Motor vehicle LED lamp thermal field distribution simulation analysis system based on digital twinning

Abstract

This invention relates to the field of motor vehicle lighting and digital twin simulation technology, specifically to a simulation and analysis system for the thermal field distribution of LED vehicle lights based on digital twins. The system includes: collecting multi-source dynamic boundary condition data of the vehicle lights during operation; inputting this data into a reduced-order model to solve the transient three-dimensional temperature field; extracting temperature gradient features and combining them with material parameters to calculate the real-time thermal deformation of key structural components using a solid mechanics model; mapping the thermal deformation to an initial model to generate a deformation geometry model; importing a ray tracing algorithm to calculate the light source focus offset and generating light distribution projection data; comparing this data with a light distribution safety threshold; if the threshold is exceeded, instructing the control unit to perform optical compensation; otherwise, instructing it to maintain the current state. This invention completely solves the problem of severe model lag in existing solutions, perfectly meeting the high timeliness requirements during motor vehicle operation.

Description

Technical Field

[0001] This invention relates to the field of motor vehicle lighting and digital twin simulation technology, specifically to a simulation and analysis system for the thermal field distribution of LED vehicle lights based on digital twins. Background Technology

[0002] In the current motor vehicle operating environment, LED vehicle lights are affected by the superposition of internal heat source drive and external meteorological heat exchange boundary under complex road conditions, resulting in violent and complex transient three-dimensional thermal field evolution. To analyze and intervene in thermally induced optical failures of automotive headlights, existing solutions generally employ full-scale three-dimensional fluid dynamics simulations or rely on single temperature feedback to perform global power derating. While these solutions have some processing capabilities in offline design analysis and static overheat protection scenarios, the long computation time of high-fidelity thermodynamic models, severe system lag in solution processing, and the failure of existing mechanisms to effectively map the temperature field to the dynamic thermal deformation of structural components and the shift of the light source focus result in lengthy online analysis processes, high control response delays, and a lack of targeted optical compensation. Consequently, they are unable to support real-time prediction and dynamic closed-loop adjustment of light distribution safety during vehicle operation. Therefore, how to achieve rapid real-time mapping simulation of the thermal field distribution of vehicle lamps to optical failure under complex working conditions, and improve the timeliness and accuracy of active light distribution compensation, has become an urgent technical problem to be solved. Summary of the Invention

[0003] The purpose of this invention is to provide a simulation and analysis system for the thermal field distribution of automotive LED headlights based on digital twins, and to solve the following technical problems: It avoids the problem that single temperature monitoring cannot cope with structural deformation and light pattern inaccuracy caused by dynamic thermal shock, and makes it easier to realize active compensation linkage adjustment from light distribution safety prediction to both heat control and light control.

[0004] The objective of this invention can be achieved through the following technical solutions: A simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins includes a processor and a memory. The memory stores a computer program, and the processor executes the computer program to perform the following steps: Collect multi-source dynamic boundary condition data of motor vehicle LED lights during operation: The motor vehicle LED lights include key structural components, LED light-emitting arrays, heat dissipation actuators, and control units; The multi-source dynamic boundary condition data is input into a preset reduced-order model to solve the transient three-dimensional temperature field data. The reduced-order model is constructed by processing the preset full-order thermodynamic equations through intrinsic orthogonal decomposition. Extract the temperature gradient features from the transient three-dimensional temperature field data; The material parameters of the key structural component are obtained, the temperature gradient characteristics are input into a preset solid mechanics mapping model, and the real-time thermal deformation of the key structural component is calculated. Based on a unified spatial coordinate system, the real-time thermal deformation is mapped to the corresponding mesh nodes of the preset initial geometric model of the key structural component to generate a deformation geometric model. The deformation geometry model is imported into a preset ray tracing algorithm to calculate the light source focus offset of the LED light-emitting array; Based on the light source focus offset, generate light distribution projection data; The light distribution projection data is compared with a preset light distribution safety threshold. In response to the light distribution projection data exceeding the light distribution safety threshold, a first control command is generated and sent to the control unit for optical compensation; In response to the light distribution projection data being within the light distribution safety threshold, a second control command is generated and sent to the control unit to maintain the current state.

[0005] Optionally, collect multi-source dynamic boundary condition data of the vehicle's LED lights during operation, including: Acquire real-time telemetry data of the vehicle's LED lights; Acquire the real-time electrical operating data of the vehicle's LED headlights; Obtain meteorological interface data of the environment in which the vehicle's LED headlights are located; The real-time telemetry data, the real-time electrical operation data, and the meteorological interface data are time-stamp aligned to generate the multi-source dynamic boundary condition data.

[0006] Optionally, the specific construction process of the order reduction model includes: Obtain high-fidelity three-dimensional fluid dynamics simulation data; The high-fidelity three-dimensional fluid dynamics simulation data were subjected to intrinsic orthogonal decomposition to extract spatial basis functions; Construct the projection matrix based on the spatial basis functions; The full-order thermodynamic equation is reduced in dimension using the projection matrix to generate the reduced-order model.

[0007] Optionally, the material parameters of the key structural component are obtained, the temperature gradient characteristics are input into a preset solid mechanics mapping model, and the real-time thermal deformation of the key structural component is calculated, including: Based on the temperature gradient characteristics, the three-dimensional temperature difference field inside the key structural component is calculated. The coefficient of thermal expansion of the material of the key structural component is obtained as the material parameter. The thermal strain field is obtained by multiplying the three-dimensional temperature difference field with the thermal expansion coefficient of the material. Combined with the preset structural constraint boundary conditions, the three-dimensional displacement field data is generated by solving the constitutive equation of elasticity. Surface node displacement vectors are extracted from the three-dimensional displacement field data, and the surface node displacement vectors are used as the real-time thermal deformation.

[0008] Optionally, the deformation geometry model is imported into a preset ray tracing algorithm to calculate the light source focus offset of the LED light-emitting array, including: Extract the optical surface profile data of the deformable geometry model; Reconstruct the optical tracing boundary conditions based on the optical surface profile data; Generate a set of virtual rays; Perform forward ray tracing calculations on the virtual ray set according to the optical tracing boundary conditions to obtain the centroid coordinates of the actual focal plane spot; Calculate the spatial vector difference between the actual focal plane spot centroid coordinates and the preset theoretical focal plane spot centroid coordinates, and use the spatial vector difference as the light source focus offset.

[0009] Optionally, based on the light source focal offset, light distribution projection data is generated, including: Obtain a preset virtual test screen model; Based on the light source focal offset, a spatial translation and rotation transformation is performed on the preset light source luminous intensity distribution matrix to generate an offset luminous intensity distribution matrix; The offset luminous intensity distribution matrix is ​​projected onto the virtual test screen model, and the illuminance value of each pixel node on the virtual test screen model is calculated. The illuminance values ​​of each pixel node are combined to generate the light distribution projection data.

[0010] Optionally, the light distribution projection data is compared with a preset light distribution safety threshold, including: The illuminance gradient value of the cutoff line is extracted from the light distribution projection data as the feature parameter of the cutoff line, and the illuminance value of the test point under the preset spatial coordinates is also extracted. The characteristic parameters of the cutoff line are compared with the preset cutoff line sharpness threshold. The illuminance value at the test point is compared with a preset upper limit threshold for glare illuminance to determine whether any of the test point illuminance values ​​exceeds the upper limit threshold for glare illuminance. In response to the fact that the characteristic parameters of the cutoff line are not lower than the cutoff line sharpness threshold and all the illuminance values ​​of the test points are not higher than the glare illuminance upper limit threshold, it is determined that the light distribution projection distribution data exceeds the light distribution safety threshold. In response to the fact that the characteristic parameter of the cutoff line is not lower than the cutoff line sharpness threshold and the illuminance value of the test point is not higher than the upper limit threshold of glare illuminance, it is determined that the light distribution projection distribution data is included within the light distribution safety threshold.

