Automobile skylight mute intelligent detection method and system fused with acoustic imaging

Abstract

The invention discloses an intelligent detection method and system for silence of an automobile skylight fused with acoustic imaging, and relates to the technical field of acoustic measurement, and the method comprises the steps: obtaining the internal space structure information of a target vehicle, and determining an optimal installation position based on the layout point location analysis of an acoustic imager, noise data in the opening and closing process of the automobile skylight are collected in real time; building acoustic data processing dual channels, and extracting a skylight noise multi-dimensional feature set and a spatial feature set; and constructing a skylight silence detection task set, mapping the multi-dimensional feature set to the spatial feature set for fusion evaluation, generating an automobile skylight silence detection report, and realizing accurate detection and quantitative analysis of dynamic noise. The technical problem that real-time and high-precision spatialization detection and quantitative evaluation cannot be performed on dynamic noise in the opening and closing process of the automobile skylight in the prior art is solved, and the technical effect of improving the detection precision and efficiency is achieved.

Description

Technical Field

[0001] The present application relates to the field of acoustic measurement technology, and in particular to an intelligent detection method and system for dynamic and static sound of a car sunroof integrated with acoustic imaging. Background Art

[0002] With the improvement of automobile NVH (noise, vibration and harshness) performance requirements, the dynamic noise problem generated by the sunroof during opening and closing has become increasingly prominent. Traditional detection methods mainly rely on wind tunnel tests and numerical simulations coupled with CFD and acoustic software, but there are significant limitations: on the one hand, sensors such as single-point microphones are difficult to capture the transient change characteristics of noise when the sunroof is opened and closed in real time, resulting in low noise source positioning accuracy; on the other hand, existing technologies cannot synchronously obtain noise signals and spatial distribution information, and can only infer the resonance area through experimental data. The optimization process requires repeated trial and error and is inefficient. In addition, the generation of sunroof wind vibration noise is closely related to the Helmholtz resonance mechanism, and its frequency is affected by factors such as the sunroof opening area and the volume of the cavity in the car. However, the existing methods are not adaptable enough to dynamic working conditions (such as vehicle speed changes and sunroof opening adjustment), and it is difficult to quantify the noise propagation path and resonance area. The above defects make it difficult for sunroof noise detection to meet the needs of real-time, spatial analysis and efficient optimization in engineering practice.

[0003] At present, there is a technical problem in the relevant technologies that it is impossible to perform real-time, high-precision spatial detection and quantitative evaluation of the dynamic noise during the opening and closing process of the car sunroof. Summary of the invention

[0004] The present application solves the technical problem that the prior art is unable to perform real-time, high-precision spatial detection and quantitative evaluation of dynamic noise during the opening and closing process of a car sunroof by providing an intelligent detection method and system for the dynamic and static noise of a car sunroof that integrates acoustic imaging.

[0005] The present application provides an intelligent detection method for automobile sunroof movement and stillness integrating acoustic imaging, including: The internal space structure information of the target vehicle and the acoustic imager to be used are obtained, and the layout point analysis of the acoustic imager to be used is performed based on the internal space structure information to determine the layout point information of the acoustic imager; the acoustic imager to be used is installed according to the layout point information of the acoustic imager, and the installed acoustic imager to be used is started to perform real-time dynamic collection of noise data during the opening and closing process of the automobile sunroof, so as to obtain a sunroof noise signal set and a sunroof noise image set; a dual-channel acoustic data processing is established, and the sunroof noise signal set and the sunroof noise image set are synchronously analyzed and processed based on the dual-channel acoustic data processing to obtain a sunroof noise multi-dimensional feature set and a sunroof noise spatial feature set; a sunroof dynamic and static sound detection task set is constructed, and based on the sunroof dynamic and static sound detection task set, the sunroof noise multi-dimensional feature set is mapped to the sunroof noise spatial feature set for fusion evaluation, and a car sunroof dynamic and static sound detection report is generated.

[0006] The present application provides an intelligent detection system for automobile sunroof movement and stillness integrating acoustic imaging, including: a deployment point analysis module, the deployment point analysis module is used to obtain the internal space structure information of the target vehicle and the acoustic imager to be applied, and perform deployment point analysis on the acoustic imager to be applied based on the internal space structure information to determine the deployment point information of the acoustic imager; a real-time dynamic acquisition module, the real-time dynamic acquisition module is used to install the acoustic imager to be applied according to the deployment point information of the acoustic imager, and start the installed acoustic imager to be applied to perform real-time dynamic acquisition of noise data during the opening and closing process of the automobile sunroof, and obtain the sunroof noise signal set and A sunroof noise image set; a feature set acquisition module, the feature set acquisition module is used to build a dual-channel acoustic data processing, based on which the sunroof noise signal set and the sunroof noise image set are synchronously analyzed and processed to obtain a sunroof noise multidimensional feature set and a sunroof noise spatial feature set; a fusion evaluation module, the fusion evaluation module is used to construct a sunroof dynamic and static sound detection task set, based on which the sunroof dynamic and static sound detection task set is mapped to the sunroof noise multidimensional feature set to the sunroof noise spatial feature set for fusion evaluation, and a car sunroof dynamic and static sound detection report is generated.

