Method and system for detecting reliability of automobile lamp and medium

Abstract

The invention discloses an automobile lamp reliability detection method and system and a medium. The method for detecting the reliability of the automobile lamp comprises the steps of applying coupled multi-factor environmental stress and a dynamically changing electric signal working state to a to-be-detected automobile lamp in a testing process; acquiring an environmental stress parameter, an electric signal working state parameter and an actual performance parameter of the to-be-tested vehicle lamp in real time; and based on a digital twin model and a machine learning model, according to the environmental stress parameter, the electric signal working state parameter and the actual performance parameter, determining a health state index of the to-be-tested vehicle lamp, and performing fault prediction. According to the method and system for detecting the reliability of the automobile lamp and the medium, multi-factor coupling accelerated aging testing is achieved, dynamic testing of dynamic working conditions is achieved, active prediction and early warning management is achieved, the testing accuracy and efficiency are improved, and predictive management of the automobile lamp is achieved.

Description

Technical Field

[0001] This invention relates to the technical field of vehicle headlight testing, and more particularly to a method, system, and medium for testing the reliability of automotive headlights. Background Technology

[0002] As a crucial component of a vehicle's active safety system, the reliability of automotive lights directly impacts driving safety at night and in adverse weather conditions. With the development of the automotive industry, lighting technology has evolved from traditional halogen and xenon lamps to current LED lights, laser lights, and intelligent adaptive high-beam systems. These advanced lights integrate sophisticated optical structures, complex electronic drive circuits, and intelligent control units, presenting them with even more stringent reliability challenges.

[0003] Currently, the reliability testing of automotive headlights primarily relies on accelerated life testing in a laboratory, simulating single or limited environmental stresses. Common testing methods include high-temperature and high-humidity testing, temperature cycling testing, vibration testing, and salt spray testing. These traditional methods typically have the following limitations: (1) The test conditions are seriously out of sync with real-world operating conditions. Existing technologies often employ sequential or single stress application methods, such as conducting high-temperature tests first, followed by vibration tests. However, in actual driving environments, vehicle lights are simultaneously subjected to the coupled effects of multiple environmental factors, including temperature, humidity, mechanical vibration, electrical fluctuations, and chemical corrosion. This "decoupled" testing method cannot realistically simulate the synergistic effects between stresses, which may lead to the failure to activate and identify key failure modes, resulting in a weak correlation between test results and actual service life.

[0004] (2) The detection methods are subjective, lagging, and not comprehensive enough. Performance evaluation is mostly carried out before and after the test, relying on manual measurement of parameters such as luminous flux and illuminance using equipment such as photometers, and visual observation of appearance changes. This method lacks continuous monitoring data during the test process, and cannot know the specific trajectory of performance degradation and the accurate time point of failure, which is not conducive to root cause analysis of failure. At the same time, manual interpretation is highly subjective, inefficient, and difficult to quantify subtle performance degradation and appearance defects.

[0005] (3) Lack of verification of the system-level interaction capabilities of intelligent vehicle lights. Modern intelligent vehicle lights (such as ADB adaptive high beams) are deeply integrated with advanced driver assistance systems and vehicle networks. Existing testing methods mainly focus on the performance of vehicle lights as independent components, and fail to effectively simulate their dynamic working state in the vehicle's electrical environment, such as frequent switching, brightness adjustment, receiving and executing control commands from the vehicle network, etc., and cannot assess their reliability when working in conjunction with the vehicle system.

[0006] Therefore, this application introduces a method, system, and medium for testing the reliability of automotive headlights. Summary of the Invention

[0007] The purpose of this invention is to provide a method, system, and medium for testing the reliability of automotive headlights. It realizes accelerated aging testing with multi-factor coupling, dynamic testing under dynamic operating conditions, and proactive predictive and early warning management, which improves the accuracy and efficiency of testing and enables predictive management of headlights.

[0008] The objective of this invention is achieved through the following technical solution: In a first aspect, the present invention provides a method for testing the reliability of automotive headlights, comprising: During the test, the test headlights are subjected to coupled multi-factor environmental stress and dynamically changing electrical signal operating states. Real-time acquisition of environmental stress parameters, electrical signal operating status parameters, and actual performance parameters of the vehicle lamp under test; Based on digital twin models and machine learning models, the health status indicators of the vehicle lamp under test are determined and fault prediction is performed according to the environmental stress parameters, electrical signal operating status parameters and actual performance parameters.