[0011] Optionally, generate first control instructions, including: Based on the difference between the light distribution projection data and the light distribution safety threshold, the target power derating factor and the target heat dissipation speed are calculated through a preset proportional-integral-derivative control logic. The target power derating factor is packaged into a light-emitting array adjustment command; The target heat dissipation speed is packaged into a heat dissipation actuator adjustment command; The first control command is generated by combining the light-emitting array adjustment command and the heat dissipation actuator adjustment command.

[0012] Optionally, when the processor executes the computer program, it also performs the following steps: Build a historical thermal cycle database; The transient three-dimensional temperature field data is continuously written into the historical thermal cycle database; Extract temperature alternation amplitude and temperature alternation frequency features from the historical thermal cycle database; Extract the position information of the target weld points preset on the key structural component from the deformation geometry model; A thermal fatigue life prediction model is constructed based on a preset material fatigue curve or the Coffin-Manson equation. The temperature alternation amplitude characteristics, the temperature alternation frequency characteristics, and the location information of the target weld point are input into the thermal fatigue life prediction model to calculate the cumulative fatigue damage degree of the target weld point. Based on the cumulative fatigue damage degree, the remaining service life prediction value of the target weld point is derived. In response to the predicted remaining service life being lower than a preset service life warning threshold, a maintenance prompt message is generated and output. In response to the remaining useful life prediction value being not lower than the useful life warning threshold, the monitoring status is maintained.

[0013] Optionally, the reduced-order model, the solid mechanics mapping model, and the ray tracing algorithm are all deployed on a cloud-based simulation computing platform. The multi-source dynamic boundary condition data is continuous time-series data obtained through the vehicle network communication interface. The first control command and the second control command are sent to the control unit through the controller local area network bus.

[0014] The beneficial effects of this invention are: 1. This invention breaks through the barrier of time-consuming iteration in traditional full-size three-dimensional thermodynamic simulation by introducing a reduced-order model based on intrinsic orthogonal decomposition to process multi-source dynamic boundary condition data. This mechanism greatly accelerates the solution process of transient three-dimensional temperature field data, completely solves the problem of serious model solution lag in existing schemes, and perfectly meets the high timeliness requirements in the process of motor vehicle driving. 2. This invention inputs temperature gradient features into a solid mechanics mapping model to calculate the real-time thermal deformation of key structural components, and then imports a ray tracing algorithm to solve the focal offset of the light source. This scheme successfully transforms complex local thermal inhomogeneity into quantifiable geometric yaw and light distribution projection data, making up for the lack of correlation mapping between structural deformation and optical focal point in existing technologies, and achieving accurate prediction of light distribution misalignment. 3. This invention performs multi-dimensional safety judgment based on the illuminance gradient of the light and dark cutoff line and the glare illuminance of the test point, and generates target power derating and heat dissipation speed commands through proportional-integral-derivative control logic. This mechanism abandons the traditional crude mode of relying on a single temperature feedback for global indiscriminate derating, and can suppress glare risk while avoiding a cliff-like attenuation in the effective lighting area in front, significantly improving the targeting of active compensation. 4. This invention innovatively performs unified timestamp alignment processing on real-time telemetry data, electrical operation data and meteorological interface data of motor vehicles; this ensures the absolute synchronization of internal heat source fluctuations and external complex heat exchange conditions on the same physical event chain, effectively avoiding phase misalignment of thermal field response caused by input errors, and establishing a precise data foundation for digital twin solutions that is highly consistent with the real road conditions. 5. This invention deploys the high-computation-power-consuming reduced-order model, solid mechanics mapping, and ray tracing algorithm on a cloud simulation platform, while the vehicle focuses on continuous time-series data acquisition and local area network command distribution. This architecture effectively breaks through the computing power and storage limitations of a single vehicle control unit, ensuring both efficient iteration of complex digital twin models and maintaining the agility and stability of the vehicle lighting's underlying control actions. 6. This invention constructs a historical thermal cycle database by continuously accumulating transient temperature field data, and calculates the cumulative fatigue damage of the target weld point under long-term temperature alternation amplitude and frequency by combining the material fatigue curve, and derives the remaining life; this mechanism expands the vehicle headlight safety management from real-time control to long-term reliability monitoring, and can output maintenance prompts in advance before sudden loss of light or pixel malfunction, preventing problems before they occur. Attached Figure Description

[0015] The invention will now be further described with reference to the accompanying drawings.

[0016] Figure 1This is a flowchart illustrating a simulation analysis method for the thermal field distribution of automotive LED headlights based on digital twins, provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 A simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins includes a processor and a memory. The memory stores a computer program. When the processor executes the computer program, it performs the following steps: collecting multi-source dynamic boundary condition data of automotive LED headlights during operation. The automotive LED headlights include key structural components, LED light-emitting arrays, heat dissipation actuators, and control units. The multi-source dynamic boundary condition data is input into a preset reduced-order model to solve the transient three-dimensional temperature field data. The reduced-order model is constructed by processing the preset full-order thermodynamic equation through intrinsic orthogonal decomposition. The temperature gradient features in the transient three-dimensional temperature field data are extracted. The material parameters of the key structural component are obtained, the temperature gradient characteristics are input into a preset solid mechanics mapping model, and the real-time thermal deformation of the key structural component is calculated. Based on a unified spatial coordinate system, the real-time thermal deformation is mapped to the corresponding mesh node of the preset initial geometric model of the key structural component to generate a deformation geometric model. The deformation geometry model is imported into a preset ray tracing algorithm to calculate the light source focus offset of the LED light-emitting array; based on the light source focus offset, light distribution projection data is generated; the light distribution projection distribution data is compared with a preset light distribution safety threshold. In response to the light distribution projection data exceeding the light distribution safety threshold, a first control command is generated and sent to the control unit for optical compensation; in response to the light distribution projection data being within the light distribution safety threshold, a second control command is generated and sent to the control unit to maintain the current state.

