An automated test system based on parametric model data processing

Abstract

The application relates to the technical field of automatic testing, in particular to an automatic testing system based on parameter model data processing, which comprises the following steps: a data acquisition module determines a to-be-tested object and a to-be-tested target, identifies a key part, determines a detection sensor and deploys the detection sensor, and acquires a plurality of real-time acquisition parameters; a data processing module receives the plurality of real-time acquisition parameters, removes abnormal values, estimates and fills up, and carries out normalization processing on the plurality of real-time acquisition parameters; a parameter modeling module determines parameter characteristics of the processing parameters, selects a plurality of processing models based on the parameter characteristics, carries out fusion modeling on the plurality of processing models, and carries out real-time optimization on the fusion model parameters through a dynamic model updating mechanism; a testing module realizes abnormal detection and fault position positioning based on the fusion model and through comparison of the real-time acquisition parameters and multi-parameter causal analysis; and a result visualization module displays a test result and generates a feedback report. The application improves the comprehensiveness of the automatic testing system.

Description

Technical Field

[0001] This application relates to the field of automated testing technology, and in particular to an automated testing system based on parametric model data processing. Background Technology

[0002] Automated testing technology plays a crucial role in the monitoring and quality control of modern industrial equipment. Traditional automated testing systems mostly rely on result reference ranges for judgment, that is, they use preset fixed threshold ranges to determine the pass / fail status of the tested parameters.

[0003] However, establishing reference intervals has limitations. The accuracy of reference intervals is highly dependent on the quality and quantity of historical data collected. If historical data contains errors, is missing, or is not representative, the established reference interval may be biased, leading to incorrect test result judgments. The working environment and usage conditions of the tested object may change, while reference intervals are often established based on specific historical environments and conditions. When environmental factors or usage conditions change significantly, the original reference intervals may no longer be applicable. Traditional platforms can only determine whether parameters exceed limits, but cannot locate the root cause of the fault, requiring manual troubleshooting based on experience, which is inefficient. Updating reference intervals relies on manual operation, which requires frequent adjustments in complex systems, is prone to human error, and is time-consuming and labor-intensive.

[0004] Therefore, existing technologies suffer from limitations in the accuracy of test result determination, low efficiency in fault location, and reliance on manual operation for updating reference intervals. Summary of the Invention

[0005] To address the aforementioned issues, this disclosure provides an automated testing system based on parametric model data processing. Through multi-model fusion modeling, dynamic updates, and deep diagnostic techniques, it achieves accurate, adaptive, and intelligent testing and fault analysis.

[0006] To achieve the above objectives, the present invention provides an automated testing system based on parametric model data processing, comprising: The data acquisition module is used to determine the test object and the test target, identify the key parts of the test object, determine the detection sensor based on the test target, deploy the detection sensor based on several key parts, and acquire several real-time acquisition parameters based on the detection sensor. A data processing module, connected to the data acquisition module, is used to receive several real-time acquisition parameters, and to perform outlier removal, estimation and filling, and normalization on the several real-time acquisition parameters to obtain several processing parameters. A parameter modeling module, connected to the data processing module, is used to determine the parameter characteristics of the processing parameters, select several processing models based on the parameter characteristics, fuse the several processing models into a model, and optimize the parameters of the fused model in real time through a dynamic model update mechanism. The testing module, connected to the parameter modeling module, is used to obtain anomaly detection and fault location based on the fusion model through real-time parameter comparison and multi-parameter causal analysis. The results visualization module is connected to the testing module to display the test results and generate feedback reports based on the test results.

[0007] Furthermore, The data acquisition module includes: A part identification unit is used to identify the structural features of the object under test and obtain historical fault data of the object under test, and to determine the key parts based on the structural features and the historical fault data; The sensor determination unit is used to match the parameter type of the target under test with the detection parameter type of the sensor in the preset sensor database, and determine the detection sensor based on the matching result; The deployment unit is connected to the part identification unit and the sensor determination unit to deploy the detection sensor based on the key part.

