fault location, fault severity, maintenance object, maintenance action and processing order. The fault description information is matched with the fault maintenance correspondence to obtain maintenance processing matching items. The maintenance processing matching items include the maintenance object, maintenance action and processing order corresponding to the fault description information. The maintenance object, maintenance action, and processing order in the maintenance processing match are combined into maintenance processing information; The fault diagnosis results and maintenance information are sent to the operation and maintenance terminal of the vacuum environment simulation equipment.

[0029] Example 1: To verify the feasibility of this invention in practice, it was applied to the operation and maintenance of vacuum environment simulation equipment at an environmental testing service organization in East China. This organization has long been responsible for vacuum environment adaptability testing of electronic components, sealing connectors, precision sensors, and small electromechanical components. The equipment routinely performs vacuuming, pressure holding, temperature control, valve switching, pressure recovery, and interlock protection processes. Due to frequent equipment use, the vacuum pump load status, temperature control circuit response status, valve action status, and pressure sensor output status change with the usage cycle. Previously, maintenance personnel mainly relied on alarm records, pressure curves, temperature curves, and manual inspections to determine anomalies. When the vacuuming response slowed down or the pressure holding curve rebounded, it was usually necessary to check the vacuum pump, door sealing structure, pipe joints, valve actuators, and pressure sensors in sequence. The troubleshooting path was long, and the curves of different faults were similar, making it easy to find anomalies but difficult to pinpoint the source.

[0030] In this scenario, the present invention is deployed between the test control system and the maintenance terminal of a vacuum environment simulation device. After the device performs functional tests, the system synchronously collects electrical test data, drive feedback data, and functional response data. It then performs time alignment, signal identification unification, invalid sampling point removal, missing sampling point completion, and data format conversion on the raw functional test data to obtain functional test response data. This allows for correlation analysis of vacuum pump electrical changes, valve action feedback, pressure changes, and temperature changes on the same time reference. Subsequently, the system extracts standard drive feedback data and standard functional response data from the functional test response data, identifies state transitions in control commands, valve actions, and relay triggers, and, combined with response inflection points in pressure and temperature response values, determines functional stage boundaries. The functional test response data is then continuously segmented to obtain a functional stage response data set.

[0031] The system further reads the historical test records that the equipment has completed and stored, filters out historical normal test records that match the equipment model, chamber volume, target pressure, target temperature and functional test procedures of the current functional test, and aligns them according to the functional stage boundaries to generate a normal functional response set. Then, the system extracts the differences between the functional stage response data set and the data with the same functional stage in the normal functional response set to obtain the pressure residual response sequence and temperature residual response sequence. Through residual direction identification, continuous residual segment statistics and residual amplitude accumulation, a stage residual response difference set is formed. This processing can isolate the normal differences caused by equipment specifications, test objectives and test procedures, highlighting the response deviations that are truly related to the fault.

[0032] During the fault diagnosis process, the system inputs the set of residual response differences of each stage into the vacuum residual drive inversion Gaussian potential model, constructs a set of Gaussian kernels according to the functional stages, and obtains the set of residual driving force characteristics through Gaussian process potential force inversion. This set of residual driving force characteristics includes residual driving force characteristics of pumping, residual driving force characteristics of pressure holding, residual driving force characteristics of temperature control, residual driving force characteristics of valve hysteresis, residual driving force characteristics of pressure recovery, and residual driving force characteristics of interlock protection. The system generates fault diagnosis results based on the occurrence stage, direction of change, duration, and combination relationship of each residual driving force characteristic.

[0033] During continuous operation verification, when the equipment exhibited a slowdown in vacuum pumping response, the system did not directly diagnose it as a pump unit failure. Instead, it attributed the cause by combining the characteristics of residual pumping driving force and residual pressure holding driving force. If the two characteristics changed in the same direction and overlapped in duration, a seal leakage fault item was generated, and the fault location was pointed to the vacuum chamber sealing area. If the residual pumping driving force characteristic persisted but did not overlap with the residual pressure holding driving force characteristic in duration, a pump unit performance degradation fault item was generated, and the fault location was pointed to the vacuum pump unit. During temperature control and valve switching processes, the system could also identify temperature control execution faults and valve execution faults by respectively analyzing the relationships between temperature control residual driving force characteristics, valve hysteresis residual driving force characteristics, and pressure recovery residual driving force characteristics.