[0019] This embodiment provides a simulation analysis mechanism for the thermal field distribution of LED vehicle lights based on digital twins. Specifically, the system is deployed on a heavy logistics vehicle that performs nighttime highway trunk line transportation tasks. The vehicle is equipped with a matrix LED headlight assembly. The headlight assembly includes key structural components such as lenses, reflectors, lamp housing brackets, and heat sinks. It includes multiple independently adjustable LED light-emitting units and a control unit connected to the vehicle's electronic and electrical architecture. The system's processor periodically reads vehicle operation information and lamp status information, and calls a preset program in memory to continuously build a digital twin of the vehicle lamp under real road conditions; Specifically, the mechanism acquires dynamic boundary data that affects the thermal behavior of vehicle lights; these boundaries are not abstract mathematical boundaries, but rather the real engineering conditions upon which heat input and dissipation depend, such as whether the vehicle is idling for a long time, whether it has just switched from low-speed congestion to high-speed cruising, whether it has experienced sudden environmental changes at tunnel entrances and exits, and whether there is external rainfall that enhances convective heat transfer on the surface of the mask. The processor inputs these boundary data into a preset reduced-order model to quickly obtain the transient three-dimensional temperature field at various locations inside the headlight. This temperature field does not reflect a single temperature value, but rather the temperature difference distribution between different materials, thicknesses, and locations. The system further extracts temperature gradient features from this data to identify whether there is significant thermal inhomogeneity in areas such as the center and edge of the lens, the connection between the heat sink and the bracket, and the area around the LED substrate and solder joints. After obtaining the temperature gradient, the system calls the solid mechanics mapping model and calculates the real-time thermal deformation by combining the material parameters of the key structural components. The physical basis is that different materials have different expansion tendencies under the same temperature change, and the key structural components usually have screws, clips, adhesives or support constraints. Therefore, they do not elongate freely after being heated, but will exhibit warping, sinking, swaying or local bulging. The processor maps the obtained surface node displacements onto the initial geometric model in a unified coordinate system to form a geometric model that reflects the current thermal deformation state. Next, the model is imported into the ray tracing algorithm to recalculate the imaging path of the LED light-emitting array after passing through the current optical surface, and the focal offset is obtained. Then, the light distribution projection data is generated based on the focal offset to determine whether the light distribution of the current road lighting area, the light-dark cutoff line and the above sensitive area still meets the safety requirements. If the comparison results show that the light distribution exceeds the safety threshold, the system generates the first control command and sends it to the control unit to perform optical compensation. This compensation can be manifested as reducing the power of some LED pixels, adjusting the driving distribution of the light-emitting array, or increasing the working intensity of the heat dissipation actuator, so as to limit the light pattern misalignment before the thermal deformation continues to expand. If the comparison result is within the safety threshold, a second control command is generated to maintain the current lighting mode and heat dissipation status, avoiding unnecessary adjustments that could cause insufficient lighting or frequent device operation. To ensure control reliability, the processor can perform a consistency check before sending instructions, such as confirming that there are no abnormal jumps in the current geometric model compared to the previous cycle, confirming that the sampling clock is continuous, and confirming that the control unit is online. Specifically, if there is a lack of multi-source boundary data in a certain period, such as a short-term interruption of the environmental interface or an abnormal fixed value returned by a temperature sensor inside the lamp, the system can roll back to the effective boundary conditions of the previous moment and reduce the control weight of the twin calculation in that period, only maintaining monitoring or performing conservative compensation. If recovery fails for several consecutive cycles, the control unit enters a safety degradation state, prioritizing preventing overheating and significant glare, rather than continuing to perform high-sensitivity dynamic optimization; if the model calculation results deviate significantly from the actual measured state of the headlights, the model credibility flag can also be triggered, pausing active compensation and retaining only the alarm output. For example, the logistics vehicle entered a mountainous highway section on a summer night. The preceding road conditions were a long period of idling in a service area while waiting for loading and unloading, and the headlights had accumulated a high amount of heat. After the vehicle left the service area and entered highway cruising, it encountered a short period of heavy rain a few minutes later. At this time, the outer cover of the headlight is cooled down rapidly by the cold rain, while the internal heat sink remains at a high temperature. A significant temperature difference is formed between the lens and the bracket. The system deduces in real time that the lens is slightly tilted forward and accompanied by a blurring of the cutoff line. Based on this, the processor generates the first control command in advance to appropriately derate the LED subarray corresponding to the upper sensitive area, while increasing the speed of the cooling fan, thereby suppressing the risk of glare from oncoming lanes. The purpose of this step is to connect the dynamic thermal shock, structural deformation, and light pattern misalignment in real roads into a single causal chain that can be calculated in real time, thereby achieving a technological shift from temperature monitoring to light distribution safety prediction and proactive compensation.

[0020] In this embodiment of the invention, the collection of multi-source dynamic boundary condition data of a vehicle LED headlight during operation includes: acquiring real-time telemetry data of the vehicle LED headlight; acquiring real-time electrical operation data of the vehicle LED headlight; acquiring meteorological interface data of the environment in which the vehicle LED headlight is located; and performing timestamp alignment processing on the real-time telemetry data, the real-time electrical operation data, and the meteorological interface data to generate the multi-source dynamic boundary condition data.

[0021] This embodiment provides a multi-source dynamic boundary condition data acquisition mechanism. Specifically, in the aforementioned nighttime logistics transportation scenario, simply collecting the lamp body temperature or the vehicle speed is insufficient to describe the true evolution of the thermal field, because the heat load of the vehicle lamp is affected by internal heating, external heat exchange conditions, and changes in vehicle posture and driving status. Therefore, this embodiment divides the data sources into three categories: real-time telemetry data, real-time electrical operation data, and environmental meteorological interface data, and then aligns the timestamps using a unified clock. Specifically, real-time telemetry data may include vehicle speed, engine compartment temperature, battery storage system status, front compartment airflow field information, frontal wind position, ambient temperature and humidity, and rain sensing signals; this type of data mainly reflects the heat transfer boundary outside the headlights and the overall vehicle operating condition. Real-time electrical operation data may include LED array partition current, drive voltage, PWM duty cycle, cooling fan drive level and control unit working mode, etc. This type of data directly reflects the intensity of the heat source inside the lamp body; meteorological interface data supplements regional weather changes that cannot be detected in advance by local sensors, such as whether there is a rainstorm belt, a sudden drop in temperature or a high humidity fog area on the road ahead. The significance of timestamp alignment is to synchronize the heat source input and external heat dissipation conditions at the same physical moment. Otherwise, the model will mistakenly regard events that occur one after another as occurring simultaneously, resulting in phase misalignment in the thermal field response. In terms of implementation, each type of data can be written into the buffer according to the sampling period. For example, vehicle speed, drive current, and weather interface can be recorded as sampling points at times T1, T2, and T3, respectively. The processor will stitch together the data closest to the same time into a set of boundary frames based on a unified clock. For example, in a simplified example, if the vehicle speed data update time is 20:15:01.100, the LED drive status is 20:15:01.120, and the weather interface refresh is 20:15:01.080, then the system can align these three to the boundary frame of 20:15:01 for subsequent thermal field solving. The key here is not to be precise to a certain mathematical value, but to ensure that the heat source, heat dissipation, and environmental changes belong to the same engineering event chain. Specifically, if the meteorological interface is temporarily unavailable, the system can prioritize using local data such as vehicle-mounted rainfall, outside temperature, humidity, and light intensity changes as a substitute; if there is clock drift between telemetry data and electrical data, clock recalibration will be performed or the most recent reliable sample will be used to complete the data. If a certain type of key data is missing for more than a set period of time, the period will be marked as a low confidence boundary input. Subsequently, only the risk trend will be output instead of directly triggering aggressive compensation, so as to avoid miscontrol due to input distortion. For example, during the stage when the logistics vehicle exits the tunnel and enters the rain area, the system simultaneously receives information that the vehicle speed increases from low speed to high speed, the driving current in the lamp maintains a high lighting level, and the weather interface returns information that there is heavy rainfall two kilometers ahead. After timestamp alignment, these data are identified as boundary changes under the same continuous working conditions, which can accurately describe the thermal shock background of sudden strong cooling of the outer surface under high heat conditions. The purpose of this step is to establish an input basis consistent with the real road conditions for subsequent digital twin solutions, thereby achieving synchronous correspondence between thermal field prediction and actual vehicle lighting conditions.

[0022] In this embodiment of the invention, the specific construction process of the reduced-order model includes: acquiring high-fidelity three-dimensional fluid dynamics simulation data; performing intrinsic orthogonal decomposition on the high-fidelity three-dimensional fluid dynamics simulation data to extract spatial basis functions; constructing a projection matrix based on the spatial basis functions; and using the projection matrix to reduce the dimensionality of the full-order thermodynamic equations to generate the reduced-order model.