[0008] Furthermore, The location identification unit includes: Extract sub-units to extract the coordinates of high-stress or vulnerable areas of the test object through finite element simulation analysis; An identification subunit, connected to the extraction subunit, is used to determine the statistical characteristics of historical fault locations based on several historical fault data, and to generate a fault priority list based on the statistical characteristics using a clustering algorithm. The location determination subunit is connected to the extraction subunit and the identification subunit to determine the critical location based on the coordinates of the high-stress area or the vulnerable area and the fault priority list.

[0009] Furthermore, The data processing module includes: The removal unit is used to identify outliers in the real-time acquired parameters using the isolated forest algorithm, so as to remove the outliers; The data imputation unit is used to select the appropriate model for data imputation based on time series correlation analysis. The processing unit is used to identify the parameter type corresponding to the real-time acquired parameters, and to match the parameter type with a preset normalization method to obtain the processing parameters.

[0010] Furthermore, The parameter modeling module includes: The feature determination unit is used to analyze the data characteristics and physical laws of the processing parameters, select a modeling method based on the data characteristics and physical laws, and then determine the processing model. A fusion unit, connected to the feature determination unit, is used to perform fusion modeling on the processing model based on a weighted fusion method to obtain a fusion model; A parameter optimization unit, connected to the fusion unit, is used to determine the full training cycle and update the fusion model based on updated test data to optimize the parameters of the fusion model.

[0011] Furthermore, The parameter optimization unit includes: A periodicity determination subunit is used to calculate the performance degradation index of the fusion model, and the full training period is calculated based on the performance degradation index. The data integration subunit, together with the period determination subunit, is used to obtain updated test data based on the full training period, and to mix the latest test data with the historical dataset according to time weights to obtain an updated training dataset. An optimization subunit, connected to the data integration subunit, is used to train the fusion model based on the updated training dataset to optimize the parameters of the fusion model.

[0012] Furthermore, The testing module includes: An anomaly detection unit is used to compare several real-time parameters with the threshold range output by the fusion model to determine anomaly parameters, determine the sampling point distribution corresponding to the anomaly parameters, and determine anomalies based on the sampling point distribution judgment results. The fault location unit, connected to the anomaly detection unit, is used to trace the root cause of abnormal parameters based on the parameter cause-effect graph determined by parameter modeling, so as to determine the fault location.

[0013] Furthermore, The anomaly detection unit includes: The threshold calculation subunit is used to calculate the dynamic upper and lower limit thresholds based on the predicted values ​​output by the fusion model and the statistical distribution of historical data, and to perform linear compensation on the upper and lower limit thresholds according to the ambient temperature to obtain the threshold range. An anomaly determination subunit, connected to the threshold calculation subunit, is used to draw a parameter change graph based on the real-time parameters, mark the abnormal region in the parameter change graph based on the threshold range, determine the distribution of sampling points in the abnormal region, and determine the detected anomaly based on the distribution judgment result.

[0014] Furthermore, The fault location unit includes: The causal reasoning subunit is used to load the parametric causal graph generated during the parametric modeling process, trace the abnormal propagation path in reverse along the edge weights, eliminate paths that violate physical laws, calculate the posterior probability of each node based on the Bayesian network, and locate the root cause of the component whose failure probability is greater than the preset failure probability. The impact assessment subunit is connected to the causal reasoning subunit to predict the fault development trend through the LSTM network, output the remaining effective time and confidence interval, and calculate the fault propagation index.

[0015] Furthermore, The results visualization module includes: The drawing unit is used to dynamically display the health status of key parts of the test object through color mapping and overlay finite element simulation stress cloud map; The interactive analysis unit, connected to the drawing unit, is used to set multi-layer switching display modes, including parameter trend graphs, spectrum analysis graphs and causal tracing path graphs, set custom time window scaling, and export local data to generate PDF / CSV format reports. The report generation unit is connected to the interactive analysis unit to automatically match the historical case library based on the fault location results and generate a decision report that includes maintenance steps, spare parts list and cost estimate. The closed-loop feedback unit is used to send maintenance records and verification data back to the parameter modeling module, triggering the incremental learning process of the model.