[0009] The beneficial effects of the above-mentioned solution are that, by simultaneously applying coupled multi-factor environmental stress and dynamically changing electrical signal operating states, this invention overcomes the shortcomings of traditional testing methods that involve applying stress only once and sequentially. It can more realistically and efficiently reproduce the operating conditions of vehicle lights under complex real-world driving environments, thereby stimulating potential failure modes that are difficult to expose under single stress, greatly improving the accuracy and efficiency of testing. Furthermore, through the collaborative analysis of digital twin models and machine learning models, this invention can not only assess the current health status of vehicle lights but also predict their future performance degradation trends and remaining lifespan based on real-time data, achieving predictive management and elevating quality control from passive response to proactive early warning.

[0010] Furthermore, the coupled multi-factor environmental stress includes a combination of at least two of the following: temperature, humidity, vibration, salt spray, and dust. The dynamically changing electrical signal operating states include at least one of the following operating states: simulated vehicle start-up, load shedding, voltage drop, steady-state fluctuation, frequent switching, brightness adjustment, and beam changes linked to advanced driver assistance systems.

[0011] The beneficial effect of the above solution is that, through a combination of multiple stresses, this invention can comprehensively evaluate the structural integrity, material aging resistance, and stability of the electronic drive system of vehicle lights. Furthermore, this invention specifically defines the types of environmental stresses (temperature, humidity, vibration, salt spray, dust) and electrical stresses (voltage drops, load dumps, beam-linkage changes), including typical and extreme conditions faced by vehicle lights in actual use. Therefore, this method can specifically verify the reliability of vehicle lights under extreme and dynamic operating conditions, ensuring that the test results are highly correlated with actual service life.

[0012] Furthermore, the coupled multi-factor environmental stress and dynamically changing electrical signal operating state applied to the lamp under test include: The multi-factor environmental stress of the coupling and the working state of the electrical signal are controlled by a preset comprehensive test profile. The integrated test profile incorporates signals simulating vehicle network interaction, and these signals follow controller area network, local interconnection network, or vehicular Ethernet protocols.

[0013] The beneficial effects of the above solution are that the present invention controls the process through a preset comprehensive test profile, ensuring the consistency and repeatability of the test process, reducing human error, and is suitable for spot checks on the production line or for consistency comparison of different batches of products.

[0014] Furthermore, the actual performance parameters include at least one of optical performance parameters, electrical performance parameters, and physical state parameters; The optical performance parameters include at least one of luminous flux and color temperature; The electrical performance parameters include at least one of operating current and operating voltage; The physical state parameters include at least one of the following: internal temperature parameters of the lamp body, appearance change parameters of the lamp housing, and appearance change parameters of the lens.

[0015] The beneficial effect of the above solution is that the present invention achieves comprehensive, quantitative, and non-destructive monitoring of vehicle headlight performance degradation (luminous flux, color temperature), functional abnormalities (current, voltage), and physical defects (appearance changes, internal temperature) by collecting three major categories of parameters: optical, electrical, and physical state.

[0016] Furthermore, the step of determining the health status indicators of the vehicle lamp under test and performing fault prediction based on the digital twin model and machine learning model, according to the environmental stress parameters, electrical signal operating state parameters, and actual performance parameters, includes: The environmental stress parameters and electrical signal operating status parameters are input into the digital twin model, and the simulated performance parameters of the vehicle lamp under test are output through simulation. The actual performance parameters are compared with the inferred performance parameters to calibrate the digital twin model; Based on the calibrated digital twin model, the health status index of the vehicle lamp under test is determined according to the environmental stress parameters and electrical signal operating status parameters. Based on the machine learning model, the fault prediction of the vehicle light under test is performed according to health status indicators and performance parameters.

[0017] The beneficial effects of the above scheme are that this invention clarifies that the digital twin model is responsible for simulation and deduction based on physical laws, while the machine learning model is responsible for fault prediction based on data patterns. The two work together, leveraging the advantages of both the physical mechanisms of pattern recognition and data-driven approaches, while also ensuring interpretability of the predictions. Furthermore, by calibrating the digital twin model, it can continuously track the state changes of the physical entity (such as material aging), preventing the model from deviating from reality due to long-term use and ensuring the long-term accuracy of the entire prediction system.

[0018] Furthermore, based on the calibrated digital twin model, the health status indicators of the vehicle lamp under test are determined according to the environmental stress parameters and electrical signal operating status parameters, including: Based on the calibrated digital twin model, simulations are performed according to the preset future test profile to obtain the performance parameter change process of the vehicle lamp under test under the continuous application of the coupled multi-factor environmental stress and the dynamically changing electrical signal working state, thus obtaining the performance degradation trajectory based on physical simulation. Based on the performance degradation trajectory, a state vector reflecting the overall health of the vehicle lights is calculated, and combined with a preset first failure threshold, a preliminary prediction of the remaining useful life is calculated to determine the health status index.

