Automatic detection system for intelligent instrument of two-wheeled vehicle
Patent Information
- Application Number
- CN202511102348.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-07
- Publication Date
- 2025-11-14
Smart Images
Figure CN120947715A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of instrument testing, and in particular to an automated testing system for intelligent instruments of two-wheeled vehicles. Background Technology
[0002] The necessity of automated testing for smart instruments on two-wheeled vehicles is becoming increasingly apparent. Traditional manual testing is inefficient, susceptible to subjective factors, and struggles to guarantee the accuracy and consistency of instrument functions. As the complexity of smart instruments increases, manual testing can no longer cover all interaction scenarios. Automated testing, through pre-programmed procedures, can achieve high-precision, high-efficiency, and uninterrupted batch testing, greatly improving product quality and production efficiency while reducing labor costs and the risk of missed inspections.
[0003] Current technology for smart meters for two-wheeled vehicles is still in a semi-manual stage. Test cases are mostly determined based on human experience, making it impossible to select test cases according to the characteristics of each meter. This results in problems such as long testing time and difficulty in achieving effective coverage. Summary of the Invention
[0004] To address the problem of low detection efficiency in existing technologies, this invention provides an automated detection system for intelligent instruments on two-wheeled vehicles.
[0005] This invention is achieved through the following technical solution: An automated detection system for intelligent instruments on two-wheeled vehicles, comprising: This includes a fault sample identification module, an injected fault package identification module, a physical instrument detection module, a comprehensive detection score calculation module, and a detection result statistics and output module. Specifically: The fault sample determination module clusters the fault causes based on the GNN network model, dividing the faults into different fault clusters; The injection fault package determination module determines the injection fault package based on a digital twin; The physical instrument detection module uses the injected fault package output by the injected fault package determination module to perform physical instrument detection and obtain detection data. The detection comprehensive score calculation module calculates the detection comprehensive score based on the detection data using the entropy weight-TOPSIS fusion algorithm. The test result statistics and output module is used to statistically analyze the test results of the same batch of products, including the detection rate, fault distribution rate, and misjudgment rate of each type of fault, and output the results.
[0006] Furthermore, the clustering of fault causes based on the GNN network model includes the definition of nodes, the definition of edges, and the determination of edge weights in the GNN network model; the nodes include fault features and environmental parameters, and the edges include electrical supply edges, signal transmission edges, and environmental coupling edges; the determination of edge weights includes fault propagation time constraints.
[0007] Furthermore, the edge weights Determined as:
[0008] in, The time delay sensitivity coefficient, This refers to the actual transmission time. For the time of theoretical dissemination, The physical time constraint window is expressed as follows: , in, and Minimum and maximum fault propagation times, respectively. 。
[0009] Furthermore, the step of determining the injected fault package based on the digital twin includes establishing a structured fault injection library based on the fault cluster classification results. The structured fault injection library is divided into multiple partitions, and the faults are managed by partition according to fault clusters and injection methods.
[0010] Furthermore, the step of determining the injected fault package based on the digital twin also includes simulating faults in the fault injection library on the digital twin, obtaining fault simulation results, using the obtained simulation results as input to the Transformer model, and outputting the determined injected fault package, which includes the fault test cases to be tested and the number of tests.
[0011] Furthermore, the objective function of the Transformer model is constructed based on a multi-objective optimization function. The objectives include fault coverage, test time, and equipment wear and tear, and the expression is as follows:
[0012] in, Let x be the objective function, and x be the configuration parameter for the injected fault package. For dynamic weights, For fault coverage, test time, and equipment wear and tear. These are constraint terms.
[0013] Furthermore, the constraints include injection equipment capability constraints, temperature constraints, test time constraints, equipment lifespan constraints, and cost budget constraints.
[0014] Furthermore, the expression for the constraint term is as follows: For adaptive penalty coefficient, The i-th constraint includes the injection equipment capability constraint, temperature constraint, test time constraint, equipment lifespan constraint, and cost budget constraint. The expression is as follows:
[0015] in, The adaptive penalty coefficient at time t+1 Let be the adaptive penalty coefficient at time t.