[0023] This embodiment provides a mechanism for constructing a reduced-order model. Specifically, if the aforementioned scheme directly uses full-size three-dimensional fluid-thermal coupling simulation for real-time calculation, it will be difficult to meet the timeliness requirements during vehicle operation. Especially in the case of continuous vehicle driving and rapid changes in boundary conditions, if a complete high-fidelity model is called in each control cycle, there will usually be a solution lag, resulting in the output results being only suitable for offline analysis and difficult to use for online compensation. Therefore, this embodiment extracts spatial basis functions that can represent the main thermal field change laws by performing intrinsic orthogonal decomposition on the high-fidelity simulation results, and establishes a reduced-order model accordingly. Specifically, the high-fidelity three-dimensional fluid dynamics simulation data comes from the offline simulation library during the development of the vehicle lights. This simulation library covers typical working conditions, such as low-temperature start-up, high-temperature idling, high-speed driving, rain cooling, and repeated on / off lighting modes. Each set of simulations contains complete information such as airflow inside the lamp, heat conduction of the radiator, lens heating, and convection heat transfer of the lamp housing. The system extracts common principal modes from a large number of simulation snapshots, which essentially identifies which temperature distribution patterns recur and contribute the most to the overall thermal behavior. After constructing the projection matrix, subsequent online calculations no longer require independent calculation of each grid point, but only solve the combination coefficients of a small number of principal modes, thereby quickly reconstructing an approximate three-dimensional transient temperature field. Specifically, the discrete temperature field snapshots obtained from high-fidelity fluid-thermal simulation are arranged into a snapshot matrix, its covariance matrix is ​​calculated and eigenvalue decomposition is performed, and the eigenvectors corresponding to the first few largest eigenvalues ​​are extracted as spatial basis functions. These basis functions represent the dominant thermal field modes with the highest energy in the system. The full-order nonlinear heat conduction and convection-diffusion equations are projected onto the low-dimensional subspace spanned by these basis functions to obtain a set of ordinary differential equations with respect to the time correlation coefficient. During online calculation, only the convective heat transfer coefficient and heat source power at the current moment and other boundary conditions need to be input. The time correlation coefficient can be quickly derived by solving the low-dimensional equation system, and then linearly combined with the spatial basis function to break the time-consuming barrier of direct iteration of high-density grid and reconstruct transient three-dimensional temperature field data. To facilitate understanding, a typical control model example can be used to illustrate the data flow; assume that there are three representative thermal field patterns in the offline sample: mode M1 corresponds to steady-state cruise heating, mode M2 ​​corresponds to external cold and internal heat when encountering sudden cold rain, and mode M3 corresponds to heat dissipation obstruction when low-speed congestion. The system extracts these three dominant spatial distributions from historical high-fidelity data. When facing new boundary frames in the online phase, it only needs to determine which combination of modes the current state is closer to in order to quickly recover the temperature trend of most areas inside the headlights. The emphasis here is on the compression of engineering mechanisms, rather than simple data fitting, because each mode comes from the real heat conduction and convection laws. Regarding the anomaly protection mechanism, if the online operating conditions exceed the coverage of the offline samples, such as later hardware modifications to the headlight assembly, radiator blockage leading to distorted flow conditions, or the vehicle operating in an extreme high-altitude environment for a long time, the system can mark the current operating conditions as an extrapolation interval and increase the weight of the measured temperature feedback in the model correction; when necessary, only the direct temperature measurement results of a limited area are used for control, and the reduced-order results far from the sample domain are not directly used for aggressive compensation; if the offline sample library is updated later, the reduced-order model can be retrained and the version can be switched. For example, before the logistics vehicle enters the long downhill section of the mountain, the system has already called up a sample library covering high-power lighting + high-speed airflow cooling + intermittent rainfall in the background; when the actual boundary frame shows that the vehicle is in a similar combined working condition, the processor quickly calls up the corresponding main mode to complete the temperature field deduction without having to rerun the complete high-fidelity fluid-heat simulation. The purpose of this step is to significantly reduce the size of the online solution while maintaining the ability to reconstruct the main thermal field features, thereby enabling real-time twin computing under vehicle or cloud conditions.

[0024] In this embodiment of the invention, the material parameters of the key structural component are obtained, the temperature gradient characteristics are input into a preset solid mechanics mapping model, and the real-time thermal deformation of the key structural component is calculated, including: calculating the three-dimensional temperature difference field inside the key structural component based on the temperature gradient characteristics. The thermal expansion coefficient of the key structural component is obtained as the material parameter; the thermal strain field is obtained by multiplying the three-dimensional temperature difference field with the thermal expansion coefficient of the material, and the three-dimensional displacement field data is generated by solving the constitutive equation of elasticity in combination with the preset structural constraint boundary conditions; the surface node displacement vector is extracted from the three-dimensional displacement field data, and the surface node displacement vector is used as the real-time thermal deformation.

[0025] This embodiment provides a thermal deformation mapping mechanism; specifically, simply knowing where the lamp body is hot and where it is cold is not enough to directly determine whether the light pattern is inaccurate, because what truly changes the optical path is the change in shape and position of the structural components after they are heated; While the aforementioned thermal field solution can identify the temperature difference distribution, it is difficult to explain why some car lights can maintain stable beam patterns at the same maximum temperature, while others show obvious focus drift, if material expansion characteristics, assembly constraints and structural coupling are ignored. Therefore, this embodiment further converts the temperature gradient into structural deformation. Specifically, the system obtains the three-dimensional temperature difference field inside the key structural components based on the transient temperature field; the difference here can be understood as the degree of thermal non-uniformity at different locations within the same structural component, rather than just looking at the average temperature; and reads the corresponding material parameters, such as the thermal expansion characteristics of PC lenses, the expansion characteristics of aluminum heat sinks, and the stiffness characteristics of steel brackets. Since the headlight is not an isolated free body, there are fixed points between the lens and the bracket, and there is an assembly fit between the reflector and the housing. Some areas can expand while others are restricted. Therefore, structural constraint boundary conditions must be applied to the system. After solving the solid mechanics problem, a three-dimensional displacement field can be obtained, and the displacement vectors of the relevant nodes of the optical surface can be extracted from it as the real-time thermal deformation. From an engineering perspective, thermal deformation does not necessarily manifest as overall magnification; more commonly, it manifests as localized warping and orientation shifts. To give a simplified example, if there are fixed points around the lens edge, and the temperature rise in the upper left region is significantly higher than that in the lower right region, the solution will not show the entire lens moving forward uniformly. Instead, it is more likely to show a localized lift in the upper left corner and a slight wobbling of the optical axis. Similarly, if there is a large temperature difference near the support feet of the reflector, it will cause changes in the surface shape, causing the originally convergent beam to diverge. Only after the system extracts these nodal displacements can subsequent ray tracing accurately reflect the changes in light distribution. Specifically, if a structural component lacks complete material parameters, the system can call a calibration template of similar materials and reduce the calculation confidence level; if the assembly constraint information is incomplete, conservative constraint boundaries can be used first to avoid underestimating the deformation risk. If the temperature field changes very little within a certain period, the displacement field can maintain the state of the previous period to reduce calculation jitter; for cases exceeding the material's allowable working range, in addition to outputting deformation data, the system can also add a material instability warning, indicating that the region may no longer be suitable for using the linear elastic approximation. For example, after the logistics vehicle continuously activated the high beam assist mode and passed through a rainy area, the system detected that the outer surface of the lens cooled down faster than the inner support ring. The support ring, in turn, restricted the lens from retracting freely through a buckle. As a result, the solid mechanics mapping result showed a slight concave deformation of the lens surface, and the displacement of the upper edge node was greater than that of the lower edge. This deformation was sent into a unified coordinate system to reconstruct the current deformation geometry model. The purpose of this step is to transform the temperature difference in the thermal field into a geometric deviation that actually plays a role in optics, thereby enabling a traceable mapping from thermal problems to structural problems.