[0016] Compared with the prior art, this disclosure has the following advantages: 1. The multi-model fusion modeling and dynamic update mechanism effectively reduces the dependence on single historical data, enhances model robustness, avoids misjudgments due to data problems, and adapts to changes in environment and conditions in real time to ensure that test results remain accurate in different scenarios.

[0017] 2. The platform is no longer limited to simply judging whether something is qualified or not. With the help of causal analysis and multi-parameter comprehensive analysis, it can deeply locate the root cause of the fault, accurately assess the impact of the fault, and automatically generate feedback reports to provide scientific guidance for maintenance, greatly shorten the troubleshooting time, and improve maintenance efficiency.

[0018] 3. Modular design and multi-model fusion architecture enable it to easily handle testing needs for new objects and parameters, reducing maintenance costs, while dynamically tracking parameter change trends to meet monitoring needs in complex scenarios. In terms of intelligent management, the platform automatically manages reference ranges, avoiding human error and improving automation levels. Combined with intelligent feedback and optimization suggestions, it achieves closed-loop optimization, continuously improving platform performance.

[0019] 4. Automated testing and intelligent diagnostics reduce manpower input, shorten equipment downtime, and improve production efficiency. At the same time, accurate fault diagnosis prevents equipment failures, extends service life, reduces equipment replacement costs, and brings long-term benefits to enterprises.

[0020] Other features and advantages of this disclosure will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the disclosure. The objects and other advantages of this disclosure may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A structural block diagram of an automated testing system based on parametric model data processing provided in this application embodiment; Figure 2 A flowchart of an automated testing system based on parametric modeling and data processing provided in an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0024] Figure 1 This diagram illustrates a structural block diagram of an automated testing system based on parametric model data processing, according to an embodiment of this application. Figure 1 As shown in the figure, an automated testing system based on parametric model data processing according to an embodiment of this application includes: Data acquisition module 10 is used to determine the test object and the test target, identify the key parts of the test object, determine the detection sensor based on the test target, deploy the detection sensor based on several key parts, and acquire several real-time acquisition parameters based on the detection sensor. Data processing module 20 is connected to data acquisition module 10 and is used to receive several real-time acquisition parameters, perform outlier removal, estimation and filling and normalization on several real-time acquisition parameters to obtain several processing parameters. The parameter modeling module 30 is connected to the data processing module 20 and is used to determine the parameter characteristics of the processing parameters, select several processing models based on the parameter characteristics, fuse the several processing models into a model, and optimize the parameters of the fused model in real time through a dynamic model update mechanism. Test module 40 is connected to parameter modeling module 30 and is used to obtain anomaly detection and fault location based on the fusion model through real-time parameter comparison and multi-parameter causal analysis. The results visualization module 50 is connected to the test module 40 to display test results and generate feedback reports based on the test results.

[0025] The data acquisition module 10 includes: A part identification unit is used to identify the structural features of the object under test and obtain historical fault data of the object under test, and to determine the key parts based on the structural features and the historical fault data; The sensor determination unit is used to match the parameter type of the target under test with the detection parameter type of the sensor in the preset sensor database, and determine the detection sensor based on the matching result; The deployment unit is connected to the part identification unit and the sensor determination unit to deploy the detection sensor based on the key part.

[0026] In this embodiment, the preset sensor database includes at least three of the following: current sensor, voltage sensor, acceleration sensor, temperature sensor, pressure sensor, and displacement sensor.

[0027] Both types of data can meet basic needs. Basic monitoring can be achieved when the two types of data can complementaryly cover the main fault characteristics. For example, two-dimensional detection of motor temperature and current. Temperature reflects thermal accumulation faults such as mechanical wear, poor lubrication, and overload (slow change); current reflects electrical short circuits, stalled rotor, and instantaneous impacts (fast change). Sudden current increases and gradual temperature rises are consistent with early characteristics of bearing wear. The limitations of these two types of data are that they cannot identify coupled faults (such as poor electrical contact caused by vibration), and isolated points are easily mistaken for noise. They also have weak ability to distinguish nonlinear modes (the triangular relationship between pressure, temperature, and flow rate), and cannot reflect the actual situation.