[0016] Furthermore, the entropy weight-TOPSIS fusion algorithm calculates the comprehensive detection score based on the detection data by objectively determining the weight of each detection index based on the entropy weight method, and further calculating the degree of closeness between each detection data and the ideal solution based on the TOPSIS method, and obtaining the comprehensive score based on the degree of closeness.
[0017] Furthermore, the calculation of weights based on the entropy weight method includes calculating the information entropy of the index and calculating the weights based on the information entropy; The information entropy of the j-th indicator is :
[0018] in, represents the contribution of the i-th sample under the j-th indicator; m is the number of samples. The weight of the j-th indicator :
[0019] in, Let be the difference coefficient of the j-th indicator.
[0020] Compared with existing technologies, the beneficial effects of this invention are as follows: Based on artificial intelligence methods, the detected fault samples are determined, and by separating the characteristics of different faults, fault tracing is achieved, solving the problem that traditional detection methods only detect normal conditions and cannot reproduce faults. Based on digital twins, the injected fault packages are dynamically determined, achieving precise targeting of real-time operating conditions and automatically generating high-value fault combinations, avoiding the inefficiency of manual exhaustive searching. Extreme cascading failures are simulated in a virtual environment, exposing system vulnerabilities in advance. Simultaneously, fault parameters are optimized based on twin feedback loops, approximating the actual failure evolution, improving test coverage, and increasing the detection rate of hidden faults. Based on the entropy weight-TOPSIS fusion algorithm, a comprehensive detection score is calculated based on the detection data. The entropy weight method dynamically allocates index weights through the data's own dispersion, avoiding subjective bias. TOPSIS quantifies the relative merits of each scheme based on dual benchmarks of positive and negative ideal solutions. The combination of these two methods ensures the objectivity of weight allocation and comprehensively reflects the global proximity of the detection data to the optimal / worst standards through geometric distance, making the comprehensive score both data-driven and consistent with decision-making logic, significantly improving the scientific nature of the evaluation and the interpretability of the results in complex detection scenarios. Attached Figure Description
[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of an automated detection system for a smart instrument panel of a two-wheeled vehicle according to an embodiment of this application; Figure 2 This is a schematic diagram illustrating the determination of the injected fault package based on a digital twin according to an embodiment of this application; Figure 3 This is a schematic diagram illustrating the process of calculating the comprehensive detection score according to an embodiment of this application. Detailed Implementation
[0022] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
[0023] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0024] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. The illustrations only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the shape, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.
[0025] See Figure 1 An automated detection system for intelligent instruments on two-wheeled vehicles includes the following modules: The fault sample determination module clusters fault causes based on the GNN network model, dividing faults into different fault clusters.
[0026] In existing technologies, fault settings during detection are based on the experience of the testing personnel. Human experience cannot cover complex, coupled faults, such as display delays caused by CAN communication interference, leading to a high false negative rate. Furthermore, there are efficiency bottlenecks; experience-driven serial testing results in long single-instrument testing times, and 70% of testing resources are wasted on repetitive routine items. Therefore, this invention uses artificial intelligence methods to determine fault samples for detection, thereby improving detection efficiency and accuracy. The specific implementation includes the following steps: S11: Obtain the production line's inspection results and user maintenance history data; S12: Determine the total fault database based on the acquired data. The total fault database includes the types of faults that have occurred in history, the fault rate, and the time when the fault occurred. S13: Clustering of fault causes based on a GNN network model, classifying faults into different fault clusters, and subsequently determining the type and quantity of fault samples to be detected based on the fault cluster category; the specific implementation method is as follows: (1) Establish a GNN network model with embedded fault propagation time constraints, including the definition of nodes, edge definitions and edge weights of the GNN network model.
[0027] The fault propagation time refers to the time delay from the occurrence of the fault root cause to the occurrence of the final observable symptoms. For example, the propagation time from the failure of the capacitor on the power supply pin of the instrument MCU to the voltage drop and then to the screen flickering is 350ms.