[0019] The beneficial effects of the above solution are that, by using a digital twin model to simulate and extrapolate based on future test profiles, this invention can proactively predict the performance changes (i.e., performance degradation trajectory) of vehicle lights during a complete test cycle or future usage scenarios. This prediction is based on a deep understanding of the physics of failure, rather than simple data extrapolation, making the results more convincing. Furthermore, by calculating the health state vector and preliminary predicted values ​​of remaining useful life, complex simulation data is transformed into concise and intuitive engineering indicators, facilitating engineers' understanding of the vehicle light status and enabling them to make maintenance or replacement decisions.

[0020] Furthermore, the step of predicting the fault of the vehicle lamp under test based on the machine learning model, according to health status indicators and performance parameters, includes: The time-series features extracted from health status indicators and performance parameters are fused with the fault mode features in the historical test database to form a fused feature vector. The fused feature vector is input into a pre-trained machine learning model, and the probability of the fault mode of the test lamp is determined by identifying the similarity between the fused feature vector and the fault mode features.

[0021] The beneficial effect of the above solution is that, by extracting time-series features and matching them with a historical fault database, the machine learning model can not only predict when it will fail, but also determine how it might fail (i.e., the probability of failure mode), providing a key basis for design improvement and precise maintenance.

[0022] Furthermore, the step of predicting the fault of the vehicle lamp under test based on the machine learning model, according to health status indicators and performance parameters, includes: The performance degradation trajectory is input into a pre-trained machine learning model. By learning the changing trend of the performance parameters of the performance degradation trajectory, the remaining useful life and performance parameter prediction values ​​of the vehicle lamp under test are determined. Based on the remaining useful life prediction value and the failure mode probability, the health level announcement information of the vehicle lamp under test is generated; When the predicted value of the performance parameter exceeds the preset second failure threshold within a preset time, a maintenance warning message is generated.

[0023] The beneficial effects of the above solution are that this invention integrates the remaining useful life prediction value, failure mode probability, and second failure threshold judgment to generate health level announcement information and maintenance early warning information, directly transforming technical prediction results into actionable maintenance suggestions, forming a closed loop from monitoring, analysis, prediction to decision-making, greatly improving vehicle safety and operational efficiency. Secondly, this invention provides a detection system for the reliability of automotive headlights, including: The environmental simulation module is used to apply coupled multi-factor environmental stress and dynamically changing electrical signal operating states to the headlights under test during the testing process. The data acquisition module is used to acquire environmental stress parameters, electrical signal operating status parameters, and the actual performance parameters of the vehicle lamp under test in real time. The detection and analysis module is used to analyze the performance degradation trend and predict the fault of the vehicle lamp under test based on the digital twin model and machine learning model, according to the environmental stress parameters, electrical signal operating state parameters and actual performance parameters.

[0024] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by at least one processor, implements the steps of the above-described detection method.

[0025] Compared with the prior art, the beneficial effects of the present invention include at least the following: This invention overcomes the shortcomings of traditional testing methods that apply stress in a single, sequential manner by simultaneously applying coupled multi-factor environmental stresses and dynamically changing electrical signal operating states. It can more realistically and efficiently reproduce the operating conditions of vehicle lights under complex real-world driving environments, thereby stimulating potential failure modes that are difficult to expose under single stress, greatly improving the accuracy and efficiency of testing. Furthermore, through the collaborative analysis of digital twin models and machine learning models, this invention can not only assess the current health status of vehicle lights but also predict their future performance degradation trends and remaining lifespan based on real-time data, achieving predictive management and elevating quality control from passive response to proactive early warning. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a method for detecting the reliability of automotive headlights according to an embodiment of the present invention.

[0027] Figure 2 This is another flowchart illustrating the method for detecting the reliability of automotive headlights according to an embodiment of the present invention. Detailed Implementation

[0028] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided to make the invention more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore repeated descriptions of them will be omitted.

[0029] The terms used to express position and direction in this invention are illustrated with reference to the accompanying drawings, but changes can be made as needed, and all such changes are included within the scope of protection of this invention.

[0030] This invention introduces a method, system, and medium for testing the reliability of automotive headlights.

[0031] Firstly, reference Figure 1 and Figure 2 The method for detecting the reliability of automotive headlights according to the present invention includes steps SS1 to SS3. Further, the detection method may also include steps SS4 to SS5.

[0032] Step SS1: During the test, apply coupled multi-factor environmental stress and dynamically changing electrical signal operating states to the headlight under test.