[0034] As can be seen from the above applications, this invention can unify electrical test data, drive feedback data, and functional response data into the same diagnostic link. It completes fault attribution layer by layer through functional stage boundaries, normal functional response sets, stage residual response difference sets, and residual driving force characteristic sets. Continuous operation records, historical normal test records, equipment maintenance records, and fault review records jointly demonstrate that this invention can map abnormalities in air extraction, pressure holding, temperature control, valve response, and interlock protection to fault type, fault location, and fault severity, and generate maintenance processing information to be sent to the operation and maintenance terminal. This improves the initiative, accuracy, and interpretability of fault diagnosis and shortens the troubleshooting path for operation and maintenance personnel.

[0035] Table 1. Comparison of Comprehensive Performance of Fault Diagnosis for Vacuum Environment Simulation Equipment

[0036] As shown in Table 1, compared with the method combining manual inspection and fixed threshold alarms, the fault type identification accuracy of the system of the present invention increased from 76.9% to 92.4%, the fault location accuracy increased from 70.5% to 89.6%, and the fault severity consistency rate increased from 68.2% to 86.8%. This indicates that the present invention can not only detect whether there are abnormalities in the vacuum environment simulation equipment, but also further determine the fault type, fault location, and fault severity corresponding to the abnormality. The reason for this is that the method combining manual inspection and fixed threshold alarms mainly relies on alarm thresholds and manual observation of curve changes, which easily leads to mixed judgments of slowed vacuuming response, pressure recovery during holding, temperature control response deviation, and valve action lag. In contrast, the present invention first divides the functional test response data into functional stages, and then constructs a set of normal functional responses that match the current functional test, so that the actual response of each functional stage has a corresponding normal reference.

[0037] Compared to conventional machine learning classification and diagnostic methods, the fault type identification accuracy of the system in this invention is improved from 84.7% to 92.4%, the fault location accuracy from 80.2% to 89.6%, and the fault severity consistency rate from 76.5% to 86.8%. While conventional machine learning classification and diagnostic methods can utilize historical data for fault classification, they typically input pressure, temperature, current, and valve feedback as overall features into the classification model, resulting in outputs biased towards fault labels and lacking explanations for the sources of response deviations in different functional stages. This invention extracts pressure residual response sequences, temperature residual response sequences, residual direction, residual duration, and residual accumulation through a set of stage residual response differences. Then, it uses a vacuum residual drive inversion Gaussian potential model to invert the residual driving force characteristics of pumping, pressure holding, temperature control, valve hysteresis, pressure recovery, and interlock protection. Therefore, it can further transform surface response anomalies into residual driving force attribution results.

[0038] In terms of false alarm rate and false negative rate, the false alarm rate of the system of this invention is 4.8%, which is lower than the 12.6% of the method combining manual inspection and fixed threshold alarm, and also lower than the 8.9% of the conventional machine learning classification and diagnosis method. The false negative rate of the system of this invention is 3.7%, which is lower than the 10.9% of the method combining manual inspection and fixed threshold alarm, and also lower than the 7.4% of the conventional machine learning classification and diagnosis method. The reason for the decrease in false alarm rate is that this invention generates a set of normal functional responses based on equipment model, cabin volume, target pressure, target temperature and functional test procedures, avoiding misjudging differences in equipment specifications and test targets as faults. The reason for the decrease in false negative rate is that this invention not only identifies instantaneous anomalies, but also counts the residual duration and residual accumulation, which can detect progressive anomalies such as seal decay, pump performance degradation and valve lag.

[0039] In terms of operational efficiency, the average time for a single diagnosis using the system of this invention is 11.8 minutes, compared to 42.5 minutes for the combination of manual inspection and fixed threshold alarms, and 24.6 minutes for the conventional machine learning classification diagnosis method. Regarding the number of invalid disassemblies per 100 functional tests, the system of this invention has 6, the combination of manual inspection and fixed threshold alarms has 18, and the conventional machine learning classification diagnosis method has 11. These results demonstrate that the fault diagnosis results generated by this invention can directly include the fault type, fault location, and fault severity, and further generate maintenance processing information. This allows maintenance personnel to directly inspect the vacuum chamber sealing area, vacuum pump group, temperature control circuit, valve drive structure, or interlock protection control circuit, thereby reducing invalid disassemblies and shortening the diagnosis time.