[0028] However, using three or more data sets has significant advantages. For example, in hydraulic system fault diagnosis, single-dimensional jumps (sensor momentary disconnection) or normal pressure, decreased flow, and slight temperature rise are due to filter blockage (feature decoupling). DBSCAN recommends a minimum core point (MinPts) value of the number of dimensions + 1 or 2 × the number of dimensions. Two-dimensional: MinPts = 3-4, more sensitive to noise, easily misclassifying small normal clusters as noise. Three-dimensional: MinPts = 4-6, requires more neighboring points to confirm core points, improving noise resistance.

[0029] In this embodiment, a modular deployment unit is also provided to provide a detachable sensor installation interface and a standardized communication protocol, supporting rapid sensor networking and parameter acquisition channel configuration for different measured objects, and real-time verification of the matching degree between the sensor deployment location and key parts.

[0030] The location identification unit includes: Extract sub-units to extract the coordinates of high-stress or vulnerable areas of the test object through finite element simulation analysis; An identification subunit, connected to the extraction subunit, is used to determine the statistical characteristics of historical fault locations based on several historical fault data, and to generate a fault priority list based on the statistical characteristics using a clustering algorithm. The location determination subunit is connected to the extraction subunit and the identification subunit to determine the critical location based on the coordinates of the high-stress area or the vulnerable area and the fault priority list.

[0031] In this embodiment, the clustering algorithm uses DBSCAN (density-based clustering) or hierarchical clustering algorithm to solve the problems of noise points (random faults) and non-spherical distribution that may exist in the fault data.

[0032] The specific process of DBSCAN (density-based clustering) is as follows: Step 1: Data acquisition and preprocessing.

[0033] Multi-dimensional time-series data (such as temperature, pressure, and current) are collected from the PLC testing system and then processed to remove dimensions.

[0034] Step 2: Feature engineering to construct density space The original signal is converted into a high-dimensional vector that reflects the fault characteristics. For example, in a bearing fault scenario, multi-channel features such as magnetic flux density and permeability can be extracted.

[0035] Step 3: DBSCAN Core Parameter Selection Two key parameters need to be dynamically adjusted based on data characteristics: ε (neighborhood radius): Usually taken as the inflection point of the k-distance curve of the sample interval. MinPts (Minimum number of core points): Generally set to the number of feature dimensions + 1 or 2 × the number of feature dimensions. Step 4: Perform clustering and noise labeling The algorithm automatically executes a three-step process: "identifying core points → expanding clusters → labeling noise". (1) Core point determination: If there are at least MinPts points in the ε neighborhood of a point, then it is marked as a core point. (2) Cluster expansion: Clusters are formed by merging core points through density reachability. (3) Noise identification: Data that is neither a core point nor a boundary point is marked as noise (labeled -1). Step 5: Result Interpretation and Fault Warning Normal operating condition cluster: High-density areas correspond to stable states under different loads (such as high-speed / low-speed motor operation modes). Noise points: Occasional sensor fluctuations or transient interference; if these occur continuously, an early warning will be triggered. Non-spherical clusters: Identify irregularly distributed gradual faults (such as temperature ring diffusion patterns caused by bearing wear). In this embodiment, the specific working process of the identification subunit includes: Identify the historical fault data, extract the fault fields corresponding to the historical fault data, and convert the fault fields into feature vectors; Density clustering is performed based on feature vectors and clustering algorithms to determine several core clusters; Calculate the cluster density corresponding to any of the core clusters, and determine a fault priority list based on the sorting results of several cluster densities.

[0036] The data processing module 20 includes: The removal unit is used to identify outliers in the real-time acquired parameters using the isolated forest algorithm, so as to remove the outliers; The data imputation unit is used to select the appropriate model for data imputation based on time series correlation analysis. The processing unit is used to identify the parameter type corresponding to the real-time acquired parameters, and to match the parameter type with a preset normalization method to obtain the processing parameters.

