Dynamic self-adaptive deployment system for vehicle-mounted phased array radar

Abstract

The invention relates to the technical field of vehicle-mounted radars, and particularly discloses a dynamic self-adaptive deployment system for a vehicle-mounted phased array radar. The system comprises an environment perception and data acquisition module, a deployment performance real-time evaluation module, a multi-target dynamic optimization decision module and a deployment parameter execution and verification module. The radar performance of a current deployment scheme is evaluated by collecting multi-dimensional data in a real vehicle body environment, optimal deployment parameters are generated based on a dynamic scene self-adaptive multi-target optimization algorithm, and finally an execution mechanism is driven to complete physical adjustment and verification to form closed-loop optimization. According to the method, the transformation of vehicle-mounted phased array radar deployment from static design to dynamic adaptive optimization is realized, and the environmental adaptability and the performance stability of the radar after actual loading are improved.

Description

Technical Field

[0001] This invention belongs to the field of vehicle-mounted radar technology, specifically relating to a dynamic adaptive deployment system for vehicle-mounted phased array radar. Background Technology

[0002] In the fields of intelligent driving and vehicle networking technologies, onboard sensors are the core foundation for environmental perception, decision-making and planning, and vehicle control. Their performance directly affects the safety and reliability of autonomous driving systems. Millimeter-wave radar, with its advantages of all-weather operation, long detection range, and high speed measurement accuracy, has become an indispensable sensing component, while phased array radar technology, due to its beam agility and multi-target tracking capabilities, is becoming an important development direction for onboard radar.

[0003] The deployment scheme of vehicle-mounted phased array radar is a key factor in determining its actual detection effectiveness. Its basic goal is to maximize its effective detection range and signal quality by optimizing the radar's installation position and attitude parameters on the vehicle body, so as to cover key driving scenarios such as forward collision warning, blind spot monitoring, and intersection assistance.

[0004] Existing technologies typically rely on static deployment designs based on vehicle CAD models and ideal electromagnetic simulation environments, which has limitations. Before deployment, the impact of the actual environment on radar performance cannot be predicted. For example, the electromagnetic scattering characteristics of complex metal and plastic components such as the vehicle's front bumper and grille can severely interfere with the radar beam, leading to increased sidelobe levels, main lobe distortion, or the formation of detection blind spots. Furthermore, the radar's detection performance is highly dependent on its relative position to other surrounding sensors and electronic equipment, making electromagnetic compatibility issues difficult to fully assess in static design.

[0005] In actual driving, changes in vehicle attitude due to load and road conditions, as well as rain, snow, and mud adhering to the vehicle's surface under different weather conditions, all dynamically alter the radar's radiation characteristics. Existing static deployment solutions are completely unable to adapt to these dynamic variables. Therefore, static deployment methods, detached from the real physical environment, often cause the radar's performance after installation in a vehicle to deviate significantly from design expectations, posing a potential safety hazard to the intelligent driving system. There is an urgent need for an efficient deployment solution that can dynamically adapt to complex real-world environments. Summary of the Invention

[0006] The purpose of this invention is to provide a dynamic adaptive deployment system for vehicle-mounted phased array radar, in order to solve the problem that existing radar deployment schemes based on static CAD models and ideal simulations cannot adapt to the complex electromagnetic environment, dynamic attitude changes and environmental interference of real vehicles, resulting in actual detection performance deviating significantly from the design expectations.

[0007] This invention provides a dynamic adaptive deployment system for vehicle-mounted phased array radar. This system operates during the final commissioning phase after vehicle production and in subsequent use and maintenance. Its core lies in constructing a closed-loop feedback and execution system from physical environment perception and real-time performance evaluation to dynamic optimization of deployment parameters. The system includes an environmental perception and data acquisition module, a real-time deployment performance evaluation module, a multi-objective dynamic optimization decision-making module, and a deployment parameter execution and verification module.

[0008] The environmental perception and data acquisition module is used to collect multi-dimensional benchmark data necessary for building a dynamic deployment optimization model under the actual physical state of the target vehicle.

[0009] The environmental perception and data acquisition module integrates a mobile, high-precision near-field scanning probe array, a high dynamic range vehicle-mounted radar target simulator array, and a set of vehicle status and environmental sensors.

[0010] The near-field scanning probe array consists of multiple broadband probes operating in the millimeter-wave band. Driven by a six-degree-of-freedom robotic arm, it performs a three-dimensional spatial scan of the vehicle surface area where candidate radar units are installed, following a preset precise trajectory, in order to collect the raw data of the actual near-field radiation pattern at the radar installation point.

[0011] The vehicle-mounted radar target simulator array consists of multiple distributed radio frequency signal sources and antennas, which are pre-deployed around the vehicle in key azimuth and elevation directions to simulate and generate standard point target echo signals with different distances, speeds and radar cross-sections in a real vehicle environment.

[0012] The vehicle status and environment sensor group includes a high-precision inertial measurement unit, wheel load sensors, optical surface contamination detection sensors, and temperature and humidity sensors, which are used to collect the vehicle's current pitch angle, roll angle, sprung mass, radome surface cleanliness and type of adhering substances, and environmental temperature and humidity parameters in real time.

[0013] All the above-mentioned collected data are strictly synchronized using timestamps and transmitted to the deployment performance real-time evaluation module.

[0014] The deployment performance real-time evaluation module is used to calculate the radar's overall performance index under the current candidate deployment scheme based on real-time multi-dimensional data provided by the environmental perception and data acquisition module.

[0015] The deployment performance real-time evaluation module incorporates a radar performance prediction engine based on actual physical parameters.

[0016] The radar performance prediction engine first performs spherical wave expansion and far-field transformation calculations on the raw radiation pattern data obtained from near-field scanning, and then superimposes the vehicle structure scattering effect based on the measured material electromagnetic parameters provided by the high-precision three-dimensional digital twin model of the vehicle to synthesize the actual far-field radiation pattern of the radar at the actual installation position under the current vehicle attitude and surface state.