[0033] In application, coupled multi-factor environmental stress involves simultaneously applying multiple environmental stresses over time and simulating their interactions. For example, when the temperature rises to a high-temperature environment (e.g., 40°C to 125°C), random vibrations corresponding to a highway road spectrum are simultaneously initiated. High temperatures soften materials such as the lamp housing, lens, and solder joints, reducing their mechanical strength, which significantly exacerbates fatigue damage caused by vibration stress. Alternatively, the temperature can be rapidly lowered to a low-temperature environment (e.g., -40°C to 10°C) and then rapidly raised to a high-temperature environment (e.g., 40°C to 125°C), while maintaining a high-humidity environment (e.g., relative humidity 75% to 98%). During the low-temperature phase of the temperature cycle, the air pressure inside the lamp decreases, making it easier for high-humidity external air to be drawn into the lamp housing; during the high-temperature phase, internal moisture condenses, leading to short circuits or fogging of optical surfaces. Another example is the simultaneous initiation of dust storms (e.g., talc powder with a particle size of 0.5μm to 150μm) when applying vibration stress at a specific frequency (e.g., 200Hz to 500Hz). Vibration causes tiny gaps in the headlight structure (such as the joint between the lamp cover and the lamp housing) to reciprocate, acting like a pump and accelerating the intrusion and erosion of sand and dust.

[0034] In practical applications, dynamically changing electrical signal operating states simulate the complex, transient, and random electrical environments and intelligent control commands in real vehicles. For example, simulating transient events in the vehicle's basic electrical environment. Specifically, this could involve a voltage drop from 12V to 6V within one second (simulating a starter motor's high current pulling down the voltage), followed by a recovery to 14V within two to three seconds (simulating the alternator starting up). It could also simulate, within milliseconds, the sudden disconnection of the battery load during normal alternator operation, generating a transient high-voltage pulse with an amplitude of 80V to 100V. It could also simulate faults such as blown fuses or loose wiring, causing the voltage to momentarily drop to 9V or even 0V, recovering after hundreds of milliseconds. Furthermore, the voltage could fluctuate between 13.5V and 14.5V at a certain frequency and amplitude to simulate the regulation characteristics of the alternator regulator. For example, simulating dynamic commands related to the vehicle's own headlight functions and intelligent interaction. Specifically, the headlights could be cyclically switched on and off at a frequency of 0.5Hz to 5Hz to simulate frequent use of lights in urban road conditions (such as passing through tunnels or frequent oncoming traffic). It can also dynamically change the duty cycle via PWM signals to smoothly or stepwise change the headlight brightness between 0% and 100%, simulating daytime running lights, taillight brightness adjustment, or a "coming home" mode. It can also integrate with ADAS: receiving network signals from the simulated ADAS controller (such as when a camera detects oncoming vehicles), it can dynamically control specific areas of the LED matrix to turn off or dim in real time, achieving adaptive high beam functionality. The beam pattern continuously and intelligently changes according to the simulated scenario (such as curves, intersections, and pedestrians).

[0035] In some embodiments, step SS1 includes: step SS11.

[0036] Step SS11: Control the coupled multi-factor environmental stress and the working state of the electrical signal through a preset comprehensive test profile.

[0037] In application, the comprehensive test profile is a preset test configuration file used to control the dynamic changes in environmental stress and electrical signal operating states. Furthermore, the comprehensive test profile also integrates signals simulating vehicle network interactions, ensuring a high degree of consistency between the test and actual vehicle operating conditions. Specifically, the comprehensive test profile includes environmental stress combination sequences (such as high temperature and humidity followed by vibration and salt spray), electrical signal operating state sequences (such as simulated startup, voltage drop, and brightness adjustment), and vehicle network signal simulation.

[0038] In practical applications, the coupled multi-factor environmental stress of the present invention includes a combination of at least two of the following environmental factors: temperature, humidity, vibration, salt spray, and dust. Specifically, the temperatures include high-temperature environments (e.g., 40°C to 125°C, simulating high temperatures or exposure to direct sunlight in an engine compartment), normal-temperature environments (e.g., 10°C to 40°C), and low-temperature environments (e.g., -40°C to 10°C, simulating winter in frigid regions); the humidity includes high-humidity environments (e.g., relative humidity 75% to 98%, simulating rainy days or high-humidity climates), normal environments (e.g., relative humidity 30% to 75%), and low-temperature dry environments (e.g., relative humidity 10% to 30%, simulating dry desert or inland climates); the vibrations include random vibrations based on actual road spectrum acquisitions (e.g., frequencies from 5Hz to 500Hz, total root mean square acceleration values ​​from 1.0Grms to 5.0Grms, simulating different levels of road surface excitation) and constant-frequency sinusoidal vibrations (used to find the system's resonance point); the salt spray environment follows neutral salt spray test standards (e.g., sodium chloride solution concentration 5wt%, chamber temperature 35°C); the dust environment simulates talc powder or Arizona dust with specific particle sizes (e.g., 0.5μm to 150μm). Experiment with dust blowing.