[0026] In this embodiment of the invention, the deformable geometry model is imported into a preset ray tracing algorithm to calculate the light source focus offset of the LED light-emitting array, including: extracting the optical surface profile data of the deformable geometry model; reconstructing the optical tracing boundary conditions based on the optical surface profile data; and generating a virtual ray set. Perform forward ray tracing calculations on the virtual ray set according to the optical tracing boundary conditions to obtain the actual focal plane spot centroid coordinates; calculate the spatial vector difference between the actual focal plane spot centroid coordinates and the preset theoretical focal plane spot centroid coordinates, and use the spatial vector difference as the light source focus offset.

[0027] This embodiment provides a focal offset calculation mechanism; specifically, if the above thermal deformation results only remain at the displacement field level, they still cannot directly serve the light distribution safety assessment, because the road lighting effect is ultimately determined by the light propagation path; Without optical layer mapping, the system can only know that the structure has changed, but not where the beam has deflected. Therefore, in this embodiment, the deformed geometric model is imported into the ray tracing algorithm to reconstruct the optical boundary conditions and obtain the focal offset. Specifically, the system first extracts surface data directly related to imaging from the deformation geometry model, such as the change of lens surface normal, the change of local curvature of the reflecting surface, and the change of the relative pose of the light-emitting array with respect to the lens. Based on the updated surface, the optical tracking boundary conditions are reconstructed. The boundary conditions here refer to the geometric state upon which the incident, refracted, and reflected paths of light on each optical surface depend. The system generates a set of virtual rays to approximate the main beam direction emitted by each LED pixel unit, and then performs forward ray tracing to obtain the centroid position of the actual focal plane. The centroid of the theoretical focal plane can be derived from the design calibration value, and the spatial vector difference between the two represents the focal offset. This can be illustrated with a very simple example; suppose that under calibration, the theoretical focal point of a certain LED subarray is located at the center point C0 of the focal plane; when the upper edge of the lens warps slightly, the ray tracing result falls on C1, then the vector from C0 to C1 is the focal point offset; this vector does not require the disclosure of complex geometric formulas, its engineering significance is clear: upward offset means that oncoming glare may increase, downward offset means that the effective illumination distance is reduced, and lateral offset means that lane coverage is asymmetrical; Specifically, if the deformation geometry model has local missing meshes within a certain period, the system can continuously repair the adjacent surfaces before performing ray tracing; if ray tracing finds that the focal offset directions of multiple subarrays are too dispersed, it indicates that there may be complex surface distortion at that moment, and the system can switch to a more conservative safety criterion, such as prioritizing the illumination risk of the sensitive area above; if the theoretical focal plane reference changes due to lamp body calibration updates, the calibration reference is updated synchronously to avoid misjudging the recalibrated state after manufacturing tolerances or subsequent repairs as thermal deformation; For example, after the logistics vehicle experienced a long downhill braking and a sudden rain, the surface shape of the upper lens of the headlight changed slightly; the system reconstructed a new refractive boundary on the deformation model, and the tracking results showed that some of the upper edge rays were lifted relative to the calibration state, and the centroid of the actual focal plane spot shifted upward; this shift was recorded as a direct input for subsequent light distribution projection and safety comparison; The purpose of this step is to further transform structural deformation into quantifiable optical consequences, thereby enabling real-time identification of focus shift risks.

[0028] In this embodiment of the invention, generating light distribution projection data based on the light source focus offset includes: acquiring a preset virtual test screen model; and performing spatial translation and rotation transformations on a preset light source luminous intensity distribution matrix based on the light source focus offset to generate an offset luminous intensity distribution matrix. The offset luminous intensity distribution matrix is ​​projected onto the virtual test screen model, and the illuminance value of each pixel node on the virtual test screen model is calculated; the illuminance values ​​of each pixel node are combined to generate the light distribution projection distribution data.

[0029] This embodiment provides a light distribution projection distribution generation mechanism; specifically, obtaining only the focus offset direction and magnitude is insufficient to determine lighting safety in a meaningful sense, because whether the light distribution is qualified is usually reflected in whether the illuminance distribution, cutoff line clarity, and sensitive point brightness on the test screen meet the requirements; therefore, this embodiment further converts the focus offset amount into light distribution projection distribution data on a virtual test screen; Specifically, the system calls a preset virtual test screen model; this screen model can be consistent with the lamp testing or internal enterprise calibration method, reflecting the spatial translation and rotation transformation of the original luminous intensity distribution on the illuminance receiving surface at a certain distance in front of the vehicle according to the focal offset. Thermal deformation does not change the total amount of light energy out of thin air, but it will change the direction of the light beam and the distribution of energy on the screen. Therefore, the light flux that was originally concentrated in the target illumination area may shift upward, to the side, or to the edge. The system calculates the illuminance value of each node on the virtual screen and then combines them to generate a complete light distribution map. For ease of understanding, a simplified example can be used; suppose the virtual screen is divided into 3×3 areas, where the lower middle area originally corresponds to the main lighting zone, and the upper middle area corresponds to the glare-sensitive zone; When the focal point shifts upwards, the energy in the lower and middle regions will migrate to the upper and middle regions, resulting in a decrease in the illuminance of the road surface ahead and an increase in the illuminance of the sensitive area opposite. The key here is not to disclose the specific mathematical transformation details, but to explain that the system uses a virtual screen to convert the abstract focal point shift into the consequences of road lighting. Specifically, if the luminous intensity distribution matrix comes from historical calibration and the current lamp body has aging light decay, the system can superimpose a real-time luminous flux correction coefficient; If the calculation results of certain pixel nodes are greatly affected by local occlusion or rain and fog scattering, an environment correction layer can be introduced during the projection stage. However, when the environment model is unreliable, the main trend judgment brought about by geometric offset should still be retained first. If the virtual screen resolution is reduced to adapt to real-time calculation, high resolution can be maintained for key test areas and coarser division can be used for non-sensitive areas. For example, in the scenario of a logistics vehicle going downhill in the rain, the system reprojects the original near-beam luminous intensity map onto the virtual test screen based on the focus shift result; the projection result shows that the bright spots in the upper sensitive area are enhanced, while the effective lighting area below the cutoff line is slightly shifted and diffused, and the generated light distribution projection data then enters the next step of safety threshold comparison; The purpose of this step is to translate the change in optical focus into a detectable and identifiable screen illuminance distribution, thereby realizing the transformation from geometric optical parameters to light distribution results.

[0030] In this embodiment of the invention, comparing the light distribution projection data with a preset light distribution safety threshold includes: extracting the illuminance gradient value of the cutoff line from the light distribution projection data as a cutoff line feature parameter, and extracting the illuminance value of the test point under preset spatial coordinates; comparing the cutoff line feature parameter with a preset cutoff line sharpness threshold; and comparing the test point illuminance value with a preset glare illuminance upper limit threshold to determine whether any of the test point illuminance values ​​is higher than the glare illuminance upper limit threshold. In response to the fact that the characteristic parameters of the light and dark cutoff line are not lower than the cutoff line sharpness threshold and the illuminance values ​​of all the test points are not higher than the glare illuminance upper limit threshold, it is determined that the light distribution projection distribution data exceeds the light distribution safety threshold; in response to the fact that the characteristic parameters of the light and dark cutoff line are not lower than the cutoff line sharpness threshold and the illuminance values ​​of the test points are not higher than the glare illuminance upper limit threshold, it is determined that the light distribution projection distribution data is included within the light distribution safety threshold.