[0017] Furthermore, the deployment of the real-time performance evaluation module utilizes the synthesized actual far-field radiation pattern, combined with the standard point target echo signal generated by the radar target simulator, to perform an end-to-end signal processing simulation link.

[0018] The signal processing simulation link fully simulates the radar's transmitted waveform modulation, spatial beamforming, echo reception, signal detection, ranging, velocity measurement, angle measurement, and point convergence process.

[0019] Based on the simulation results, a real-time performance evaluation module is deployed to calculate and output a set of quantitative performance evaluation metrics, which include at least the following: The key focus areas include the main lobe gain decrease, the highest side lobe level, the root mean square error of angle measurement accuracy, the target detection probability and false alarm probability in multi-target simulation scenarios, and the predicted electromagnetic compatibility interference level between the radar and surrounding known vehicle-mounted electronic equipment.

[0020] The multi-objective dynamic optimization decision module receives the set of performance indicators output by the real-time deployment performance evaluation module, compares and calculates them with preset performance thresholds and optimization objectives, and generates the optimal radar deployment parameter adjustment command by solving a multi-objective optimization problem with multiple constraints.

[0021] The core of the multi-objective dynamic optimization decision module is a multi-objective optimization solver with embedded adaptive weight factors. The multi-objective optimization solver defines the radar deployment parameters as optimization variables. The deployment parameters include at least the three-dimensional installation position of the radar unit in the vehicle coordinate system, the installation attitude angles in three axes, and the installation distance between the radar array and the vehicle surface.

[0022] The objective function constructed by the multi-objective optimization solver aims to simultaneously maximize the overall detection performance while minimizing the cost of modifying the vehicle structure and the risk of electromagnetic interference.

[0023] Specifically, the first sub-objective function is positively correlated with the main lobe gain and negatively correlated with the highest sidelobe level and angle error; The second sub-objective function is negatively correlated with the distance of the radar installation location from the original design reference point of the vehicle and the required structural modification complexity metric. The third sub-objective function is negatively correlated with the predicted electromagnetic compatibility interference level. The weights of each sub-objective function are not fixed, but are adaptively allocated by the weight factor dynamic adjustment unit according to the priority of the current driving scenario.

[0024] For example, when the system determines that the vehicle is in a highway cruising scenario, the weight factor for improving forward long-range detection performance will automatically increase; when in a congested urban scenario, the weight factor for improving lateral short-range multi-target resolution will take the lead.

[0025] The multi-objective optimization solver also needs to meet a series of hard constraints during the optimization process, including geometric constraints that the radar unit must not physically interfere with the internal structure of the vehicle body, coverage constraints that the main lobe of the radar beam must completely cover the minimum detection airspace required by regulations, and performance constraints that all performance evaluation indicators must be better than the preset safety threshold.

[0026] The multi-objective dynamic optimization decision module uses a genetic algorithm based on non-dominated sorting to iteratively solve the above optimization problem, and finally outputs a set of radar deployment parameters that are optimal under the current environment and scenario.

[0027] The deployment parameter execution and verification module is used to receive the optimal deployment parameter adjustment instructions generated by the multi-target dynamic optimization decision module, drive the high-precision automated actuator to complete the physical adjustment of the radar unit, and then start a round of verification tests to confirm the performance improvement.

[0028] The deployment parameter execution and verification module is connected to a high-rigidity multi-axis linkage radar mounting bracket adjustment mechanism, which can make precise adjustments to the position and attitude of the radar unit at the micrometer and milliradian level according to the instructions.

[0029] After the adjustment instructions are executed, the module automatically triggers the environmental awareness and data acquisition module to start a streamlined verification scan and test process.

[0030] The verification process will collect key samples of the adjusted near-field radiation pattern and use some radar target simulators for rapid performance retesting.

[0031] The obtained verification data is sent back to the deployment performance real-time evaluation module for rapid evaluation. If the evaluation results confirm that the performance indicators have reached or exceeded the optimization expectations, the process is terminated, the system records the final deployment parameters and generates a calibration report. If the expected results are not achieved, the system will use the verification data as new input and feed it back to the multi-objective dynamic optimization decision module to start a new round of optimization and fine-tuning until a convergent and satisfactory final deployment plan is formed.

[0032] As one embodiment of the present invention, the near-field scanning probe array in the environmental perception and data acquisition module has its scanning trajectory planned by an adaptive path planning algorithm based on the surface curvature of the vehicle body digital twin model and the expected radar beam direction.

[0033] The algorithm first calculates the basic scanning point cloud based on the geometric features of the vehicle body surface in the candidate radar installation area to ensure the uniformity of the spatial distribution of sampling points and the accuracy of normal alignment.

[0034] Furthermore, the algorithm will focus on the azimuth and elevation angles of the radar's expected main beam scan, as well as areas prone to strong scattering such as vehicle body structure edges and seams, and automatically increase the density of scan points to ensure the accuracy of capturing the electromagnetic characteristics of these key areas.

[0035] As one embodiment of the present invention, the radar performance prediction engine in the real-time performance evaluation module is specifically integrated with the surface contamination electromagnetic effect correction sub-model when synthesizing the actual far-field radiation pattern.

[0036] The surface contamination electromagnetic effect correction sub-model uses the contamination type and thickness distribution information identified by the optical surface contamination detection sensor to call the preset dielectric constant database to calculate the transmission loss, phase shift and surface wave excitation effect of rain film, thin ice or mud layer on millimeter wave signals. These effects are then superimposed on the radiation path from the radar array to free space in the form of a transfer function, so that the synthesized radiation pattern can truly reflect the performance degradation caused by surface contamination.

[0037] As one embodiment of the present invention, the adaptive weight factor dynamic adjustment unit in the multi-objective dynamic optimization decision module determines the driving scenario priority based on vehicle bus signals and a predefined scenario rule base.

