Virtual auxiliary troubleshooting return-to-zero method for large aircraft test scene

Abstract

The invention discloses a virtual auxiliary troubleshooting return-to-zero method for a large aircraft test scene. According to the virtual auxiliary troubleshooting return-to-zero method, multiple fault information is supplemented for field testers by using a virtual test technology, a return-to-zero mechanism based on man-machine interaction is constructed, and a troubleshooting task of a large aircraft test environment is efficiently assisted. The virtual auxiliary troubleshooting return-to-zero method comprises the following three steps of: firstly, screening fault performance parameters by combining a directed propagation relation of the performance parameters with an incidence relation between a fault mode and a measuring point anomaly; based on an LRU-level performance parameter fault injection interface, parameter offset is injected into fault performance parameters, and reproduction of the faults in the virtual simulation environment is completed; and finally, through a high-parallel virtual isolation test experiment, a fault correlation evaluation mechanism based on response cosine similarity is adopted to assist field troubleshooting personnel to realize rapid attribution of LRU-level faults.

Description

Technical Field

[0001] The present invention relates to a test process optimization method, in particular to a virtual auxiliary troubleshooting and zeroing method for large aircraft test scenarios in discrete industrial scenarios. Background Art

[0002] For discrete industry products, the functional testing tasks in the assembly phase and the performance testing tasks in the use phase are generally presented in the form of a measurement point verification process under fixed test steps. For example: testing the flap control function of a large aircraft is essentially to give an excitation signal through the joystick, observe whether the control surface is deflected within the specified range, and then determine whether there is a fault in the components of the control link unit where the elevator is located. Therefore, similar to the above test steps, the core of the test item test steps is to give a test basis such as the joystick offset range, the object of the measurement point is the deflection angle of the elevator, and the abnormal handling situation is to troubleshoot and repair the various components of the control link unit where the elevator is located.

[0003] At present, in the functional testing of large aircraft assembly stage and the performance testing of the use stage, the fault troubleshooting and zeroing methods mainly include:

[0004] (1) Manually design test items and test steps, and on-site troubleshooters analyze the cause of the failure based on their experience and test results. Due to the large number of test steps and the difficulty in defining component failure modes, traditional manual analysis cannot guarantee the optimality of test design in actual engineering environments. In addition, test step anomalies caused by poor performance of different functional components still require troubleshooting reset, component replacement, component retesting, and other operations outside the test process. On the one hand, the traditional troubleshooting process based on physical inspection increases the number of times each component of the product is run during the production test phase, which has a negative impact on the service life of the components after leaving the factory; on the other hand, the testing and loading and unloading operations during the troubleshooting process also increase the execution time of testers, resulting in a deterioration in cost and efficiency during the assembly-subassembly-final assembly process.

[0005] (2) Semi-physical simulation testing, which embeds the physical object into the software environment for real-time simulation testing. Although semi-physical simulation testing reduces the cost compared to the above-mentioned manual design of test items and full-process physical testing, as a real-time simulation system, semi-physical simulation testing still depends on the running speed of the physical object in terms of test time, and is relatively limited in reducing time costs. In addition, since the physical object is involved in the testing process, the factory life of the physical object will still be affected to a certain extent.

[0006] (3) A fault-finding method based on machine learning algorithms. This method analyzes historical fault data and establishes a fault diagnosis model to assist troubleshooters in fault diagnosis. Machine learning algorithms usually rely on the quality and quantity of historical data when processing data and making predictions. If there are no samples related to occasional mechanism failures in the historical data,

[0007] The fault diagnosis model is unable to identify the characteristics of such faults, resulting in blind spots in practical applications.

[0008] In the universal test scenario, at the mechanism level, all test operations can be implemented in a virtual environment through modeling and simulation (such as the case of the flap test function mentioned above). Although the virtual simulation test model cannot be directly substituted into the real test process of the engineering environment to directly interact with the physical equipment, it can replace the physical equipment to go through the test process, and can reproduce the fault phenomenon and troubleshoot the fault performance parameters through the flexible and autonomous fault injection mechanism in the virtual environment, thereby reducing the time cost and improving the efficiency of test troubleshooting and zeroing. Therefore, combined with the specific discrete industrial scenario of large aircraft assembly, in view of its long and expensive test process, the negative impact of actual testing on the life of components, and the deterioration of cost and efficiency in the assembly-subassembly-final assembly process, a virtual assisted troubleshooting and zeroing method is constructed, so as to assist on-site troubleshooting personnel in fault zeroing at a lower cost and working hours, and effectively solve the above-mentioned inefficient and expensive test execution problems. Summary of the invention

[0009] The present invention provides a virtual auxiliary troubleshooting and zeroing method to solve the negative impact on component life in actual large aircraft testing, as well as the cost and efficiency deterioration caused by component assembly-subassembly-final assembly during testing. The method can assist testers in achieving fault reproduction and zeroing in a virtual environment, thereby effectively solving the problem of inefficient and costly test execution and improving the efficiency of the test process.