[0031] This embodiment provides a light distribution safety determination mechanism. Specifically, the aforementioned projection distribution only describes where the light falls, but actual control requires a clear safety conclusion. Without determination logic, the system cannot decide whether to maintain the current state or perform compensation. Furthermore, simply looking at the total illuminance is not enough, because nighttime driving safety is highly dependent on the clarity of the cutoff line and whether glare occurs in the sensitive area above it. Therefore, this embodiment introduces two types of criteria: cutoff line characteristic parameters and test point illuminance values. Specifically, the system extracts the illuminance gradient value of the cutoff line from the light distribution projection to characterize whether the transition between the bright and dark areas is steep. Physically, the clearer the cutoff line, the more stable the beam boundary control, and the less likely the oncoming driver is to be disturbed by stray uplight. Meanwhile, the system also extracts the illuminance values ​​of test points under preset spatial coordinates. These test points are usually located in sensitive areas of concern to enterprise standards and are used to directly reflect glare risks. When the cutoff line characteristic parameter is lower than the threshold, it indicates that the beam boundary is blurred and that even if the total luminous flux remains unchanged, upward spillage may occur. When the illuminance of the test point is higher than the upper limit, it means that some sensitive areas are over-illuminated and should be judged as having safety risks. A specific test grid example can be used to illustrate the judgment logic; assuming a cutoff line area G and a sensitive test point P are set on the screen: if the transition between light and dark in area G becomes slower, it indicates that light spot diffusion occurs in the cutoff line transition area, resulting in a decrease in clarity; if the brightness of point P increases significantly, it indicates that the glare above increases. The system adopts or determines the relationship; if any risk occurs, the current light distribution is deemed to be out of limit. This determination method conforms to the conservative principle in vehicle headlight safety control, because cutoff line distortion and excessive glare can both cause road safety problems on their own. For the fault tolerance mechanism for environmental interference and data loss, if the cutoff line area is blurred as a whole due to rain and fog scattering, but the sensitive test points have not risen, the system can reduce the probability of false alarms by combining the environmental conditions. Conversely, if the cutoff line is still relatively clear but some test points above exceed the limit, the conclusion of exceeding the limit should still be maintained, because this means that there may be local beam upward; if the reliability of the beam distribution diagram is low due to missing input, the system can temporarily not output a qualified conclusion, but enter a pending confirmation state and prioritize conservative heat dissipation control. For example, after the logistics vehicle enters a section of road with continuous rainfall, the system extracts from the virtual screen results that the edge of the low beam cutoff line becomes gentler, and at the same time, it detects an increase in illuminance at the test point above the corresponding lane in the opposite direction; although the main lighting area still has a certain brightness, due to the glare trend, the system determines that the current light distribution exceeds the safety threshold. The purpose of this step is to identify the risk of thermally induced optical failure by using road light distribution quality as a direct criterion, thereby enabling evidence-based triggering of control actions.

[0032] In this embodiment of the invention, generating a first control command includes: calculating a target power derating factor and a target heat dissipation rotation speed based on the difference between the light distribution projection data and the light distribution safety threshold using a preset proportional-integral-derivative control logic; packaging the target power derating factor into a light-emitting array adjustment command; packaging the target heat dissipation rotation speed into a heat dissipation actuator adjustment command; and combining the light-emitting array adjustment command and the heat dissipation actuator adjustment command to generate the first control command. This embodiment provides an active compensation control mechanism for over-limit scenarios. Specifically, the aforementioned scheme can identify the risk of light distribution exceeding limits, but if only an alarm is output without forming a control closed loop, thermal deformation may still increase as the vehicle continues to drive, resulting in persistent glare or insufficient lighting. Furthermore, if the global indiscriminate reduction of luminous power is simply adopted, the driver's forward lighting may be sacrificed. Therefore, this embodiment converts the light distribution deviation into a combined control command of luminous array adjustment and heat dissipation actuator adjustment. Specifically, the system calculates the target power derating factor and target heat dissipation speed based on the difference exceeding the safety threshold through preset control logic; its engineering meaning is: when the main problem comes from an excessively strong heat source and the structural temperature difference is still expanding, the heat dissipation intensity should be increased and the heat load should be appropriately reduced. When the main problem is that the beam rises in the sensitive area above, the corresponding LED subarray can be derating locally instead of reducing the overall brightness. Proportional adjustment reflects a rapid response to the current deviation, integral adjustment helps to handle continuous deviations, and derivative adjustment is used to suppress overshoot when changes are too rapid. In practice, the system sets the excessive value of the light distribution illuminance as the error input and sends it to the derating control loop of the light-emitting array and the speed-up control loop of the heat dissipation actuator respectively. For the proportional loop, the system directly calculates the basic derating ratio and the fan speed-up percentage based on the excessive value. The greater the excessive value, the stronger the adjustment force applied immediately. For the integral phase, the system continuously accumulates the out-of-limit residuals from the historical period. When the local temperature rise is difficult to suppress quickly by the fan, causing the deviation to persist, the integral term will cause the depreciation coefficient to increase further to avoid long-term inaccuracy. For the differential stage, the system calculates the rate of change of the current over-limit value. If it finds that the over-limit amplitude is rapidly converging, it reduces the control output in advance to prevent excessive power reduction that could lead to a precipitous drop in lighting. Specifically, the difference between the current period's light distribution projection data and the safety threshold is recorded as the error input. The target power derating factor is calculated using the following formula. :

[0033] in, , and These represent the preset proportional gain, integral gain, and derivative gain, respectively, where t is the current time. The system encapsulates power adjustment as an illuminator adjustment command and heat dissipation requirements as a heat dissipation actuator adjustment command, and combines them into a unified control message to send to the control unit. Similarly, the system calculates the target cooling speed compensation amount through an independent cooling control loop. The formula is:

[0034] in, , and These represent the preset proportional gain, integral gain, and derivative gain of the heat dissipation control loop, respectively. The system compares the base speed with the speed compensation amount. The combined values ​​are used as the final target heat dissipation speed. This can be illustrated through specific driver configuration examples. Assuming the current risk is mainly concentrated in the upper area, the control strategy can derate subarray A and increase the fan speed, while keeping subarray B unchanged, thereby suppressing glare without significantly sacrificing the main lighting strip. For example, when the cutoff line blurring is due to excessive overall temperature rise, a gentle derate can be applied to the entire array while simultaneously increasing the heatsink speed. The core here is to transfer the consequences of thermal-mechanical-optical failure back to executable lighting control actions, rather than simply using excessive temperature as the control basis. Regarding degradation strategies in fault modes, if the heat dissipation actuator is in a faulty or limited state, the system can increase the derating ratio of the light-emitting array and output a maintenance mark; if the current vehicle speed is very low and the environment is extremely hot, and the heat exchange cannot be significantly improved even after the fan is upgraded, the system should prioritize the use of a conservative lighting mode to avoid further damage to the devices. If the light distribution deviation is small and close to the threshold boundary, frequent power increases and decreases can be avoided by controlling the hysteresis zone, thereby reducing user-perceived flicker and actuator lifespan loss. For example, in a mountainous rainy night scenario for the logistics vehicle, the system determines that the illuminance of the sensitive area above the low beam exceeds the allowable range, while the high temperature inside the lamp body has not yet subsided; The processor then generates the first control instruction: on the one hand, to implement controlled derating of the relevant LED subarray at the upper edge, and on the other hand, to increase the speed of the cooling fan to a higher speed in order to reduce the temperature difference between the lens and the bracket as soon as possible; after the control unit executes the instruction, the light distribution gradually returns to the safe range in the following cycles; The purpose of this step is to immediately establish a dual-channel compensation closed loop for the thermal state of the lamp body and the light distribution results after identifying the risks, so as to achieve linkage adjustment that controls both heat and light.