[0038] The adaptive weight factor dynamic adjustment unit monitors the vehicle's speed signal, turn signal, navigation path information, and preliminary fusion results from other perception sensors in real time. By matching these with the scene rule base, it determines which one or a combination of typical scenarios such as highways, urban roads, and parking the vehicle is currently in. Based on this, it selects the corresponding weight factor combination from the preset weight configuration matrix and loads it into the multi-objective optimization solver.

[0039] As one embodiment of the present invention, the entire system operates within a hierarchical iterative optimization framework.

[0040] The first layer is the coarse optimization layer, which operates in a static environment after the vehicle is fully assembled and rolls off the production line. It quickly determines the macroscopic feasible domain of radar deployment parameters with a large optimization step size.

[0041] The second layer is the fine-tuning layer, which runs during dynamic road tests of the vehicle. It incorporates real driving posture, vibration, and environmental disturbance data to fine-tune the deployment parameters to optimize robustness under dynamic conditions.

[0042] The third layer is the online calibration layer, which serves as a vehicle after-sales maintenance function. When radar performance is detected to have drifted due to long-term use or accident repair, this system can be invoked to perform local recalibration and optimization.

[0043] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention completely changes the traditional static, offline radar deployment paradigm. By introducing an environmental perception and data acquisition module, it obtains the near-field radiation characteristics, vehicle scattering effect and interference source information of the radar installation location in a real vehicle and environment. This allows the performance evaluation to be based on measured physical data rather than ideal simulation, ensuring the accuracy and reliability of the evaluation model and providing real input conditions for subsequent optimization.

[0044] 2. By deploying a real-time performance evaluation module, this invention constructs an end-to-end simulation evaluation link from raw electromagnetic data to the final radar system-level performance indicators. It can quantitatively analyze the specific impact of multiple dynamic factors such as vehicle structure, attitude change, and surface contamination on key performance aspects such as radar detection accuracy, resolution, and false alarm rate, and achieve accurate prediction and diagnosis of radar performance under the coupled effects of complex factors.

[0045] 3. The multi-objective dynamic optimization decision module designed in this invention formalizes the deployment problem into a multi-objective optimization problem subject to multiple constraints, and innovatively introduces an adaptive weight factor mechanism, so that the optimization objective can be dynamically adjusted according to the driving scenario, thereby ensuring that the deployment scheme generated under any working condition is the global optimal solution after comprehensively balancing detection performance, engineering cost and electromagnetic compatibility, rather than the local optimal solution under a fixed scenario.

[0046] 4. This invention constructs a complete closed-loop system of "perception-evaluation-optimization-execution-verification." By deploying parameter execution and verification modules, it achieves the physical implementation and effect confirmation of optimization results, and supports iterative fine-tuning, ensuring the consistency and effectiveness of the optimization scheme from the digital domain to the physical domain. The entire system possesses full lifecycle deployment, optimization, and calibration capabilities from production debugging to use and maintenance, improving the environmental adaptability and performance stability of vehicle-mounted phased array radar after actual vehicle installation, and providing a solid perception foundation for the safe and reliable operation of advanced intelligent driving systems. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the overall technical architecture of the vehicle-mounted phased array radar dynamic adaptive deployment system proposed in this invention; Figure 2This is a schematic diagram of the core principle framework of the multi-objective dynamic optimization decision module in this invention; Figure 3 This is a logical flowchart of the environmental perception and data acquisition module in this invention; Figure 4 This is a logical flow diagram of the deployment of the real-time performance evaluation module in this invention; Figure 5 This is a logical flow diagram of the deployment parameter execution and verification module in this invention. Detailed Implementation

[0048] This embodiment details the specific technical implementation of the vehicle-mounted phased array radar dynamic adaptive deployment system.

[0049] Please refer to the attached document. Figures 1 to 5 This system is a closed-loop feedback and execution system that operates during the final debugging stage after vehicle production and during subsequent use and maintenance. Its core is to achieve adaptive adjustment of radar deployment parameters by sensing the real physical environment in real time, evaluating the current deployment performance, dynamically optimizing decisions and accurately executing verification.

[0050] The system consists of an environmental perception and data acquisition module, a real-time deployment performance evaluation module, a multi-objective dynamic optimization decision-making module, and a deployment parameter execution and verification module. Each module is connected to the central control unit through a high-speed data bus, forming a collaborative and organic whole.

[0051] The environmental perception and data acquisition module is responsible for collecting multi-dimensional, high-precision benchmark data necessary for building a dynamic deployment optimization model under the actual physical state of the target vehicle. Please refer to the appendix. Figure 3 This module integrates three core data acquisition subsystems: a movable, high-precision near-field scanning probe array, a high dynamic range vehicle-mounted radar target simulator array, and a vehicle status and environment sensor group.

[0052] Near-field scanning probe arrays are key equipment for obtaining the actual electromagnetic radiation characteristics of radar installation locations.

[0053] The array consists of 128 broadband probes operating in the 76 to 81 gigahertz millimeter-wave band, each probe having an independent RF transceiver channel and analog-to-digital converter.

[0054] The probe array is driven by an industrial-grade high-precision robotic arm with six degrees of freedom. The robotic arm has a repeatability accuracy of better than 5 micrometers and an angular resolution of better than 0.001 degrees.

[0055] The scanning trajectory is planned by an adaptive path planning algorithm based on the surface curvature of the vehicle body digital twin model and the expected radar beam direction.

[0056] The algorithm first imports a three-dimensional computer-aided design model of the vehicle and identifies the vehicle surface area where candidate radar units are planned to be installed. This area is usually located behind the front bumper grille, inside the rearview mirror housing, or at the four corners of the vehicle.