[0039] Furthermore, the dynamically changing electrical signal operating states of the invention include at least one of the following operating states: simulating vehicle startup (e.g., voltage momentarily dropping from 12V to 6V and recovering), load dumping (e.g., generating a transient high-voltage pulse with an amplitude of up to 80V or more), voltage drop (e.g., short-term power outage or voltage dropping below 9V), steady-state fluctuation (e.g., voltage fluctuating in the range of 13.5V to 14.5V, simulating generator regulation), frequent switching (e.g., cyclically switching vehicle lights on and off at a frequency of 0.5Hz to 5Hz), brightness adjustment (e.g., brightness adjustment from 0% to 100% via PWM modulation), and beam changes linked to advanced driver assistance systems (e.g., dynamically obscuring part of the beam according to the simulated target position).

[0040] Furthermore, the signals for the simulated vehicle network interaction of this invention follow Controller Area Network (CAN / CAN FD, communication rate up to 5Mbps), Local Area Network (LIN, communication rate typically 20kbps), or in-vehicle Ethernet protocol (following IEEE 802.3 standard, rate up to 100Mbps / 1Gbps). In addition, the signals for the simulated vehicle network interaction include standardized data frames for controlling headlight switching, mode switching, and adaptive beam adjustment.

[0041] In actual implementation, the headlight under test is placed in a comprehensive environmental simulation device, which integrates a comprehensive test profile to simulate the working state of the headlight in the whole vehicle environment.

[0042] Specifically, the integrated environmental simulation device is used to simulate the multi-factor coupled stress that vehicle lights experience under real driving conditions in a laboratory environment. The integrated environmental simulation device integrates multiple environmental simulation units and electrical signal control units, capable of simultaneously applying environmental stresses such as temperature, humidity, vibration, salt spray, and dust, as well as dynamically changing electrical signal operating states (such as voltage fluctuations and frequent switching). Furthermore, the integrated environmental simulation device has a built-in comprehensive test profile, which controls the stress (environmental stress and electrical signal stress) application process through preset programs, ensuring test consistency and repeatability.

[0043] Step SS2: Real-time acquisition of environmental stress parameters, electrical signal operating status parameters, and the actual performance parameters of the vehicle lamp under test.

[0044] When applied, the actual performance parameters include at least one of the following: optical performance parameters, electrical performance parameters, and physical state parameters.

[0045] In practical applications, the optical performance parameters include at least one of luminous flux and color temperature. The electrical performance parameters include at least one of operating current and operating voltage. The physical state parameters include at least one of the following: internal lamp temperature, lamp housing appearance change parameters, and lens appearance change parameters.

[0046] The internal temperature parameters of the lamp can include the surface temperature of the LED chip or light source module, the hot spot temperature of the driver circuit board, the air temperature inside the lamp housing, and the temperature of the heat sink or heat-conducting components, in order to monitor the internal thermal management performance of the vehicle lamp and prevent overheating from causing light decay or electrical failure.

[0047] Appearance change parameters can include: crack-related parameters, scratch / wear-related parameters, color-related parameters, haze-related parameters, deformation-related parameters, and corrosion-related parameters. Specifically, crack-related parameters can include: total crack length, maximum crack width, number of cracks, and crack distribution density (number or length of cracks per unit area). Scratch-related parameters can include: total scratch length, average scratch depth, and surface wear area ratio (percentage of wear area to total lens area). Color-related parameters can include: color difference and yellow index (an indicator used to quantify the degree of yellowing of materials). Haze-related parameters can include: haze value (calculated by analyzing the grayscale distribution of acquired images to determine the degree of surface haze and opacity). Deformation-related parameters can include: curvature change and flatness error. Corrosion-related parameters can include: number of corrosion spots and area of ​​corrosion zone.

[0048] Step SS3: Based on the digital twin model and machine learning model, determine the health status index of the vehicle lamp under test and perform fault prediction according to the environmental stress parameters, electrical signal working status parameters and actual performance parameters.

[0049] When applied, the health status indicators include a state vector reflecting the overall health of the vehicle lights and a preliminary prediction of the remaining useful life.

[0050] In practical applications, the training method for digital twin models includes: establishing an initial digital twin model based on the physical characteristics of the vehicle headlights (such as thermodynamics, optics, and electrical properties); inputting historically measured environmental stress parameters and historical electrical signal operating states into the initial digital twin model, outputting historically projected performance parameters; comparing the historically projected performance parameters with historically actual performance parameters (such as luminous flux, current, and temperature) to calculate the error; and adjusting the model parameters of the initial digital twin model using an inversion algorithm (such as the least squares method) to make the historically projected performance parameters approximate the historically actual performance parameters.