[0010] The virtual assisted troubleshooting and zeroing method is mainly aimed at large aircraft test scenarios in discrete industries. Specifically, the virtual assisted troubleshooting and zeroing method includes three core steps: fault performance parameter screening, interactive fault hypothesis verification, and replaceable unit (Line Replaceable Unit, LRU) level fault attribution.

[0011] The screening of fault performance parameters is carried out by reversely determining the fault-related performance parameters according to the fault location through the directed propagation relationship of each performance parameter in the virtual simulation mechanism model. The interactive fault hypothesis verification assists the tester in performing comparative verification of the hypothesis of the fault cause at the sensor signal level through the LRU-level performance parameter fault injection interface, thereby achieving the reproduction of the fault phenomenon in the virtual simulation link. The LRU-level fault attribution is achieved through the high-parallel virtual isolation test of the fault simulation module in the case where multiple performance parameter offsets occur during the reproduction process and the source of the LRU-level fault cannot be directly located.

[0012] The characteristics of the present invention are:

[0013] (1) The present invention constructs a virtual simulation mechanism model to replace the actual physical component to undergo the test process, thereby reducing the negative impact of actual physical testing on component life and labor costs, and improving test efficiency;

[0014] (2) The present invention screens performance parameters reflecting functional component failures through the directed propagation relationship between performance parameters and fault locations in the virtual simulation mechanism model, and designs an interactive fault hypothesis verification human-computer interaction interface based on the LRU-level performance parameter fault injection interface to achieve the reproduction of fault phenomena in the virtual simulation link;

[0015] (3) The present invention uses a virtual isolation test method to evaluate the correlation of the fault performance parameter offsets with the fault phenomenon one by one in the case where multiple performance parameter offsets occur during the fault reproduction process and the source of the LRU-level fault cannot be directly located, thereby assisting on-site troubleshooting personnel to quickly locate components with abnormal performance and achieve LRU-level troubleshooting zeroing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Virtual assisted troubleshooting and zeroing method steps for large aircraft test scenarios

[0017] Figure 2 Schematic diagram of the directional propagation of performance parameters in the virtual simulation mechanism model

[0018] Figure 3 Prototype of an interactive interface for mining the relationship between fault modes and measurement point anomalies

[0019] Figure 4 Auxiliary zeroing mechanism based on virtual isolation test DETAILED DESCRIPTION

[0020] The virtual assisted zeroing method provided by the present invention is described in detail below with reference to the accompanying drawings.

[0021] The present invention provides a virtual auxiliary troubleshooting and zeroing method for large aircraft test scenarios. Figure 1As shown, the virtual assisted troubleshooting and zeroing method for large aircraft test scenarios includes three core steps: fault performance parameter screening, interactive fault hypothesis verification, and LRU-level fault attribution.

[0022] 1. Fault performance parameter screening steps

[0023] The fault performance parameter screening step can reversely determine all performance parameters related to the fault manifestation through the directed propagation relationship between the fault manifestation and each performance parameter in the virtual simulation mechanism model, as shown in the attached figure. Figure 2 As shown. The failure mode is defined by performance parameters, specifically defined as a significant abnormality in which the performance parameter has a positive / negative offset of more than 25% relative to the theoretical value; the measurement point and its judgment criteria are defined by the actual product sensor signal, specifically defined as the judgment result of whether the sensor signal stability value is based on the theoretical threshold. The correlation between the failure mode and the measurement point abnormality is constructed by the test designer through FMECA and other test data, and the basic qualitative fault propagation assumption needs to be met: in the test, if the failure mode occurs, the measurement point related to the failure mode in the correlation relationship will definitely show an abnormal state; conversely, if the measurement point shows an abnormal state, a failure mode related to the measurement point must occur.

[0024] The main steps of the fault performance parameter screening step are as follows:

[0025] 1) Based on the correlation between the fault mode and the abnormal measuring point, according to the current abnormal measuring point status, the fault modes related to the abnormal measuring point are positively sorted out to construct a preliminary set of suspected fault modes;

[0026] 2) Based on the correlation between the fault mode and the abnormal measurement point, the fault modes related to the normal measurement points are filtered out from the preliminary suspected fault mode set according to the normal measurement points that have passed, and a screened suspected fault mode set is constructed.

[0027] 2. Interactive Fault Hypothesis Verification Steps

[0028] The interactive fault hypothesis verification step assists the tester in performing a comparative verification of the hypothesis of the fault cause at the sensor signal level through a fault injection interface based on the LRU level performance parameter, thereby reproducing the observed fault phenomenon in the virtual simulation link. The human-computer interaction interface of the interactive fault hypothesis verification step is shown in the attached figure. Figure 3 shown.