[0057] The algorithm calculates a set of basic scanning point clouds based on the Gaussian curvature distribution of the surface in the region, ensuring the uniformity of the distribution of scanning points in three-dimensional space, and adjusts the posture of the end of the robotic arm at each scanning point so that the probe axis is always perpendicular to the local tangent plane of the vehicle surface, in order to ensure the normal alignment accuracy.

[0058] Furthermore, based on the radar's intended functions, the algorithm will automatically increase the scanning point density to twice the base density within the key azimuth and elevation angle ranges that its main beam needs to scan.

[0059] Meanwhile, for areas known to generate electromagnetic wave diffraction and strong scattering, such as the edges of the vehicle structure, sheet metal seams, screw holes, and the junctions of different materials, the algorithm will additionally mark and deploy a high-density scanning dot matrix. The scanning dot density can be up to 4 times the base density to ensure accurate capture of the characteristics of these electromagnetically abrupt change areas.

[0060] During the scanning process, each probe transmits a calibrated continuous wave or frequency-modulated continuous wave signal at a preset frequency point and simultaneously receives signals reflected or radiated back from the radar unit array installed at the candidate location.

[0061] The acquired raw data is a complex scattering parameter matrix containing amplitude and phase information, with a data sampling rate of up to 2 gigabits per second.

[0062] All raw data is appended with a global timestamp with nanosecond precision issued by the central control unit during acquisition.

[0063] The vehicle-mounted radar target simulator array is used to generate standard point target echoes in a real vehicle environment to evaluate the radar's end-to-end detection performance.

[0064] The array consists of 24 distributed radio frequency signal sources and corresponding horn antennas, which are pre-deployed in key azimuth and elevation directions within a hemispherical space with a radius of 50 meters centered on the vehicle.

[0065] The selection of deployment locations is based on the minimum detection airspace required by regulations and high-risk areas in typical driving scenarios, such as the distance directly in front of the vehicle, blind spots to the sides, and rear-end collision areas.

[0066] Each signal source can be independently programmed to simulate standard point targets at different distances from 1 meter to 200 meters, at different radial velocities from negative 50 meters per second to positive 50 meters per second, and at different radar cross-sections from 0.1 square meters to 10 square meters.

[0067] The radio frequency signal generated by the simulator is amplified and modulated before being radiated toward the vehicle via a horn antenna.

[0068] These signals pass through the real vehicle environment and are received by the candidate radar units, thereby injecting target components with known characteristics into the radar echo signal.

[0069] The simulator array's triggering and parameter settings are strictly synchronized with the near-field scanning process to ensure the consistency of performance evaluation data over time.

[0070] The vehicle status and environment sensor group is used to collect the vehicle's current physical status and surrounding environmental parameters in real time.

[0071] The sensor suite includes a high-precision inertial measurement unit installed at the vehicle's center of gravity, which can measure the vehicle's pitch angle, roll angle, and angular velocity in three-dimensional space in real time, with an angle measurement accuracy better than 0.01 degrees.

[0072] High-precision wheel load sensors are installed at each of the four wheels to monitor changes in the sprung mass of each wheel in real time, and the data is transmitted via the controller's local area network bus.

[0073] Four optical surface contamination detection sensors are deployed at key locations on the radome surface. These sensors use multispectral imaging technology to identify the types of contaminants adhering to the radome surface, such as water, rainwater, thin ice, snow, or mud, and estimate their average thickness and distribution uniformity, with a thickness resolution of up to 0.1 mm.

[0074] In addition, temperature and humidity sensors are installed inside the radar installation cabin and in the external near-field area to continuously monitor the ambient temperature and relative humidity. The temperature measurement range is from -40 degrees Celsius to 85 degrees Celsius, with an accuracy of ±0.5 degrees Celsius.

[0075] All sensor data is collected at a sampling rate of 1000 times per second and synchronized with near-field scanning and target simulation data at the microsecond level through a unified time base system. Finally, the data is packaged into a standard format data frame and transmitted to the deployment performance real-time evaluation module.

[0076] The core task of the real-time performance evaluation module is to calculate the radar's overall performance indicators under the current candidate deployment scheme based on the real-time multi-dimensional data provided by the aforementioned modules. Please refer to the appendix. Figure 4 The module incorporates a radar performance prediction engine based on actual physical parameters, which runs on a high-performance graphics processor cluster to handle the massive computational load.

[0077] The first step of the engine is to preprocess and transform the raw scattering parameter matrix data obtained from near-field scanning.

[0078] Preprocessing includes time-domain windowing filtering of the raw data to suppress noise, and system error calibration to eliminate inherent errors introduced by the probe and cable.

[0079] Subsequently, the engine performs spherical wave expansion calculations, decomposing the near-field data collected by the probe into a set of spherical wave mode coefficients.

[0080] This process involves solving a large system of linear equations, the matrix dimension of which is related to the number of scan points and the order of expansion.

[0081] After completing the spherical wave expansion, the engine extrapolates the spherical wave coefficients to the far-field region using a rigorous far-field transformation formula, thereby calculating the theoretical far-field radiation pattern radiated from the radar installation point under free-space conditions.

[0082] The radiation pattern is three-dimensional, containing gain, phase, and polarization information in two dimensions: azimuth and elevation.

[0083] However, a spatial radiation map alone is insufficient to reflect the effects of the actual vehicle environment. Therefore, the engine needs to integrate the scattering effects of the vehicle structure.

[0084] The system has a high-precision three-dimensional digital twin model of the vehicle model. This model not only contains geometric information, but more importantly, the electromagnetic parameters of the materials of each part of its surface are based on measured data, such as the conductivity of sheet metal, the complex permittivity of plastic parts, and the transmittance of glass.

[0085] The engine uses an algorithm that combines physical optics with uniform diffraction theory to calculate the scattering field generated when the radar beam hits different parts of the vehicle body.

[0086] The calculation process discretizes the vehicle surface into millions of triangular elements, calculates the induced current for each element based on its material properties and incident wave characteristics, and then integrates to obtain the total scattered field.