[0037] In this embodiment, the removal unit utilizes advanced data mining and analysis algorithms to automatically identify and remove outliers in the collected data. For example, it uses cluster analysis and isolated forest algorithms to detect and remove noise points that significantly deviate from the normal data distribution.

[0038] The imputation unit intelligently estimates and imputes missing data. Based on the characteristics of the time series or the correlation of relevant parameters, it adopts appropriate data interpolation methods, such as linear interpolation, spline interpolation, or machine learning-based prediction models, to ensure data integrity and provide a high-quality data foundation for subsequent modeling. The processing unit automatically selects an appropriate standardization method, such as Z-score standardization or Min-Max standardization, based on the parameter type. This transforms data with different dimensions and distributions to the same scale range, eliminating the impact of differences in dimensions and data distributions on modeling, improving the stability and accuracy of the model, and facilitating subsequent parameter modeling and comprehensive analysis.

[0039] The parameter modeling module 30 includes: The feature determination unit is used to analyze the data characteristics and physical laws of the processing parameters, select a modeling method based on the data characteristics and physical laws, and then determine the processing model. A fusion unit, connected to the feature determination unit, is used to perform fusion modeling on the processing model based on a weighted fusion method to obtain a fusion model; A parameter optimization unit, connected to the fusion unit, is used to determine the full training cycle and update the fusion model based on updated test data to optimize the parameters of the fusion model.

[0040] The parameter modeling module combines statistical, machine learning, and physical models to construct accurate parameter models for different types of parameters and test objects. For example, for parameters with clear physical laws, a model based on physical equations is established, using known physical laws and material properties to describe the relationships between parameters. For complex, nonlinear parameters that are difficult to describe accurately using physical models, machine learning methods, such as neural networks and support vector machines, are employed to learn the inherent patterns and laws of parameters from a large amount of historical data, uncovering hidden information in the data and achieving accurate parameter modeling. Ensemble learning techniques are used to fuse multiple different types of models, leveraging strategies such as weighted averaging and voting to fully utilize the advantages of each model, improving its generalization ability and prediction accuracy, enabling the model to more accurately reflect the parameter characteristics and variation patterns of the tested object under various working conditions.

[0041] The system monitors the operating status and environmental conditions of the tested object in real time. When significant changes in environmental factors (such as temperature, humidity, and pressure) or operating conditions (such as load and speed) are detected, the model update process is automatically triggered. Utilizing incremental learning algorithms, newly acquired data is dynamically integrated into the existing model, and model parameters are adjusted in real time to ensure the model always adapts to the current testing environment and operating conditions, maintaining its accuracy and effectiveness and avoiding misjudgments caused by changes in environment and conditions.

[0042] Regularly retrain and optimize the model as a whole, and combine the latest test data and expert feedback to make in-depth adjustments to the model structure and parameters, continuously improving the model's performance and adaptability, so that the platform can stably cope with various complex and ever-changing test scenarios in the long term.

[0043] The parameter optimization unit includes: A periodicity determination subunit is used to calculate the performance degradation index of the fusion model, and the full training period is calculated based on the performance degradation index. The data integration subunit, together with the period determination subunit, is used to obtain updated test data based on the full training period, and to mix the latest test data with the historical dataset according to time weights to obtain an updated training dataset. An optimization subunit, connected to the data integration subunit, is used to train the fusion model based on the updated training dataset to optimize the parameters of the fusion model.

[0044] In this embodiment, the performance degradation index is the recent error change rate.

[0045] The formula for calculating the full training cycle is: In the formula, To preset the training period, Let λ be the rate of change of recent error, and λ be the attenuation coefficient.

[0046] In this embodiment, the time weight is 7:3 for the latest test data and the historical dataset.