[0040] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A fault diagnosis system for vacuum environment simulation equipment based on functional testing, characterized in that, include: The data acquisition module is used to collect raw functional test data of the vacuum environment simulation equipment during the functional testing process; The data processing module is used to preprocess the raw functional test data to obtain functional test response data; The phase segmentation module is used to extract phase identification data from the functional test response data and segment the functional test response data to obtain a functional phase response data set. The normal response construction module is used to read the historical test records that have been completed and stored by the vacuum environment simulation equipment, filter the normal test records and align them according to the functional stage boundaries, and extract the normal response data corresponding to each functional stage to obtain the normal functional response set. The residual difference extraction module is used to extract the differences between the functional stage response data set and the normal functional response set to obtain the stage residual response difference set. The residual drive inversion module is used to input the set of stage residual response differences into the vacuum residual drive inversion Gaussian potential model, perform Gaussian process potential force inversion on the set of stage residual response differences, and obtain the set of residual driving force characteristics. The fault diagnosis module is used to generate fault diagnosis results based on each residual driving force feature in the residual driving force feature set; The maintenance output module is used to generate maintenance processing information based on the fault diagnosis results and send the fault diagnosis results and maintenance processing information to the operation and maintenance terminal of the vacuum environment simulation equipment.

2. The fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The data acquisition module includes: The functional testing process includes vacuuming, pressure holding, temperature control, valve switching, pressure recovery, and interlock protection. The raw data for the functional tests include electrical test data, drive feedback data, and functional response data; The drive feedback data is used to characterize the execution feedback status of valves, relays, and control commands.

3. The fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The data processing module includes: Preprocessing includes time alignment, signal identification unification, invalid sample point removal, missing sample point completion, and data format conversion to obtain functional test response data; The functional test response data includes standard electrical test data, standard drive feedback data, and standard functional response data.

4. The fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The phase division module includes: The drive feedback status and function response values ​​are read from the phase identification data according to the acquisition time to form a phase identification sequence; State transition identification is performed on the driving feedback states in the stage identification sequence. The acquisition time corresponding to the state transition is determined as the driving trigger point. The driving trigger points are arranged according to the acquisition time to obtain the driving trigger point sequence. The adjacent sampling difference is calculated for the functional response values ​​in the stage identification sequence. The acquisition time when the sign of the adjacent sampling difference changes or the adjacent sampling difference changes from non-zero to zero is determined as the response inflection point. The response inflection points are arranged according to the acquisition time to obtain the response inflection point sequence. Based on the order of acquisition time, each drive trigger point in the drive trigger point sequence is paired with a response inflection point that is adjacent to the acquisition time after that drive trigger point to obtain stage boundary candidate pairs. The candidate pair of stage boundaries with the smallest acquisition time difference is selected from the candidate pairs of stage boundaries as the stage boundary pair. The acquisition time of the driving trigger point in the stage boundary pair is used as the functional stage boundary. The functional stage boundaries are arranged according to the acquisition time to obtain the functional stage boundary sequence. The functional test response data is continuously segmented according to the functional phase boundary sequence to obtain the functional phase response data set.

5. A fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The normal response construction module includes: Read the equipment test conditions corresponding to the current functional test, and combine the equipment model, cabin volume, target pressure, target temperature and functional test procedure into the current test condition identifier; Read historical normal test records and their historical test condition identifiers from the historical test records that have been completed and stored by the vacuum environment simulation equipment. Match the historical test condition identifiers with the current test condition identifiers. Retain historical normal test records that are consistent in terms of equipment model, chamber volume, target pressure, target temperature and functional test procedures to obtain the matched normal test records. Determine the start and end points of each functional phase in the current functional test according to the functional phase boundaries, and combine the phase start and end points into a phase alignment benchmark. The normal response data in the matching normal test record is stage-aligned according to the stage alignment benchmark to obtain stage-aligned normal response data. Extract the normal response data corresponding to each functional stage from the stage alignment normal response data, and combine the normal response data corresponding to each functional stage into a normal functional response set.