[0087] Finally, the calculated scattered field is vector-superimposed with the aforementioned free-space far-field radiation pattern to synthesize the actual far-field radiation pattern of the radar at the actual installation position under the current vehicle attitude.

[0088] The vehicle attitude data here comes directly from the pitch and roll angles collected in real time by the inertial measurement unit. The engine dynamically rotates the digital twin model based on these angles to ensure that the scattering calculation is consistent with the actual attitude of the vehicle.

[0089] To further enhance the realism of the directional pattern synthesis, the engine has integrated a surface contamination electromagnetic effect correction sub-model.

[0090] This sub-model receives information on the type and thickness distribution of contamination from an optical surface contamination detection sensor.

[0091] The system maintains an internal dielectric constant database, which stores curves showing the changes in the complex dielectric constant of water, ice, and soils of different compositions as a function of temperature in the millimeter-wave band.

[0092] The modified sub-model obtains the corresponding complex permittivity by interpolation from the database based on the detected type of contamination.

[0093] For contamination layers of non-uniform thickness, the model discretizes them into multi-layer media.

[0094] Subsequently, based on transmission line theory and wave equations, the transmission and reflection coefficients of millimeter-wave signals passing through the contamination layer were calculated. These coefficients are complex numbers that include amplitude loss and phase shift.

[0095] At the same time, the model will also calculate the perturbation of the radiation pattern by surface waves that may be excited on the surface of the contamination layer.

[0096] All these effects are synthesized into an equivalent transfer function from the radar array to the external free space.

[0097] When synthesizing the actual far-field radiation pattern, the engine performs a convolution operation between this transfer function and the radiation characteristics of the radar unit itself, so that the final radiation pattern can truly reflect the performance degradation caused by surface contamination, such as the decrease in main lobe gain and the increase in side lobe level.

[0098] After obtaining the accurate actual far-field radiation pattern, the engine starts the end-to-end signal processing simulation link. This link completely simulates the workflow of a phased array radar.

[0099] First, based on the radar waveform design, the transmitted signal, such as a linear frequency modulated pulse, is simulated and generated.

[0100] Using synthesized actual far-field radiation pattern data, the weighting and delay of the transmitted signal by the space beamforming network are simulated to form a transmitted beam pointing in a specific direction.

[0101] Next, the engine calls upon the standard point target echo signal injected by the onboard radar target simulator array. This signal already includes the propagation path attenuation and multipath effect in the real vehicle environment.

[0102] The simulated receiving link performs down-conversion and analog-to-digital conversion on the echo signal, and then applies the actual far-field radiation pattern to perform receiving beamforming.

[0103] Subsequently, a complete set of signal processing algorithms, including pulse compression, moving target detection, constant false alarm rate detection, ranging, velocity measurement, angle measurement, and point aggregation, are executed.

[0104] The entire simulation process was conducted in the digital domain, but all parameters were consistent with the actual radar hardware parameters.

[0105] Based on the above simulation results, a real-time performance evaluation module is deployed to calculate and output a set of quantitative performance evaluation metrics.

[0106] This set of indicators serves as the direct basis for subsequent optimization decisions and includes at least the following: Within the key detection airspace required by regulations, the decrease in the actual main lobe gain of the radar relative to the theoretical maximum gain under ideal unobstructed conditions, expressed in decibels; The highest actual sidelobe level value, in decibels, that occurs across the entire scanned spatial domain. After measuring the angles of multiple standard point targets generated by the simulator, the root mean square error of the angle measurement accuracy is calculated in degrees. In a complex simulated scenario containing multiple targets with different radar cross-sections, the statistical average detection probability and average false alarm probability of the system are obtained. And based on the electromagnetic compatibility analysis model, the predicted interference level of the radar transmission signal to known onboard electronic devices around the vehicle, such as GPS receivers, other radars, and onboard communication modules, is expressed in decibels and milliwatts.

[0107] After all indicators have been calculated, they are encapsulated into structured data packets and sent to the multi-objective dynamic optimization decision-making module.

[0108] The multi-objective dynamic optimization decision-making module is the system's intelligent decision-making center. It receives a set of performance evaluation indicators and generates optimal radar deployment parameter adjustment commands by solving a complex optimization problem with multiple constraints. Please refer to the appendix. Figure 2 The core of this module is a multi-objective optimization solver with embedded adaptive weighting factors.

[0109] The solver first defines the optimization variables, namely the radar deployment parameters. These parameters are described in the vehicle coordinate system and include at least six degrees of freedom: the three-dimensional mounting position of the radar unit reference point relative to the vehicle design origin, i.e., the translations in the X, Y, and Z directions; the three axial mounting attitude angles of the radar unit coordinate system relative to the vehicle coordinate system, i.e., yaw, pitch, and roll angles; and the mounting distance between the radar array antenna plane and the vehicle mounting surface. These variables collectively determine the final physical state of the radar on the vehicle.

[0110] The solver needs to construct a multi-objective function, aiming to simultaneously optimize multiple competing objectives. The objective function is composed of a linear weighted sum of three sub-objective functions.

[0111] The first sub-objective function aims to maximize the overall detection performance. It is positively correlated with the main lobe gain and negatively correlated with the highest sidelobe level and the root mean square error of the angle measurement accuracy.

[0112] Specifically, this sub-function can be expressed as a weighted sum of these performance metrics after normalization.

[0113] The second sub-objective function aims to minimize engineering modification costs and risks. It is negatively correlated with the Euclidean distance between the radar installation location and the original vehicle design reference point, and also negatively correlated with the structural modification complexity metric.

[0114] This complexity metric is derived from analysis using a digital twin model, taking into account factors such as whether sheet metal needs to be cut, whether reinforcement brackets need to be added, and whether the wiring harness needs to be rerouted. Each operation is assigned a complexity weight.