[0051] Furthermore, the training methods for machine learning models include: extracting fault mode features (such as light decay curves and voltage anomaly modes) from historical real-vehicle test databases; and extracting historical time-series features (such as degradation rates and fluctuation characteristics) from historical health status indicators and historical actual performance parameters.

[0052] Historical time-series features and historical failure features are fused into a historical feature vector. This historical feature vector is then used as input to train machine learning models (such as classification and regression algorithms) to identify historical failure mode probabilities, predict historical remaining useful life, and predict historical performance parameters.

[0053] In actual implementation, step SS3 includes steps SS31 to SS34.

[0054] Step SS31: Input the environmental stress parameters and electrical signal operating status parameters into the digital twin model, and output the inferred performance parameters of the vehicle lamp under test through simulation.

[0055] Step SS32: Compare the actual performance parameters with the inferred performance parameters to calibrate the digital twin model.

[0056] When applied, if the error between the actual performance parameters and the inferred performance parameters exceeds a preset threshold (e.g., 5%), the physical parameters (e.g., material aging coefficient, thermal conductivity coefficient) in the digital twin model are adjusted.

[0057] Step SS33: Based on the calibrated digital twin model, determine the health status index of the vehicle lamp under test according to the environmental stress parameters and electrical signal operating status parameters. When applied, step SS33 includes steps SS331 to SS332.

[0058] Step SS331: Based on the calibrated digital twin model, perform simulation and deduction according to the preset future test profile to obtain the performance parameters (which can be deduced performance parameters) of the vehicle lamp under test under the continuous application of the coupled multi-factor environmental stress and the dynamically changing electrical signal working state, and obtain the performance degradation trajectory based on physical simulation.

[0059] In application, the performance degradation trajectory based on physical simulation is obtained according to the change process of the performance parameters of the vehicle headlight under test.

[0060] In practical applications, the future test profile (i.e., the continuous application of the coupled multi-factor environmental stress and the dynamically changing electrical signal operating state) is input into a calibrated digital twin model. Based on its built-in physical laws (such as thermodynamic models, material aging models, and photoelectric conversion models), the digital twin model calculates how the performance parameters of the vehicle lamp under test (i.e., extrapolated performance parameters, such as luminous flux, color temperature, and internal temperature) will change at each future time step. Arranging the extrapolated performance parameters in chronological order yields the performance degradation trajectory based on physical simulation.

[0061] Step SS332: Based on the performance degradation trajectory, calculate the state vector reflecting the overall health of the headlights, and combine it with a preset first failure threshold to calculate a preliminary prediction of the remaining useful life, so as to determine the health status index.

[0062] In application, a first feature reflecting the degree and trend of degradation is extracted from the performance degradation trajectory. This first feature constitutes the various dimensions of the state vector. The feature values ​​of the first feature are combined to form a state vector reflecting the overall health of the vehicle lights.

[0063] In practical applications, the performance degradation trajectory based on physical simulation is used to continue extrapolating along its trend until the performance parameter value on the performance degradation trajectory first reaches the preset first failure threshold. The time interval between the point when the first failure threshold is reached and the current point in time is the preliminary prediction of the remaining useful life.

[0064] In practice, the first failure threshold can be a set of thresholds for different failure modes. Each value in the set is set with reference to the minimum requirements for vehicle lighting performance in automotive industry standards (such as ISO and SAE), vehicle manufacturer standards, or national regulations.

[0065] Step SS34: Based on the machine learning model, and according to the health status indicators and the performance parameters (which may be extrapolated performance parameters), perform fault prediction for the vehicle light under test.

[0066] In application, fault prediction can include fault mode recognition to determine how the headlight under test might fail and output the probability of each fault mode. For example: light decay of LED chips leads to insufficient brightness, overheating of the driver circuit leads to unstable performance, lenses crack or yellow severely due to thermal stress, and connectors loosen due to vibration.

[0067] In practical applications, step SS34 includes steps SS341 to SS342. Further, step SS34 may also include steps SS343 to SS345.

[0068] Step SS341: Extract time-series features from the health status indicators and performance parameters (which may be extrapolated performance parameters), and fuse the time-series features with the fault mode features in the historical test database (which may be a historical real vehicle test database) to form a fused feature vector.

[0069] In application, a data sequence within a time window is extracted from continuously acquired health status indicators and performance parameters, and features that reflect the dynamic change pattern of the data sequence are extracted as time-series features.

[0070] In practical applications, the extracted time-series features are concatenated and combined with the fault mode features to form a fused feature vector.

[0071] Step SS342: Input the fused feature vector into the pre-trained machine learning model, and determine the fault mode probability of the test lamp by identifying the similarity between the fused feature vector and the fault mode features.

[0072] When applied, the similarity between the current fused feature vector and each fault mode category is calculated to obtain the probability of each fault mode.