[0028] The nodes of the GNN network model include fault characteristics and environmental parameters; The fault characteristics include voltage and current timing signals of key components; Optionally, the key components include an MCU, a power supply, a display module, and a sensor; The environmental parameters include temperature, humidity, and vibration.
[0029] The edge types of GNN network models include electrical supply edges, signal transmission edges, and environmental coupling edges.
[0030] The electrical supply side relates to the relationship between each chip and the power supply chip; the signal transmission side relates to the signal transmission between the MCU and the sensor and display module; the thermal coupling side relates to the relationship between environmental parameters and each chip.
[0031] Create edges with fault propagation time constraints, where the edge weights are defined as follows:
[0032] in, The time delay sensitivity coefficient, This refers to the actual transmission time. For the time of theoretical dissemination, The physical time constraint window is expressed as follows:
[0033] (2) Input the collected fault characteristics and environmental parameters into the GNN network model; (3) Obtain the fault cluster classification results; Optionally, the fault cluster includes electrical faults, communication faults, and environmental faults.
[0034] The fault clustering method of this invention enables fault location based on the root cause. By separating the characteristics of different faults, it achieves fault tracing and solves the problem that traditional detection methods can only detect normal conditions and cannot reproduce faults.
[0035] The injection fault package determination module, which determines the injection fault package based on a digital twin, is implemented through the following steps: S21: Digital Twin Creation The creation of a digital twin is a complete mapping process from the physical world to the virtual space, and its core inputs include two parts: physical world data input and virtual model construction.
[0036] Regarding physical world data input, the system needs to acquire voltage and current signals of key components in real time. For example, it needs to obtain electrical parameters such as current waveforms and voltage fluctuations of instruments through high-precision sensors to ensure that the data sampling frequency meets the requirements of dynamic simulation. At the same time, environmental parameters (such as temperature, humidity, and vibration intensity) also need to be monitored synchronously to reflect the actual working conditions of the instruments.
[0037] Construct a virtual model, including a 3D digital model of the instrument display, to accurately reproduce the instrument panel structure, pointer dynamics, or digital display and other visual elements. Simultaneously, define signal input points (such as CAN bus interfaces and voltage sampling points) to map physical world sensor data to corresponding nodes in the virtual model.
[0038] S22: Establish a structured fault injection library based on fault cluster classification results.
[0039] The structured fault injection library is divided into multiple partitions, and faults are managed according to fault clusters and injection methods. The injection methods include hardware injection and software injection. Hardware injection is for faults that require an external fault injection device to trigger, while software injection is for faults that can be simulated by software.
[0040] By partitioning the fault injection library, we can quickly retrieve and locate fault clusters and make flexible expansion possible. When a new fault mode needs to be added, we only need to add an entry to the corresponding partition without affecting the global architecture and reducing the redundancy of test cases.
[0041] S23: Dynamically Determining Injected Fault Packages Based on Digital Twins
[0042] For instrument panels of different models of two-wheeled vehicles, due to differences in composition and function, different detection strategies are adopted when performing fault detection. It is necessary to determine the injected fault package by combining the faults in different fault clusters. This invention, based on the characteristics of digital twins, uses the Transformer model to dynamically determine the injected fault package. The process includes: The faults in the fault injection library are simulated on the digital twin to obtain the fault simulation results, which include the fault rate, fault time, and components involved in various types of faults.
[0043] The obtained simulation results are used as input to the Transformer model, and the output is the determined fault package, which includes the fault test cases to be tested and the number of tests.
[0044] Furthermore, the objective function of the Transformer model is constructed based on a multi-objective optimization function. Specifically, the objectives include fault coverage, test time, and equipment wear and tear, expressed as follows:
[0045] Where x represents the configuration parameter for the injected fault package. For dynamic weights, For fault coverage, test time, and equipment wear and tear. These are constraint terms.
[0046] The dynamic weight update method is as follows:
[0047] in, The dynamic weights at time t+1 , The dynamic weights at time t For learning rate, , The improvement rate for the target , The standard deviation is denoted as .