[0115] The third sub-objective function aims to minimize the risk of electromagnetic interference, and it is negatively correlated with the maximum electromagnetic compatibility interference level predicted by the deployment performance real-time assessment module.

[0116] The weights of each sub-objective function are not fixed, but are dynamically allocated in real time by an adaptive weight factor adjustment unit.

[0117] The decision-making logic of this unit is based on the current driving scenario of the vehicle.

[0118] It monitors the vehicle's speed signal, turn signal switch status, path curvature information from the navigation system, and preliminary fusion results from other perception sensors in real time via the controller area network bus.

[0119] The system has a pre-stored scenario rule library, which defines the characteristic signal combinations of typical scenarios such as highway cruising, urban road following, intersection crossing, and automatic parking.

[0120] For example, when the vehicle speed is consistently greater than 80 kilometers per hour and the navigation path is a straight highway, the unit determines it to be a highway cruise scenario.

[0121] In this scenario, improving forward long-range detection performance is crucial. Therefore, the weighting factor assigned to the terms related to forward main lobe gain and long-range target detection probability in the first sub-objective function will be automatically increased to 0.6, while the weighting factor related to lateral detection performance will be reduced accordingly.

[0122] When the vehicle speed is less than 30 kilometers per hour, the turn signal is frequently activated, and multiple slow-moving targets are detected in the vicinity, the unit determines it to be an urban congestion scenario. At this time, the weight factors related to the multi-target resolution capability and lateral detection range in the first sub-objective function will become dominant and may be increased to 0.7.

[0123] The output of the weight factor dynamic adjustment unit is a weight vector, which is loaded into the multi-objective optimization solver in real time to ensure that the optimization direction is highly matched with the priority of the current driving scenario.

[0124] The optimization process also needs to meet a series of inviolable hard constraints.

[0125] Geometric constraints require that the physical envelope of the radar unit must maintain a minimum safe clearance of at least 5 millimeters between it and the internal structure of the vehicle body, such as anti-collision beams, radiators, and wiring harnesses. This clearance is verified through the collision detection function of the three-dimensional digital twin model.

[0126] Coverage constraints require that the radar beam, in the adjusted actual far-field pattern, must have a main lobe beamwidth of 3 dB that completely covers the minimum detection airspace mandated by regulations, which is usually defined by the range of azimuth and elevation angles.

[0127] Performance constraints require that all performance metrics calculated by the deployment performance real-time evaluation module must be better than preset safety thresholds, such as a detection probability greater than 0.9 and a maximum sidelobe level less than -20 dB.

[0128] To solve the aforementioned complex, nonlinear, multi-objective, constrained optimization problem, this module employs a genetic algorithm based on non-dominated sorting.

[0129] During algorithm initialization, a population of 200 individuals is randomly generated, with each individual representing a set of encoded radar deployment parameters.

[0130] In each generation of evolution, the algorithm first calculates the objective function value and the degree of constraint violation for each individual.

[0131] Then, a non-dominated sorting process is performed, dividing the population into multiple frontier levels, with individuals that are not superior to any other individual in all objectives placed at the highest frontier.

[0132] At the same time, the crowding degree of each individual is calculated to measure its distribution density in the target space.

[0133] The selection operation is based on the frontier rank and crowding of individuals, prioritizing individuals with high frontier rank and high crowding to promote the population's evolution toward the Pareto optimal frontier and maintain diversity.

[0134] Crossover and mutation operations are performed in the parameter encoding space, simulating biological evolution. The algorithm is set to a maximum of 500 generations.

[0135] Once the termination condition is met, the algorithm selects the final solution from the highest frontier. This solution can be selected based on scene weights, such as selecting the solution with the best forward detection performance in a highway scene.

[0136] Finally, the solver outputs a set of radar deployment parameter adjustment instructions that are optimal for the current environment and scenario, with precision in millimeters and milliradians.

[0137] The deployment parameter execution and verification module is responsible for translating optimization instructions from the digital world into precise actions in the physical world and verifying the adjustment effects.

[0138] Please refer to the attached document. Figure 5 This module connects to and controls the adjustment mechanism of the high-rigidity, multi-axis linkage radar mounting bracket.

[0139] This mechanism typically consists of a high-precision linear slide and a rotary table driven by a servo motor, integrated into the radar's mounting base.

[0140] The mechanism receives adjustment instructions from the multi-objective dynamic optimization decision module, which contain the target displacement and angle values ​​for six degrees of freedom.

[0141] The motion controller inside the module decomposes the commands into motion trajectories for each axis and adopts a closed-loop control strategy. It uses a grating ruler or encoder to provide real-time feedback of position information, ensuring that the final positioning accuracy reaches the micrometer and milliradian level.

[0142] After the adjustment instructions are executed, the system does not immediately end the process, but automatically triggers a round of verification testing. The deployment parameter execution and verification module sends instructions to the environment awareness and data acquisition module to start a streamlined verification scan and testing process.

[0143] This process differs from the initial full scan. It only targets the key areas of the optimized radar position, executes an optimized fast scan path, and collects a small amount of representative near-field pattern sample data.

[0144] Meanwhile, the vehicle-mounted radar target simulator array also generates standard target echoes at a few key locations. The entire verification process is shorter than the initial acquisition process.

[0145] The collected verification data is immediately sent back to the deployment performance real-time evaluation module for rapid evaluation.

[0146] The evaluation engine uses a simplified process with the same core algorithm to quickly calculate adjusted key performance indicators, such as main lobe gain, highest sidelobe level, and detection probability in the main direction.

[0147] The module will quickly compare the evaluation results with the expected optimization objectives of the multi-objective dynamic optimization decision-making module.

[0148] If the evaluation results confirm that all key performance indicators have reached or exceeded the expected optimization values, the system determines that the optimization is successful.

[0149] The system will encrypt and store the final radar deployment parameters, including all position, attitude and spacing data, along with complete environmental data snapshots, performance evaluation reports and optimization process logs, in the non-volatile memory of the vehicle's electronic control unit, and generate a standard-format calibration report that can be read by the production line management system or maintenance diagnostic system.