[0073] Step SS343: Input the performance degradation trajectory into the pre-trained machine learning model, and determine the remaining useful life prediction value and performance parameter prediction value of the vehicle lamp under test by learning the changing trend of the performance parameters of the performance degradation trajectory.

[0074] When applied, the machine learning model should include at least one regression-type machine learning model, which uses data-driven methods to capture complex nonlinear degradation behaviors that may be ignored by the digital twin model (physical model), making predictions more accurate. Step SS344: Generate the health level announcement information of the vehicle lamp under test based on the remaining useful life prediction value and the failure mode probability.

[0075] In application, the machine learning model pre-constructs a decision tree. For example: when A > 1000 h and all B values ​​< 0.1, the health level is green; when 500 h ≤ A ≤ 1000 h, or any B ≥ 0.1, the health level is yellow; when 100 h ≤ A < 1000 h, or any B ≥ 0.5, the health level is orange; when A < 100 h, or any B ≥ 0.8, the health level is red. Here, A represents the predicted remaining useful life; B represents the failure mode probability. Step SS345: When the predicted performance parameter value exceeds the second failure threshold within a preset time, a maintenance warning is generated.

[0076] In application, the second failure threshold is a set of multiple parameters and modes, allowing for the setting of thresholds for key parameters such as luminous flux, color temperature, and current. Furthermore, the settings of each value in the set can comply with the mandatory minimum requirements for vehicle lighting performance in international (such as ECE, SAE), national, or automotive industry standards.

[0077] In practical applications, maintenance early warning information can include the warning type, predicted time of exceeding the limit, the involved failure mode, and recommended measures.

[0078] In actual implementation, if any projected performance parameter within a certain period of time exceeds the upper limit of the corresponding second failure threshold or falls below the lower limit of the second failure threshold, an alarm will be issued and maintenance warning information will be generated. For example, a warning for insufficient luminous flux indicates that the performance is expected to fall below the industry standard after 20 hours. This warning is highly correlated with the light decay failure mode, with a probability of 85%, and it is recommended to check and replace the headlights in a timely manner.

[0079] Step SS4: Assemble the headlights under test onto the real vehicle test platform, conduct long-term testing in a real road environment, and collect real vehicle verification datasets.

[0080] In application, the test headlights that have completed laboratory simulation tests are mounted on a real vehicle test platform (such as a test vehicle) and subjected to long-term operational verification tests in a real road environment. Using an on-board data acquisition system, the working data, environmental data, and vehicle bus data of the test headlights under actual working conditions are collected in real time through vehicle networking technology to form a real vehicle verification dataset.

[0081] Step SS5: Validate and optimize the digital twin model and machine learning model using the collected real-vehicle verification dataset.

[0082] In application, the digital twin model and machine learning model are jointly validated and optimized based on the collected real-vehicle verification dataset. Specifically, this includes: comparing real-vehicle data with model prediction results to evaluate the model's prediction accuracy and robustness under real-world conditions; and using real-vehicle data to retrain and adaptively correct model parameters to improve the model's generalization ability and long-term prediction accuracy in complex real-world environments.

[0083] Secondly, the automotive headlight reliability testing system of the present invention includes: an environmental simulation module, a data acquisition module, and a testing and analysis module. Furthermore, the testing system may also include: a real-vehicle testing module and a verification and optimization module.

[0084] The system comprises the following modules: an environmental simulation module, used to apply coupled multi-factor environmental stress and dynamically changing electrical signal operating states to the headlight under test during testing; a data acquisition module, used to acquire environmental stress parameters, electrical signal operating state parameters, and the actual performance parameters of the headlight under test in real time; a detection and analysis module, used to analyze the performance degradation trend and predict faults of the headlight under test based on the environmental stress parameters, electrical signal operating state parameters, and actual performance parameters, using a digital twin model and a machine learning model; a real vehicle testing module, used to assemble the headlight under test onto a real vehicle testing platform for long-term testing in actual road environments and collect real vehicle verification datasets; and a verification and optimization module, used to verify and optimize the digital twin model and machine learning model using the collected real vehicle verification datasets.

[0085] The specific implementation of the above functional modules refers to the specific steps of the above-mentioned automotive headlight reliability testing method.

[0086] Thirdly, the computer-readable storage medium of the present invention stores a computer program that, when executed by at least one processor, implements the steps of the detection method.

[0087] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the invention without departing from the principles and spirit of the invention, and all such changes should fall within the protection scope of the claims of the present invention.

Claims

1. A method for testing the reliability of automotive headlights, characterized in that, include: During the test, the test headlights are subjected to coupled multi-factor environmental stress and dynamically changing electrical signal operating states. Real-time acquisition of environmental stress parameters, electrical signal operating status parameters, and actual performance parameters of the vehicle lamp under test; Based on digital twin models and machine learning models, the health status indicators of the vehicle lamp under test are determined and fault prediction is performed according to the environmental stress parameters, electrical signal operating status parameters and actual performance parameters.