6. The fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The residual difference extraction module includes: The functional stage response data in the functional stage response data set is matched with the normal response data in the normal functional response set according to the functional stage to obtain the stage matching data set. The functional stage response data and normal response data in the stage matching data set are sampled and matched according to the collection time sequence to obtain the stage matching data set; Read the stage matching data corresponding to each functional stage from the stage matching data set, extract the functional stage pressure response and functional stage temperature response from the functional stage response data, extract the normal pressure response and normal temperature response from the normal response data, and perform sampling point difference calculation to obtain the pressure residual response sequence and temperature residual response sequence. The residual direction, continuous residual segment statistics, and residual amplitude accumulation are performed on the pressure residual response sequence to obtain the pressure residual direction, pressure residual duration, and pressure residual accumulation. The residual direction, continuous residual segment statistics, and residual amplitude accumulation are performed on the temperature residual response sequence to obtain the temperature residual direction, temperature residual duration, and temperature residual accumulation. The residual pressure direction, residual pressure duration, residual pressure accumulation, residual temperature direction, residual temperature duration, and residual temperature accumulation are combined according to the corresponding functional stages to obtain the stage residual response item set. According to the functional phase sequence corresponding to the functional testing process, the set of residual response items for each phase is collected to obtain the set of residual response differences for each phase.

7. A fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The residual drive inversion module includes: The set of stage residual response differences is input into the vacuum residual drive inversion Gaussian potential model, which includes a residual input construction layer, a stage Gaussian kernel construction layer, a potential force inversion layer, a residual driving force feature extraction layer, and a residual driving force convergence output layer. In the residual input construction layer, the residual response difference set of the stage is sorted according to the functional stage order to obtain the residual input sequence carrying the functional stage identifier, and the residual input sequence is sent to the stage Gaussian kernel construction layer and the potential force inversion layer. In the stage Gaussian kernel construction layer, based on the functional stages and residual response differences corresponding to the residual input sequence, stage Gaussian kernels corresponding to each functional stage are constructed, and the stage Gaussian kernels corresponding to each functional stage are combined into a stage Gaussian kernel set. In the latent force inversion layer, the Gaussian kernel set of the stages is called to generate the covariance matrix corresponding to each functional stage, and the Gaussian process latent force inversion is performed on the residual response difference in the residual input sequence based on the covariance matrix to obtain the residual driving force sequence. In the residual driving force feature extraction layer, the residual driving force sequence is marked with the occurrence stage, the driving force direction is identified, the driving force duration is statistically analyzed, and the cumulative driving force intensity is extracted to obtain the residual driving force stage feature set. In the residual driving force convergence output layer, the residual driving force stage feature set is aggregated according to the functional stage to obtain the residual driving force feature set.

8. A fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The fault diagnosis module includes: The residual driving force features of air extraction, pressure holding, temperature control, valve hysteresis, pressure recovery, and interlock protection are read from the residual driving force feature set. The occurrence stage, change direction, and duration of each residual driving force feature are extracted to obtain residual driving force discrimination data. Based on the stage of occurrence, direction of change, and duration, the residual driving force characteristics in the residual driving force discrimination data are identified by cross-stage association, same-stage association, and duration overlap to obtain the combination relationship of residual driving forces. From the residual driving force combination relationship, extract the cross-stage combination of the residual driving force characteristics of air extraction and the residual driving force characteristics of pressure holding with the same change direction and overlapping duration to obtain the sealing leakage fault item; The pump set performance degradation fault item is obtained by extracting cross-stage combinations of residual driving force characteristics that have a duration but do not overlap with the duration of residual driving force characteristics of pressure holding. Extract the temperature control residual driving force characteristics of the same stage combination of the duration of existence to obtain the temperature control execution fault item; The valve execution failure item is obtained by extracting the valve hysteresis residual driving force characteristics and the pressure recovery residual driving force characteristics with a cross-stage combination of duration overlap. The interlocking protection residual driving force characteristics and any residual driving force characteristics with the same stage having a duration overlap length are combined to obtain the interlocking protection associated fault item; The fault diagnosis results are obtained by categorizing the fault items related to sealing leakage, pump performance degradation, temperature control, valve operation, and interlock protection.

9. A fault diagnosis system for vacuum environment simulation equipment based on functional testing according to claim 1, characterized in that, The maintenance output module includes: Read the fault type, fault location, and fault severity from the fault diagnosis results, and combine the fault type, fault location, and fault severity into fault description information; Read the fault maintenance correspondence stored in the system; Match the fault description information with the corresponding fault maintenance to obtain maintenance processing matching items; The maintenance object, maintenance action, and processing order in the maintenance processing match are combined into maintenance processing information; The fault diagnosis results and maintenance information are sent to the operation and maintenance terminal of the vacuum environment simulation equipment.

Images