[0035] In this embodiment of the invention, when the processor executes the computer program, it further performs the following steps: constructing a historical thermal cycle database; continuously writing the transient three-dimensional temperature field data into the historical thermal cycle database; and extracting temperature alternation amplitude features and temperature alternation frequency features from the historical thermal cycle database. The location information of the target weld point on the key structural component is extracted from the deformation geometry model; a thermal fatigue life prediction model is constructed based on the preset material fatigue curve or the Coffin-Manson equation; the temperature alternation amplitude feature, the temperature alternation frequency feature and the location information of the target weld point are input into the thermal fatigue life prediction model; the cumulative fatigue damage degree of the target weld point is calculated; and the remaining service life prediction value of the target weld point is derived based on the cumulative fatigue damage degree. If the predicted remaining service life is lower than a preset service life warning threshold, a maintenance prompt message is generated and output; if the predicted remaining service life is not lower than the service life warning threshold, the monitoring status is maintained.

[0036] This embodiment provides a thermal fatigue life prediction mechanism for target solder joints. Specifically, the aforementioned solution mainly addresses the light distribution safety issue at the current moment, but for vehicles in long-term service, real-time compensation alone is still insufficient. Especially in high-frequency nighttime operation scenarios such as logistics vehicles, the headlights repeatedly experience cycles of lighting up, high load, sudden environmental changes, and shutdown. The LED substrate solder joints, driver connection solder joints, and other parts are prone to accumulated damage due to long-term thermal expansion and contraction. If the remaining lifespan cannot be predicted in advance, sudden loss of light, intermittent failure, or local pixel malfunction may occur in actual use. Specifically, the system constructs a historical thermal cycle database and continuously writes transient three-dimensional temperature field data for each cycle; through long-term recording, the system can identify the magnitude of temperature alternation experienced by certain key areas and how frequently such alternation occurs. In engineering, the greater the temperature alternation amplitude, the more intense the material expansion and contraction; the higher the alternation frequency, the more thermomechanical cycles the weld joint undergoes per unit time; the system then combines the target weld joint location information in the deformation geometry model to identify whether the weld joint is located in a high-stress region or a relatively mild region; using a preset fatigue curve or thermal fatigue life model, the cumulative damage degree is estimated and the remaining service life prediction value is derived. In the specific business logic of fatigue damage deduction based on the Coffin-Manson equation, the system converts the extracted temperature alternation amplitude characteristics into the plastic strain range experienced by the target weld point. According to the Coffin-Manson mechanism, low-cycle thermal fatigue life is mainly controlled by the plastic strain amplitude. The system substitutes the obtained plastic strain range into the following formula:

[0037] Therefore, the maximum number of reversals that the weld joint can withstand under the current thermal alternation state can be determined. ,in The fatigue ductility coefficient of the material is represented by c, and the fatigue ductility index is represented by φ. The system extracts the temperature alternation frequency characteristics in actual operation and counts the actual number of thermal cycles under the operating condition i. The fatigue damage degree at the current stage is calculated using Miner's linear cumulative damage rule.

[0038] in, This represents the limit cycle number calculated using the aforementioned equation under operating condition i; the total damage is obtained by summing the damage levels at each stage throughout the entire lifespan.

[0039] Where m is the total number of working conditions experienced throughout the entire life cycle. When the total damage degree approaches the safety threshold, for example (D=1), the corresponding expected remaining working cycle can be used as the predicted value of the remaining service life. A simplified example can be used to illustrate this: if solder joint W1 is located at the edge of the LED substrate and near a heat source, it will experience a cycle of significant temperature rise followed by rapid external cooling by cold rain during each nighttime task, while solder joint W2 is located in a relatively neutral area, so the damage accumulation of W1 will be faster; the system does not simply record the highest temperature, but pays more attention to the amplitude and rhythm of repeated temperature fluctuations, because fatigue failure is often caused by cyclic strain, rather than determined by a single high temperature. Specifically, if historical data for a certain period is incomplete, the system can maintain a conservative estimate and indicate a decrease in the reliability of the life prediction; if the vehicle has had its lamp module replaced or its solder joints repaired, database segmentation management should be implemented to avoid mixing and accumulating the thermal history of new and old components. If the difference between the lifespan model and the actual repair statistics increases, the fatigue curve can be corrected through later calibration. However, before correction, the system should still output maintenance prompts with a more conservative warning threshold. For example, the logistics vehicle has been carrying out cross-regional nighttime transportation for a long time, and has experienced alternating working conditions such as idling in high-temperature service areas in summer, high-speed cruising at night, and cooling in mountainous areas during heavy rain in the past three months; the historical thermal cycle database shows that the area where a certain LED driver solder joint is located has been in a state of high alternating amplitude and high alternating frequency for a long time. Based on this, the system determined that its remaining service life was close to the warning threshold, and therefore output a maintenance prompt on the fleet operation and maintenance platform and the driver's cab diagnostic terminal, suggesting that the corresponding module be checked or replaced at the next maintenance window; The purpose of this step is to extend the real-time thermal twin capability to the entire life cycle of reliability assessment, thereby enabling the extension of vehicle lights from immediate control to predictive maintenance.

[0040] In this embodiment of the invention, the reduced-order model, the solid mechanics mapping model, and the ray tracing algorithm are all deployed on a cloud simulation computing platform. The multi-source dynamic boundary condition data is continuous time series data obtained through the vehicle network communication interface. The first control command and the second control command are sent to the control unit through the controller local area network bus.

[0041] This embodiment provides a mechanism for cloud-based collaborative deployment and vehicle-side execution. Specifically, as vehicle headlight digital twin models gradually incorporate three types of calculations—thermal field, structural deformation, and optical tracking—if all of these are deployed on a single vehicle-side control unit, they may be limited by onboard computing power, storage space, and heat dissipation capabilities. This is especially true for large-scale fleet operation scenarios, where model version updates and lifetime database management can become difficult. Therefore, in this embodiment, the reduced-order model, solid mechanics mapping model, and ray tracing algorithm are deployed on a cloud simulation computing platform, while the vehicle is responsible for data acquisition, communication, local caching, and control command execution. Specifically, the vehicle uploads multi-source dynamic boundary condition data to the cloud platform in a continuous time series manner through the vehicle-to-everything (V2X) communication interface; the cloud platform completes transient thermal field solution, thermal deformation mapping and optical tracking based on stronger computing resources, and sends back the light distribution safety conclusions and corresponding control commands to the vehicle. After receiving the control command, the vehicle sends it to the headlight control unit via the controller local area network bus to achieve rapid adjustment of the LED array and heat dissipation actuator. This architecture can utilize the high computing power of the cloud while preserving the real-time performance of the vehicle. For headlights of the same model, the cloud can also uniformly maintain the model version, environmental sample library and fatigue life database, which facilitates the reuse of experience across vehicles. To illustrate the data link, a simplified example can be used: after the boundary frame F1 is formed at the vehicle end, it is uploaded to the cloud, and the cloud returns the result R1, which includes three types of fields: light distribution safety risk warning and control recommendations. The vehicle only needs to parse the control fields in R1 and send them to the control unit via the bus. If the next frame F2 is delayed in communication, the vehicle can use the effective control state of the previous frame first, and then switch smoothly after the result in the cloud is restored. This can avoid communication jitter directly causing frequent changes in lighting mode. Specifically, if vehicle-to-everything (V2X) communication is interrupted for a short period of time, the vehicle can use a local simplified model or a single valid model result to maintain basic monitoring; if the cloud return delay exceeds the control time limit, only a conservative strategy will be executed in the current cycle, without waiting for expired results; if bus transmission fails, the vehicle should record the fault code and switch to the lamp's own thermal protection mode. For critical controls related to driving safety, the vehicle can be set to the highest priority channel, so that control commands take precedence over non-critical diagnostic data transmission. For example, in the above-mentioned cross-provincial nighttime transportation task of logistics vehicles, the vehicle continuously uploads the vehicle speed, rainfall, in-light driving status and environmental parameters to the cloud simulation platform; after the cloud completes the thermo-mechanical-optical joint solution, it determines that the current vehicle light has a blurred cutoff line trend and generates new light-emitting array adjustment instructions and cooling fan speed instructions. The command is transmitted back to the vehicle via the vehicle network, and then sent to the headlight control unit via the controller area network bus for execution. If there is no signal coverage in the mountainous area for a short period of time, the vehicle will continue to operate according to the most recent effective safety policy until communication is restored. The purpose of this step is to balance the computational power of complex models, fleet-level unified management, and vehicle-side control reliability, thereby achieving stable collaboration between digital twin analysis and actual vehicle lighting execution.