2. The method for testing the reliability of automotive headlights according to claim 1, characterized in that, The coupled multi-factor environmental stress includes a combination of at least two of the following: temperature, humidity, vibration, salt spray, and dust. The dynamically changing electrical signal operating states include at least one of the following operating states: simulated vehicle start-up, load shedding, voltage drop, steady-state fluctuation, frequent switching, brightness adjustment, and beam changes linked to advanced driver assistance systems.

3. The method for testing the reliability of automotive headlights according to claim 2, characterized in that, The operating state of the multi-factor environmental stress and dynamically changing electrical signal coupled to the headlight under test includes: The multi-factor environmental stress of the coupling and the working state of the electrical signal are controlled by a preset comprehensive test profile. The integrated test profile incorporates signals simulating vehicle network interaction, and these signals follow controller area network, local interconnection network, or vehicular Ethernet protocols.

4. The method for testing the reliability of automotive headlights according to claim 1, characterized in that, The actual performance parameters include at least one of optical performance parameters, electrical performance parameters, and physical state parameters; The optical performance parameters include at least one of luminous flux and color temperature; The electrical performance parameters include at least one of operating current and operating voltage; The physical state parameters include at least one of the following: internal temperature parameters of the lamp body, appearance change parameters of the lamp housing, and appearance change parameters of the lens.

5. The method for testing the reliability of automotive headlights according to claim 1, characterized in that, The method based on digital twin and machine learning models, determining the health status indicators of the vehicle lamp under test and predicting faults according to the environmental stress parameters, electrical signal operating status parameters, and actual performance parameters, includes: The environmental stress parameters and electrical signal operating status parameters are input into the digital twin model, and the simulated performance parameters of the vehicle lamp under test are output through simulation. The actual performance parameters are compared with the inferred performance parameters to calibrate the digital twin model; Based on the calibrated digital twin model, the health status index of the vehicle lamp under test is determined according to the environmental stress parameters and electrical signal operating status parameters. Based on the machine learning model, the fault prediction of the vehicle light under test is performed according to health status indicators and performance parameters.

6. The method for testing the reliability of automotive headlights according to claim 5, characterized in that, The calibrated digital twin model determines the health status indicators of the vehicle lamp under test based on the environmental stress parameters and electrical signal operating status parameters, including: Based on the calibrated digital twin model, simulations are performed according to the preset future test profile to obtain the performance parameter change process of the vehicle lamp under test under the continuous application of the coupled multi-factor environmental stress and the dynamically changing electrical signal working state, thus obtaining the performance degradation trajectory based on physical simulation. Based on the performance degradation trajectory, a state vector reflecting the overall health of the vehicle lights is calculated, and combined with a preset first failure threshold, a preliminary prediction of the remaining useful life is calculated to determine the health status index.

7. The method for testing the reliability of automotive headlights according to claim 6, characterized in that, The step of predicting the fault of the vehicle light under test based on the machine learning model, according to health status indicators and performance parameters, includes: The time-series features extracted from health status indicators and performance parameters are fused with the fault mode features in the historical test database to form a fused feature vector. The fused feature vector is input into a pre-trained machine learning model, and the probability of the fault mode of the test lamp is determined by identifying the similarity between the fused feature vector and the fault mode features.

8. The method for testing the reliability of automotive headlights according to claim 7, characterized in that, The step of predicting the fault of the vehicle light under test based on the machine learning model, according to health status indicators and performance parameters, includes: The performance degradation trajectory is input into a pre-trained machine learning model. By learning the changing trend of the performance parameters of the performance degradation trajectory, the remaining useful life and performance parameter prediction values ​​of the vehicle lamp under test are determined. Based on the predicted remaining useful life and the failure mode probability, the health level announcement information of the vehicle lamp under test is generated; When the predicted value of the performance parameter exceeds the preset second failure threshold within a preset time, a maintenance warning message is generated.

9. A system for testing the reliability of automotive headlights, characterized in that, include: The environmental simulation module is used to apply coupled multi-factor environmental stress and dynamically changing electrical signal operating states to the headlights under test during the testing process. The data acquisition module is used to acquire environmental stress parameters, electrical signal operating status parameters, and the actual performance parameters of the vehicle lamp under test in real time. The detection and analysis module is used to analyze the performance degradation trend and predict the fault of the vehicle lamp under test based on the digital twin model and machine learning model, according to the environmental stress parameters, electrical signal operating state parameters and actual performance parameters.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by at least one processor, implements the steps of the detection method according to any one of claims 1 to 8.

Images