[0048] The constraints consider injection equipment capacity constraints, temperature constraints, test time constraints, equipment lifespan constraints, and cost budget constraints. The expressions for the constraints are as follows: For adaptive penalty coefficient, The i-th constraint includes the injection equipment capability constraint, temperature constraint, test time constraint, equipment lifespan constraint, and cost budget constraint. The expression is as follows:
[0049] in, The adaptive penalty coefficient at time t+1 Let be the adaptive penalty coefficient at time t.
[0050] By dynamically adjusting the constraint terms based on the adaptive penalty coefficient, a light exploration is allowed in the early stage, and strict convergence is enforced in the later stage, thereby improving optimization efficiency and solution quality.
[0051] This invention uses digital twins to dynamically determine injected fault packages, enabling precise targeting of real-time operating conditions and automatic generation of high-value fault combinations, avoiding the inefficiency of manual exhaustive searching. By simulating extreme cascading failures in a virtual environment, system vulnerabilities are exposed in advance. Simultaneously, fault parameters are optimized based on twin feedback loops to approximate the actual failure evolution, improving test coverage and increasing the detection rate of latent faults.
[0052] The physical instrument detection module is used to perform physical instrument detection by using the injected fault package output by the injected fault package determination module to obtain detection data. The detection comprehensive score calculation module is used to calculate the detection comprehensive score based on the detection data using the entropy weight-TOPSIS fusion algorithm. Specifically, it objectively determines the weight of each detection index based on the entropy weight method, and further calculates the degree of closeness between each detection data and the ideal solution based on the TOPSIS method. The comprehensive score is obtained based on the degree of closeness. The implementation process includes the following steps: S41: Acquire the detection data and perform data standardization processing to obtain the standardization matrix. This is to eliminate the influence of different indicator units and orders of magnitude, making the data comparable. Here, m is the sample size, and n is the number of indicators.
[0053] S42: Calculate weights based on the entropy weight method; This includes calculating the information entropy of the indicators and calculating the weights based on the information entropy; The information entropy of the j-th indicator is :
[0054] in, denoted as , where is the contribution of the i-th sample under the j-th indicator; m is the number of samples.
[0055] The weight of the j-th indicator :
[0056] in, Let be the difference coefficient of the j-th indicator. .
[0057] S43: Constructing the weighted normalization matrix V:
[0058] S44: Calculate distance and proximity based on TOPSIS
[0059] The ideal solution is obtained based on the weighted normalization matrix. and negative ideal solution A positive ideal solution is composed of the maximum value of each column element, while a negative ideal solution is composed of the minimum value of each column element.
[0060]
[0061] The distances from sample i to the positive ideal solution and the negative ideal solution are respectively:
[0062] The degree of proximity is expressed as follows: The closer to 0, the better the evaluation object.
[0063] S45: Calculate the overall test score C
[0064] The test result statistics and output module is used to statistically analyze the test results of the same batch of products, including the detection rate, fault distribution rate, and false positive rate of each type of fault, and output the results.
[0065] In this implementation, automated detection of smart instruments for two-wheeled vehicles is achieved. Fault samples are identified using artificial intelligence methods, and fault tracing is achieved by separating the characteristics of different faults, solving the problem of traditional detection methods where individual faults are detected but cannot be reproduced. Based on digital twins, the injected fault packages are dynamically determined, enabling precise targeting of real-time operating conditions and automatic generation of high-value fault combinations, avoiding the inefficiency of manual exhaustive searching. Extreme cascading failures are simulated in a virtual environment to expose system vulnerabilities in advance. Simultaneously, fault parameters are optimized based on twin feedback loops to approximate the actual failure evolution, improving test coverage and increasing the detection rate of hidden faults. An entropy-weighted TOPSIS fusion algorithm is used to calculate a comprehensive detection score based on the detection data. The entropy-weighted method dynamically allocates index weights based on the data's inherent dispersion, avoiding subjective bias. TOPSIS quantifies the relative merits of each solution based on dual benchmarks of positive and negative ideal solutions. The combination of these two methods ensures the objectivity of weight allocation and comprehensively reflects the global proximity of the detection data to the optimal / worst standards through geometric distance, giving the comprehensive score both data-driven characteristics and consistency in decision-making logic, significantly improving the scientific rigor and interpretability of the evaluation in complex detection scenarios.