[0042] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins, characterized in that, Includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to perform the following steps: Collect multi-source dynamic boundary condition data of motor vehicle LED lights during operation: The motor vehicle LED lights include key structural components, LED light-emitting arrays, heat dissipation actuators, and control units; The multi-source dynamic boundary condition data is input into a preset reduced-order model to solve the transient three-dimensional temperature field data. The reduced-order model is constructed by processing the preset full-order thermodynamic equations through intrinsic orthogonal decomposition. Extract the temperature gradient features from the transient three-dimensional temperature field data; The material parameters of the key structural component are obtained, the temperature gradient characteristics are input into a preset solid mechanics mapping model, and the real-time thermal deformation of the key structural component is calculated. Based on a unified spatial coordinate system, the real-time thermal deformation is mapped to the corresponding mesh nodes of the preset initial geometric model of the key structural component to generate a deformation geometric model. The deformation geometry model is imported into a preset ray tracing algorithm to calculate the light source focus offset of the LED light-emitting array; Based on the light source focus offset, generate light distribution projection data; The light distribution projection data is compared with a preset light distribution safety threshold. In response to the light distribution projection data exceeding the light distribution safety threshold, a first control command is generated and sent to the control unit for optical compensation; In response to the light distribution projection data being within the light distribution safety threshold, a second control command is generated and sent to the control unit to maintain the current state.

2. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The collection of multi-source dynamic boundary condition data of vehicle LED lights during operation includes: Acquire real-time telemetry data of the vehicle's LED lights; Acquire the real-time electrical operating data of the vehicle's LED headlights; Obtain meteorological interface data of the environment in which the vehicle's LED headlights are located; The real-time telemetry data, the real-time electrical operation data, and the meteorological interface data are time-stamp aligned to generate the multi-source dynamic boundary condition data.

3. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The specific construction process of the reduced-order model includes: Obtain high-fidelity three-dimensional fluid dynamics simulation data; The high-fidelity three-dimensional fluid dynamics simulation data were subjected to intrinsic orthogonal decomposition to extract spatial basis functions; Construct the projection matrix based on the spatial basis functions; The full-order thermodynamic equation is reduced in dimension using the projection matrix to generate the reduced-order model.

4. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The process of obtaining the material parameters of the key structural component, inputting the temperature gradient characteristics into a preset solid mechanics mapping model, and calculating the real-time thermal deformation of the key structural component includes: Based on the temperature gradient characteristics, the three-dimensional temperature difference field inside the key structural component is calculated. The coefficient of thermal expansion of the material of the key structural component is obtained as the material parameter. The thermal strain field is obtained by multiplying the three-dimensional temperature difference field with the thermal expansion coefficient of the material. Combined with the preset structural constraint boundary conditions, the three-dimensional displacement field data is generated by solving the constitutive equation of elasticity. Surface node displacement vectors are extracted from the three-dimensional displacement field data, and the surface node displacement vectors are used as the real-time thermal deformation.

5. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The step of importing the deformation geometry model into a preset ray tracing algorithm to calculate the light source focus offset of the LED light-emitting array includes: Extract the optical surface profile data of the deformable geometry model; Reconstruct the optical tracing boundary conditions based on the optical surface profile data; Generate a set of virtual rays; Perform forward ray tracing calculations on the virtual ray set according to the optical tracing boundary conditions to obtain the centroid coordinates of the actual focal plane spot; Calculate the spatial vector difference between the actual focal plane spot centroid coordinates and the preset theoretical focal plane spot centroid coordinates, and use the spatial vector difference as the light source focus offset.

6. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The step of generating light distribution projection data based on the light source focal offset includes: Obtain a preset virtual test screen model; Based on the light source focal offset, a spatial translation and rotation transformation is performed on the preset light source luminous intensity distribution matrix to generate an offset luminous intensity distribution matrix; The offset luminous intensity distribution matrix is ​​projected onto the virtual test screen model, and the illuminance value of each pixel node on the virtual test screen model is calculated. The illuminance values ​​of each pixel node are combined to generate the light distribution projection data.

7. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The step of comparing the light distribution projection data with a preset light distribution safety threshold includes: The illuminance gradient value of the cutoff line is extracted from the light distribution projection data as a feature parameter of the cutoff line, and the illuminance value of the test point under the preset spatial coordinates is extracted; the feature parameter of the cutoff line is compared with the preset cutoff line sharpness threshold; the illuminance value of the test point is compared with the preset glare illuminance upper limit threshold to determine whether there is any illuminance value of the test point that is higher than the glare illuminance upper limit threshold. In response to the fact that the characteristic parameter of the cutoff line is lower than the cutoff line sharpness threshold, or that the illuminance value of any of the test points is higher than the glare illuminance upper limit threshold, it is determined that the light distribution projection distribution data exceeds the light distribution safety threshold. In response to the fact that the characteristic parameters of the cutoff line are not lower than the cutoff line sharpness threshold and all the illuminance values ​​of the test points are not higher than the glare illuminance upper limit threshold, it is determined that the light distribution projection distribution data is included within the light distribution safety threshold.

8. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The generation of the first control command includes: Based on the difference between the light distribution projection data and the light distribution safety threshold, the target power derating factor and the target heat dissipation speed are calculated through a preset proportional-integral-derivative control logic. The target power derating factor is packaged into a light-emitting array adjustment command; The target heat dissipation speed is packaged into a heat dissipation actuator adjustment command; The first control command is generated by combining the light-emitting array adjustment command and the heat dissipation actuator adjustment command.

9. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, When the processor executes the computer program, it also performs the following steps: Build a historical thermal cycle database; The transient three-dimensional temperature field data is continuously written into the historical thermal cycle database; Extract temperature alternation amplitude and temperature alternation frequency features from the historical thermal cycle database; Extract the position information of the target weld points preset on the key structural component from the deformation geometry model; A thermal fatigue life prediction model is constructed based on a preset material fatigue curve or the Coffin-Manson equation. The temperature alternation amplitude characteristics, the temperature alternation frequency characteristics, and the location information of the target weld point are input into the thermal fatigue life prediction model to calculate the cumulative fatigue damage degree of the target weld point. Based on the cumulative fatigue damage degree, the remaining service life prediction value of the target weld point is derived. In response to the predicted remaining service life being lower than a preset service life warning threshold, a maintenance prompt message is generated and output. In response to the remaining useful life prediction value being not lower than the useful life warning threshold, the monitoring status is maintained.

10. The simulation and analysis system for thermal field distribution of automotive LED headlights based on digital twins according to claim 1, characterized in that, The reduced-order model, the solid mechanics mapping model, and the ray tracing algorithm are all deployed on a cloud-based simulation computing platform. The multi-source dynamic boundary condition data is continuous time-series data obtained through the vehicle network communication interface. The first control command and the second control command are sent to the control unit through the controller local area network bus.

Images