[0029] Attached Figure 3The functional unit-measurement point abnormal fault entity relationship diagram on the left side further summarizes the fault modes corresponding to the performance parameters into LRU-level faults through the correspondence between the performance parameters and the LRU-level components; further, through the positive-abnormal possibility of the measurement points and the faults, the reasoning results of the fault performance parameter screening step are intuitively presented to the testers, supporting the testers to understand the brief reasoning process of the LRU-level faults that may cause the abnormal state of the measurement points.

[0030] And attached Figure 3 The interactive panel on the right lists the fault modes involved in specific abnormal measurement points based on the inference results of the fault performance parameter screening step; the tester injects the corresponding performance parameter offset in proportion to the theoretical value through the fault injection panel, and the fault simulation module further connects the parameter offset configuration to the simulation model. After obtaining the simulation results, the response action of the object during the test is reproduced through the 3D visualization panel, which supports the tester to compare the simulation results with the actual fault performance, and gradually complete the fault location and reproduction.

[0031] 3. LRU level fault attribution steps

[0032] The LRU-level fault attribution step is implemented through a high-parallel virtual isolation test of the fault simulation module to address the situation where multiple performance parameter offsets occur during the reproduction process and the source of the LRU-level fault cannot be directly located. Figure 4 As shown, the main steps are as follows:

[0033] 1) Determine whether the multiple performance parameter offsets that occur during the recurrence process all belong to the same LRU object. If so, directly attribute the fault to the LRU object. Otherwise, proceed to step (2);

[0034] 2) The high-parallel virtual isolation test of the fault simulation module masks the performance parameters of each offset in the recurrence state (i.e., keeps the offsets of other performance parameters unchanged and sets the offset of the performance parameter to be studied to 0), thereby forming a corresponding fault injection configuration for different performance parameters;

[0035] 3) The fault simulation module simulates the test configuration corresponding to each performance parameter through a highly parallel architecture, and uses the absolute value of the cosine similarity of the response timing before and after shielding to represent the performance parameter offset to measure the degree of irrelevance of the corresponding fault mode with the current measurement point abnormality. The lower the cosine similarity of the response timing before and after shielding, the higher the relevance of the shielded performance parameter to the fault phenomenon, and the higher the attribution probability of the performance parameter.

[0036] 4) By evaluating the correlation between all performance parameter offsets and faults, the correlation levels are displayed on the interactive interface in order from low to high, thereby providing quantitative auxiliary information for on-site troubleshooting personnel's LRU-level physical troubleshooting inspection operations, allowing on-site troubleshooting personnel to more effectively locate the specific LRU object with performance abnormalities.

Claims

1. A virtual auxiliary troubleshooting and zeroing method for large aircraft test scenarios, characterized by: The virtual assisted troubleshooting and zeroing method includes a fault performance parameter screening step of reversely determining the preliminary troubleshooting scope through the performance parameter-fault location relationship, an interactive fault hypothesis verification step of assisting testers in verifying and reproducing the fault phenomenon in a virtual simulation environment through an LRU-level performance parameter fault injection interface and interface, and an LRU-level fault attribution step of evaluating the degree of relevance of each fault performance parameter to the fault phenomenon through a highly parallel virtual isolation test.

2. The virtual auxiliary troubleshooting and zeroing method according to claim 1 is characterized in that: The fault performance parameter screening step in the virtual assisted troubleshooting and zeroing method defines the fault mode as a positive / negative deviation of the performance parameter relative to the theoretical value of more than 25%, and defines the positive / abnormality of the measuring point as whether the stable value of the sensor signal is within the theoretical threshold. According to the current positive / abnormal measuring point state combined with the propagation relationship of the performance parameters, the fault modes related to the abnormal measuring points are positively sorted out to construct a set of suspected fault modes that have been screened.

3. The virtual auxiliary troubleshooting and zeroing method according to claim 1 is characterized in that: The interactive fault hypothesis verification step in the virtual assisted troubleshooting and zeroing method further summarizes the fault mode corresponding to the performance parameter into the LRU level fault according to the corresponding relationship between the performance parameter and the LRU level component; through the fault injection panel, the corresponding performance parameter offset is injected in proportion to the relative theoretical value, and the simulation results are displayed through three-dimensional visualization to gradually complete the reproduction of the fault.

4. The virtual auxiliary troubleshooting and zeroing method according to claim 1 is characterized in that: The LRU-level fault attribution step in the virtual assisted troubleshooting and zeroing method conducts a highly parallel virtual isolation test for different performance parameters, and evaluates the correlation between the performance parameter offset and the fault mode based on the cosine similarity of the response before and after the performance parameter is shielded; the evaluation results provide quantitative auxiliary information for on-site troubleshooting personnel to effectively locate the specific LRU object with performance abnormalities.

Images