[0066] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.
Claims
1. An automated detection system for intelligent instruments on two-wheeled vehicles, characterized in that, This includes a fault sample identification module, an injected fault package identification module, a physical instrument detection module, a comprehensive detection score calculation module, and a detection result statistics and output module. Specifically: The fault sample determination module clusters the fault causes based on the GNN network model, dividing the faults into different fault clusters; The injection fault package determination module determines the injection fault package based on a digital twin; The physical instrument detection module uses the injected fault package output by the injected fault package determination module to perform physical instrument detection and obtain detection data. The detection comprehensive score calculation module calculates the detection comprehensive score based on the detection data using the entropy weight-TOPSIS fusion algorithm. The test result statistics and output module is used to statistically analyze the test results of the same batch of products, including the detection rate, fault distribution rate, and misjudgment rate of each type of fault, and output the results.
2. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 1, characterized in that, The clustering of fault causes based on the GNN network model includes the definition of nodes, the definition of edges, and the determination of edge weights in the GNN network model; the nodes include fault characteristics and environmental parameters, and the edges include electrical supply edges, signal transmission edges, and environmental coupling edges; the determination of edge weights includes fault propagation time constraints.
3. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 2, characterized in that, The edge weight Determined as: in, The time delay sensitivity coefficient, This refers to the actual transmission time. For the time of theoretical dissemination, The physical time constraint window is expressed as follows: ， in, and Minimum and maximum fault propagation times, respectively. 。 4. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 1, characterized in that, The method of determining the injected fault package based on the digital twin includes establishing a structured fault injection library based on the fault cluster classification results. The structured fault injection library is divided into multiple partitions, and the faults are managed by partition according to fault clusters and injection methods.
5. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 4, characterized in that, The method of determining the injected fault package based on the digital twin also includes simulating faults in the fault injection library on the digital twin, obtaining fault simulation results, using the obtained simulation results as input to the Transformer model, and outputting the determined injected fault package, which includes the fault test cases to be tested and the number of tests.
6. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 5, characterized in that, The objective function of the Transformer model is constructed based on a multi-objective optimization function. The objectives include fault coverage, test time, and equipment wear and tear, and the expression is as follows: in, Let x be the objective function, and x be the configuration parameter for the injected fault package. For dynamic weights, For fault coverage, test time, and equipment wear and tear. These are constraint terms.
7. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 6, characterized in that, The constraints include injection equipment capability constraints, temperature constraints, test time constraints, equipment lifespan constraints, and cost budget constraints.
8. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 7, characterized in that, The constraint term expression is as follows: For adaptive penalty coefficient, The i-th constraint includes the injection equipment capability constraint, temperature constraint, test time constraint, equipment lifespan constraint, and cost budget constraint. The expression is as follows: in, The adaptive penalty coefficient at time t+1 Let be the adaptive penalty coefficient at time t.
9. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 1, characterized in that, The entropy-weighted TOPSIS fusion algorithm calculates the comprehensive detection score based on the detection data by objectively determining the weight of each detection index based on the entropy weight method, and further calculating the degree of closeness between each detection data and the ideal solution based on the TOPSIS method, and obtaining the comprehensive score based on the degree of closeness.
10. The automated detection system for intelligent instruments of two-wheeled vehicles according to claim 9, characterized in that, The calculation of weights based on the entropy weight method includes calculating the information entropy of the index and calculating the weights based on the information entropy. The information entropy of the j-th indicator is : in, represents the contribution of the i-th sample under the j-th indicator; m is the number of samples. The weight of the j-th indicator : in, Let be the difference coefficient of the j-th indicator. .