[0047] The test module 40 includes: An anomaly detection unit is used to calculate the variance of the residual sequence of real-time parameters through a sliding window based on the threshold range output by the fusion model, determine the abnormal parameters based on the variance of the residual sequence, determine the distribution of sampling points corresponding to the abnormal parameters, and determine the anomaly based on the sampling point distribution judgment result. The fault location unit, connected to the anomaly detection unit, is used to trace the root cause of abnormal parameters based on the parameter cause-effect graph determined by parameter modeling, so as to determine the fault location.

[0048] In this embodiment, the anomaly detection unit compares the real-time collected parameter values ​​with the normal parameter range calculated based on the parameter model. Utilizing a highly efficient real-time data processing algorithm, it quickly determines whether parameters exceed the normal range. For parameters exceeding the range, it immediately issues a preliminary warning signal, indicating a potential anomaly and ensuring that potential faults are detected at the earliest possible time.

[0049] When the fault location unit detects abnormal parameters, it initiates a deep fault diagnosis process. Using causal analysis algorithms, combined with the correlations and causal logic between parameters discovered during parameter modeling, it traces the root cause of the abnormal parameters to determine the specific component or process causing the fault. For example, by constructing causal graphs or Bayesian networks between parameters, it analyzes the dependencies and transmission paths between parameters, accurately locating the fault source from numerous parameters, providing a clear direction and basis for subsequent maintenance and adjustments.

[0050] This embodiment also includes a multi-parameter comprehensive analysis method, considering the coordinated change characteristics of multiple related parameters, to further analyze the severity and development trend of the fault. By analyzing the parameter change rate, change law, and correlation between parameters, the impact of the fault on the overall system performance is assessed, and the possible development trend of the fault is predicted. This provides a scientific basis for formulating reasonable maintenance strategies and plans, effectively improving the reliability and operating efficiency of the equipment.

[0051] The anomaly detection unit includes: The threshold calculation subunit is used to calculate the dynamic upper and lower limit thresholds based on the predicted values ​​output by the fusion model and the statistical distribution of historical data, and to perform linear compensation on the upper and lower limit thresholds according to the ambient temperature to obtain the threshold range. An anomaly determination subunit, connected to the threshold calculation subunit, is used to draw a parameter change graph based on the real-time parameters, mark the abnormal region in the parameter change graph based on the threshold range, determine the distribution of sampling points in the abnormal region, and determine the detected anomaly based on the distribution judgment result.

[0052] In this embodiment of the disclosure, the calculation of dynamic upper and lower thresholds includes: Upper limit = k × standard deviation of historical data + predicted value Lower limit = predicted value - k × standard deviation of historical data In the formula, k is an adjustable coefficient, usually taken as 2 to 3, to control the threshold width; Based on the real-time collected ambient temperature, linear compensation for the upper and lower thresholds includes: Threshold compensation = α × (T) T 基准 ), In the formula, α is the temperature compensation coefficient, T is the current ambient temperature, and T 基准 Reference temperature; The threshold compensation is added to the upper and lower thresholds to obtain the final dynamic threshold range.

[0053] In this embodiment, the fault location unit includes: The causal reasoning subunit is used to load the parametric causal graph generated during the parametric modeling process, trace the abnormal propagation path in reverse along the edge weights, eliminate paths that violate physical laws, calculate the posterior probability of each node based on the Bayesian network, and locate the root cause of the component whose failure probability is greater than the preset failure probability. The impact assessment subunit is connected to the causal reasoning subunit to predict the fault development trend through the LSTM network, output the remaining effective time and confidence interval, and calculate the fault propagation index.

[0054] In this embodiment, the preset failure probability is 85%.

[0055] The results visualization module includes: The drawing unit is used to dynamically display the health status of key parts of the test object through color mapping and overlay finite element simulation stress cloud map; The interactive analysis unit, connected to the drawing unit, is used to set multi-layer switching display modes, including parameter trend graphs, spectrum analysis graphs and causal tracing path graphs, set custom time window scaling, and export local data to generate PDF / CSV format reports. The report generation unit is connected to the interactive analysis unit to automatically match the historical case library based on the fault location results and generate a decision report that includes maintenance steps, spare parts list and cost estimate. The closed-loop feedback unit is used to send maintenance records and verification data back to the parameter modeling module to trigger the incremental learning process of the model.