[0150] If the rapid evaluation results do not meet expectations, for example, if the level of a certain sidelobe is still greater than the threshold, the system will use this round of verification data as new input and feed it back to the multi-objective dynamic optimization decision module.

[0151] The optimization module combines this data with historical data and, based on the results of the previous round of optimization, initiates a new round of fine-tuning optimization with a smaller search step size and a more refined optimization range.

[0152] This iterative cycle of "execution-verification-feedback-re-optimization" can continue. Usually, after 2 to 3 iterations, the system can converge to the final deployment scheme that meets all performance and constraint requirements, thereby ensuring a high degree of consistency and reliability from optimization decision to physical implementation.

[0153] The entire system operates within a hierarchical iterative optimization framework to adapt to different stages of the vehicle's lifecycle.

[0154] The first layer is the coarse optimization layer, which operates in a static environment after the vehicle rolls off the assembly line. At this time, the vehicle is in a horizontal position, and the environment is controllable. The system executes the complete initial scan and optimization process, but the optimization algorithm has a relatively large step size to quickly determine the macroscopic feasible domain of radar deployment parameters and solve basic installation interference and performance compliance issues.

[0155] The second layer is the fine-tuning layer, which operates during dynamic road tests of the vehicle. During actual vehicle operation, the system incorporates real-time data on actual driving attitude changes collected by the inertial measurement unit, mechanical vibration data collected by vibration sensors, and dynamically changing environmental data. Driven by this dynamic data, the optimization module fine-tunes the deployment parameters, focusing on optimizing the radar's performance robustness under pitch, roll, and vibration conditions, ensuring the stability of beam pointing and consistency of detection performance under dynamic conditions.

[0156] The third layer is the online calibration layer, embedded in the vehicle diagnostic system as a vehicle after-sales maintenance function. After routine vehicle maintenance or minor collision repairs, the system can trigger a calibration process through the diagnostic interface. When the vehicle sensors detect that radar performance has drifted due to long-term use, component aging, or repair / installation errors, the system can call upon local functions within the system. In a repair shop environment, using portable near-field scanning equipment and a target simulator, it can perform local recalibration and optimization of the radar deployment parameters to restore its optimal performance.

[0157] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0158] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A vehicle-mounted phased array radar dynamic adaptive deployment system, characterized in that, include: The environmental perception and data acquisition module is used to collect multi-dimensional benchmark data necessary for building a dynamic deployment optimization model under the actual physical state of the target vehicle. The deployment performance real-time evaluation module is used to calculate the radar comprehensive performance index under the current candidate deployment scheme based on the real-time multi-dimensional data provided by the environmental perception and data acquisition module. The multi-objective dynamic optimization decision module is used to receive the set of performance indicators output by the real-time deployment performance evaluation module, compare and calculate them with preset performance thresholds and optimization targets, and generate the optimal radar deployment parameter adjustment command by solving the multi-objective optimization problem with multiple constraints. The deployment parameter execution and verification module is used to receive the optimal deployment parameter adjustment instructions generated by the multi-target dynamic optimization decision module, drive the high-precision automated actuator to complete the physical adjustment of the radar unit, and then start a round of verification tests to confirm the performance improvement.

2. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 1, characterized in that, The environmental perception and data acquisition module integrates a movable, high-precision near-field scanning probe array, a high dynamic range vehicle-mounted radar target simulator array, and a set of vehicle status and environmental sensors. The near-field scanning probe array consists of multiple broadband probes operating in the millimeter-wave band. Driven by a six-degree-of-freedom robotic arm, it performs a three-dimensional spatial scan of the vehicle surface area where candidate radar units are installed, according to a preset precision trajectory, in order to collect the raw data of the actual near-field radiation pattern at the radar installation point. The vehicle-mounted radar target simulator array consists of multiple distributed radio frequency signal sources and antennas, which are pre-deployed around the vehicle in key azimuth and elevation directions to simulate and generate standard point target echo signals with different distances, speeds and radar cross-sections in a real vehicle environment. The vehicle status and environment sensor group includes a high-precision inertial measurement unit, wheel load sensors, optical surface contamination detection sensors, and temperature and humidity sensors, which are used to collect the vehicle's current pitch angle, roll angle, sprung mass, radome surface cleanliness and type of adhering substances, and environmental temperature and humidity parameters in real time. The environmental perception and data acquisition module strictly synchronizes all collected data through timestamps and transmits it to the deployment performance real-time evaluation module.

3. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 1, characterized in that, The deployment performance real-time evaluation module has a built-in radar performance prediction engine based on actual physical parameters; The radar performance prediction engine first performs spherical wave expansion and far-field transformation calculations on the raw radiation pattern data obtained from near-field scanning, and then superimposes the vehicle structure scattering effect based on the measured material electromagnetic parameters provided by the high-precision three-dimensional digital twin model of the vehicle to synthesize the actual far-field radiation pattern of the radar at the actual installation position under the current vehicle attitude and surface state. The deployment performance real-time evaluation module further utilizes the synthesized actual far-field radiation pattern, combined with the standard point target echo signal generated by the vehicle-mounted radar target simulator array, to perform an end-to-end signal processing simulation link. The simulation link completely simulates the radar's transmitted waveform modulation, spatial beamforming, echo reception, signal detection, ranging, velocity measurement, angle measurement, and point convergence process. The deployment performance real-time evaluation module calculates and outputs a set of quantitative performance evaluation indicators based on simulation results. These indicators include at least the main lobe gain reduction value, the highest side lobe level value, the root mean square error of angle measurement accuracy, the target detection probability and false alarm probability in multi-target simulation scenarios, and the predicted electromagnetic compatibility interference level between the radar and surrounding known vehicle-mounted electronic equipment.

4. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 1, characterized in that, The core of the multi-objective dynamic optimization decision module is a multi-objective optimization solver with embedded adaptive weight factors; The multi-objective optimization solver defines the radar deployment parameters as optimization variables. The deployment parameters include at least the three-dimensional installation position of the radar unit in the vehicle coordinate system, the installation attitude angles of the three axes, and the installation distance between the radar array and the vehicle surface. The objective function constructed by the multi-objective optimization solver aims to simultaneously maximize the overall detection performance and minimize the cost of modifying the vehicle structure and the risk of electromagnetic interference. The objective function is composed of a linear weighted sum of a first sub-objective function, a second sub-objective function, and a third sub-objective function. The first sub-objective function is positively correlated with the main lobe gain and negatively correlated with the highest sidelobe level and angle error. The second sub-objective function is negatively correlated with the distance of the radar installation position from the original design reference point of the vehicle and the required structural modification complexity metric. The third sub-objective function is negatively correlated with the predicted electromagnetic compatibility interference level; The weights of each sub-objective function are adaptively allocated by the weight factor dynamic adjustment unit according to the priority of the current driving scenario; The multi-objective optimization solver must also meet a series of hard constraints during the optimization process, including geometric constraints that the radar unit must not physically interfere with the internal structure of the vehicle body, coverage constraints that the main lobe of the radar beam must completely cover the minimum detection airspace required by regulations, and performance constraints that all performance evaluation indicators must be better than the preset safety threshold. The multi-objective dynamic optimization decision-making module uses a genetic algorithm based on non-dominated sorting to iteratively solve the optimization problem, and finally outputs a set of radar deployment parameters that are optimal under the current environment and scenario.

5. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 1, characterized in that, The deployment parameter execution and verification module is connected to a high-rigidity multi-axis linkage radar mounting bracket adjustment mechanism, which can make precise adjustments to the position and attitude of the radar unit at the micrometer and milliradian levels according to instructions. After the adjustment instructions are executed, the deployment parameter execution and verification module automatically triggers the environment perception and data acquisition module to start a simplified verification scan and test process. The verification process will collect key samples of the adjusted near-field radiation pattern and use some radar target simulators for rapid performance retesting; The obtained verification data is sent back to the deployment performance real-time evaluation module for rapid evaluation; if the evaluation result confirms that the performance indicators have reached or exceeded the optimization expectations, the process terminates, the system records the final deployment parameters and generates a calibration report; If the expected results are not achieved, the system will use the verification data as new input and feed it back to the multi-objective dynamic optimization decision module to start a new round of optimization and fine-tuning until a convergent final deployment plan that meets the requirements is formed.

6. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 2, characterized in that, The scanning trajectory planning of the near-field scanning probe array is generated by an adaptive path planning algorithm based on the surface curvature of the vehicle body digital twin model and the expected radar beam direction. The adaptive path planning algorithm first calculates the basic scanning point cloud based on the vehicle body surface geometry features of the candidate radar installation area to ensure the uniformity of the spatial distribution of sampling points and the accuracy of normal alignment. The adaptive path planning algorithm further automatically increases the density of scanning points within the azimuth and elevation angles of the radar's expected main beam scan, as well as in the vehicle body structure edges and seam areas.

7. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 3, characterized in that, The radar performance prediction engine integrates a surface contamination electromagnetic effect correction sub-model when synthesizing the actual far-field radiation pattern. The surface contamination electromagnetic effect correction sub-model, based on the contamination type and thickness distribution information identified by the optical surface contamination detection sensor, calls a preset dielectric constant database to calculate the transmission loss, phase shift, and surface wave excitation effect of rain film, thin ice, or mud layers on millimeter wave signals, and superimposes these effects onto the radiation path from the radar array to free space in the form of a transfer function.

8. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 4, characterized in that, The driving scenario priority determination logic of the weight factor dynamic adjustment unit is based on vehicle bus signals and a predefined scenario rule library; The weight factor dynamic adjustment unit monitors the vehicle's speed signal, turn signal, navigation path information, and preliminary fusion results from other perception sensors in real time. By matching these with the scene rule base, it determines which state (highway, urban road, or parking scenario) the vehicle is currently in. Based on this, it selects the corresponding weight factor combination from the preset weight configuration matrix and loads it into the multi-objective optimization solver.

9. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 8, characterized in that, The optimization problem solving process constructed by the multi-objective optimization solver is as follows: Initialize a population containing multiple individuals, each representing a set of encoded radar deployment parameters; in each generation of evolution, calculate the objective function value and constraint violation degree for each individual; Then, non-dominated sorting was performed to divide the population into multiple frontier levels; Simultaneously calculate the crowding level for each individual; select operations based on the individual's frontier level and crowding level; Crossover and mutation operations are performed in the parameter encoding space; The algorithm iterates until a preset termination condition is reached, and then selects the final solution from the highest frontier. The non-dominated sorting process is as follows: individuals in the population that are not better than any other individual in all objective functions are placed at the highest frontier; after removing the individuals at the highest frontier from the population, this process is repeated to determine the next frontier level, until all individuals are assigned to a frontier.

10. The vehicle-mounted phased array radar dynamic adaptive deployment system according to claim 1, characterized in that, The system operates within a hierarchical iterative optimization framework; The hierarchical iterative optimization framework includes a first coarse optimization layer, a second fine optimization layer, and a third online calibration layer. The coarse optimization layer operates in a static environment after the vehicle is fully assembled and rolls off the production line, and quickly determines the macroscopic feasible domain of radar deployment parameters with a large optimization step size. The fine-tuning layer runs during dynamic road tests of the vehicle, introducing real driving posture, vibration and environmental interference data, and fine-tuning the deployment parameters to optimize robustness under dynamic conditions. The online calibration layer serves as a vehicle after-sales maintenance function. When radar performance is detected to have drifted due to long-term use or accident repair, this system is invoked to perform local recalibration and optimization.

Images