[0056] In this embodiment, the incremental learning process involves integrating the maintenance records and verification data with the latest test data to form an incremental training dataset when the maintenance records and verification data are sent back to the parameter modeling module or the system detects a decline in model performance (such as the recent error change rate exceeding a set threshold).

[0057] The fusion model is locally trained using incremental learning algorithms (such as online gradient descent, incremental SVM, incremental random forest, etc.). Update the trained model parameters to the online model; Use a validation set to evaluate model performance and ensure that the model's accuracy improves or remains stable after incremental learning.

[0058] In this embodiment, the drawing unit dynamically displays the health status of each key part of the object under test using a color mapping method. Red indicates a failure probability >80%, and green indicates normal.

[0059] Figure 2 A flowchart illustrating the automated testing system based on parametric modeling and data processing provided in this application embodiment. (See also...) Figure 2 As shown, the detailed steps of the process include: Platform initialization: Load sensor configuration and initialize data storage.

[0060] Data Acquisition and Preprocessing: Initiate data acquisition and clean and standardize the acquired data.

[0061] Parametric modeling: Update the parameter model based on the preprocessed data.

[0062] Automated testing: Collect real-time data and compare it with the normal range predicted by the model.

[0063] Fault diagnosis: If the data exceeds the normal range, initiate in-depth fault diagnosis to locate the cause of the fault and assess its impact.

[0064] Results Feedback and Optimization: Generate a feedback report, send it to relevant personnel, record the repair status and evaluate the repair effectiveness, and update the model parameters. If the data is within the normal range, continue testing or terminate the test.

[0065] Based on the same inventive concept as the above disclosure, this disclosure also provides an electronic device. The electronic device of this disclosure includes at least one processor and at least one memory electrically connected to the processor. The memory is electrically connected to the processor, wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method described above.

[0066] It should be noted that the electrical connection between the above-mentioned units does not necessarily mean the connection between lines. The indirect connection method can be applied to the embodiments of this disclosure as long as it achieves the purpose of this disclosure.

[0067] Based on the same inventive concept, this disclosure also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the above method.

[0068] Although the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. An automated testing system based on parametric model data processing, characterized in that, include: The data acquisition module is used to determine the test object and the test target, identify the key parts of the test object, determine the detection sensor based on the test target, deploy the detection sensor based on several key parts, and acquire several real-time acquisition parameters based on the detection sensor. A data processing module, connected to the data acquisition module, is used to receive several real-time acquisition parameters, and to perform outlier removal, estimation and filling, and normalization on the several real-time acquisition parameters to obtain several processing parameters. A parameter modeling module, connected to the data processing module, is used to determine the parameter characteristics of the processing parameters, select several processing models based on the parameter characteristics, fuse the several processing models into a model, and optimize the parameters of the fused model in real time through a dynamic model update mechanism. The testing module, connected to the parameter modeling module, is used to obtain anomaly detection and fault location based on the fusion model through real-time parameter comparison and multi-parameter causal analysis. The testing module includes: An anomaly detection unit is used to compare several real-time parameters with the threshold range output by the fusion model to determine anomaly parameters, determine the sampling point distribution corresponding to the anomaly parameters, and determine anomalies based on the sampling point distribution judgment results. The fault location unit is connected to the anomaly detection unit and is used to trace the root cause of abnormal parameters based on the parameter cause-effect graph determined by parameter modeling in order to determine the fault location. The results visualization module is connected to the testing module to display the test results and generate feedback reports based on the test results.

2. The automated testing system based on parametric model data processing according to claim 1, characterized in that, The data acquisition module includes: A part identification unit is used to identify the structural features of the object under test and obtain historical fault data of the object under test, and to determine the key parts based on the structural features and the historical fault data; The sensor determination unit is used to match the parameter type of the target under test with the detection parameter type of the sensor in the preset sensor database, and determine the detection sensor based on the matching result; The deployment unit is connected to the part identification unit and the sensor determination unit to deploy the detection sensor based on the key part.

3. The automated testing system based on parametric model data processing according to claim 2, characterized in that, The location identification unit includes: Extract sub-units to extract the coordinates of high-stress or vulnerable areas of the test object through finite element simulation analysis; An identification subunit, connected to the extraction subunit, is used to determine the statistical characteristics of historical fault locations based on several historical fault data, and to generate a fault priority list based on the statistical characteristics using a clustering algorithm. The location determination subunit is connected to the extraction subunit and the identification subunit to determine the critical location based on the coordinates of the high-stress area or the vulnerable area and the fault priority list.

4. The automated testing system based on parametric model data processing according to claim 1, characterized in that, The data processing module includes: The removal unit is used to identify outliers in the real-time acquired parameters using the isolated forest algorithm, so as to remove the outliers; The data imputation unit is used to select the appropriate model for data imputation based on time series correlation analysis. The processing unit is used to identify the parameter type corresponding to the real-time acquired parameters, and to match the parameter type with a preset normalization method to obtain the processing parameters.

5. The automated testing system based on parametric model data processing according to claim 1, characterized in that, The parameter modeling module includes: The feature determination unit is used to analyze the data characteristics and physical laws of the processing parameters, select a modeling method based on the data characteristics and physical laws, and then determine the processing model. A fusion unit, connected to the feature determination unit, is used to perform fusion modeling on the processing model based on a weighted fusion method to obtain a fusion model; A parameter optimization unit, connected to the fusion unit, is used to determine the full training cycle and update the fusion model based on updated test data to optimize the parameters of the fusion model.

6. The automated testing system based on parametric model data processing according to claim 5, characterized in that, The parameter optimization unit includes: A periodicity determination subunit is used to calculate the performance degradation index of the fusion model, and the full training period is calculated based on the performance degradation index. The data integration subunit, connected to the period determination subunit, is used to obtain updated test data based on the full training period, and to mix the latest test data with the historical dataset according to time weights to obtain an updated training dataset. An optimization subunit, connected to the data integration subunit, is used to train the fusion model based on the updated training dataset to optimize the parameters of the fusion model.

7. The automated testing system based on parametric model data processing according to claim 1, characterized in that, The anomaly detection unit includes: The threshold calculation subunit is used to calculate the dynamic upper and lower limit thresholds based on the predicted values ​​output by the fusion model and the statistical distribution of historical data, and to perform linear compensation on the upper and lower limit thresholds according to the ambient temperature to obtain the threshold range. An anomaly determination subunit, connected to the threshold calculation subunit, is used to draw a parameter change graph based on the real-time parameters, mark the abnormal region in the parameter change graph based on the threshold range, determine the distribution of sampling points in the abnormal region, and determine the detected anomaly based on the distribution judgment result.

8. The automated testing system based on parametric model data processing according to claim 1, characterized in that, The fault location unit includes: The causal reasoning subunit is used to load the parametric causal graph generated during the parametric modeling process, trace the abnormal propagation path in reverse along the edge weights, eliminate paths that violate physical laws, calculate the posterior probability of each node based on the Bayesian network, and locate the root cause of the component whose failure probability is greater than the preset failure probability. The impact assessment subunit is connected to the causal reasoning subunit to predict the fault development trend through the LSTM network, output the remaining effective time and confidence interval, and calculate the fault propagation index.

9. The automated testing system based on parametric model data processing according to claim 1, characterized in that, The results visualization module includes: The drawing unit is used to dynamically display the health status of key parts of the test object through color mapping and overlay finite element simulation stress cloud map; The interactive analysis unit, connected to the drawing unit, is used to set multi-layer switching display modes, including parameter trend graphs, spectrum analysis graphs and causal tracing path graphs, set custom time window scaling, and export local data to generate PDF / CSV format reports. The report generation unit is connected to the interactive analysis unit to automatically match the historical case library based on the fault location results and generate a decision report that includes maintenance steps, spare parts list and cost estimate. The closed-loop feedback unit is used to send maintenance records and verification data back to the parameter modeling module, triggering the incremental learning process of the model.

Images