[0007] The intelligent detection method and system for automobile sunroof movement and stillness integrated with acoustic imaging proposed in this application first obtains the internal space structure information of the target vehicle, determines the optimal installation position based on the layout point analysis of the acoustic imager, collects the noise data of the automobile sunroof in real time during the opening and closing process, and obtains the sunroof noise signal set and image set; builds a dual-channel acoustic data processing, synchronously analyzes the noise signal and image, and extracts the multi-dimensional feature set and spatial feature set of the sunroof noise; constructs a sunroof movement and stillness detection task set, maps the multi-dimensional feature set to the spatial feature set for fusion evaluation, and finally generates a car sunroof movement and stillness detection report, realizes accurate detection and quantitative analysis of dynamic noise, and realizes real-time spatial detection and quantitative evaluation of dynamic noise during the opening and closing process of the automobile sunroof through optimization of the acoustic imager layout and dual-channel data processing, achieving the technical effect of improving detection accuracy and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0009] Figure 1 A schematic diagram of the process of an intelligent detection method for automobile sunroof movement and stillness integrating acoustic imaging provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of an intelligent detection system for automobile sunroof movement and stillness integrating acoustic imaging provided in an embodiment of the present application.

[0010] Explanation of the reference numerals: layout point analysis module 10 , real-time dynamic acquisition module 20 , feature set acquisition module 30 , fusion evaluation module 40 . DETAILED DESCRIPTION

[0011] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0012] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0013] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0014] The present application embodiment provides an intelligent detection method for the movement and stillness of a car sunroof by integrating acoustic imaging, such as Figure 1 As shown, the method includes: Step S100, obtain the internal space structure information of the target vehicle and the acoustic imager to be applied, analyze the layout points of the acoustic imager to be applied based on the internal space structure information, and determine the layout point information of the acoustic imager. Specifically, to determine the layout point information of the acoustic imager in the target vehicle, it is necessary to first collect basic information, use measurement tools to measure on the spot and combine with technical documents to master the internal space structure of the vehicle, and investigate the parameters and performance of the imager. Then clarify the detection target, extract the acoustic detection influencing factors such as coverage, overlapping area, measurement accuracy and sensitivity. Then use modeling software to build the internal space model of the vehicle. After verification and calibration, perform acoustic analysis marking on it, divide the monitoring area, and plan points in the area with influencing factors and equipment attributes as constraints to generate optional layout point sets. Finally, use finite element simulation technology to simulate noise propagation, determine the noise source capture effectiveness parameters, and optimize the initial point information based on the parameters through iterative optimization algorithm until the parameters reach the optimal standard, determine the final acoustic imager layout point information, covering the precise coordinates, orientation and position relationship with surrounding components, and record the optimized parameters.

[0015] In one possible implementation, the internal space structure information of the target vehicle and the acoustic imager to be used are obtained, and the layout point analysis of the acoustic imager to be used is performed based on the internal space structure information to determine the layout point information of the acoustic imager. Step S100 further includes step S110, obtaining the dynamic and static sound detection target of the automobile sunroof, extracting acoustic influencing factors based on the dynamic and static sound detection target of the automobile sunroof, and determining the acoustic detection influencing factor set, wherein the acoustic detection influencing factor set includes coverage, overlapping area, measurement accuracy and sensitivity. Specifically, in order to achieve accurate dynamic and static sound detection of automobile sunroofs, the detection target must be clarified first. By communicating with automobile manufacturers, testing institutions, parts suppliers and scientific research teams, the detection content is determined, such as identifying types of mechanical noise, wind noise, air leakage noise, etc., quantifying noise intensity, positioning and frequency detection accuracy, and clarifying the goal of providing a basis for design optimization, quality control and after-sales service. Based on this, the acoustic influencing factors are extracted. When determining the coverage range, the skylight and surrounding potential noise source areas should be fully covered, and changes in different working conditions should be taken into consideration. When multiple devices are detected, a reasonable overlapping area ratio is set to improve the accuracy of noise source positioning. The noise intensity, source positioning and frequency measurement accuracy are determined based on the detection target and environmental factors. High-sensitivity sensors and optimized circuits are used to balance sensitivity and dynamic range to meet the needs of weak noise detection. Finally, a set of acoustic detection influencing factors that includes coverage range, overlapping area, measurement accuracy and sensitivity is constructed.

[0016] Step S120, based on the internal space structure information, generate a vehicle internal space model, and obtain the device attribute information of the acoustic imager to be applied. Specifically, in the dynamic and static detection scheme of the automobile sunroof, generating the vehicle internal space model and obtaining the device attribute information of the acoustic imager are important preliminary work. First, collect the internal space structure information, measure the length, width, height and position dimensions of each component of the vehicle on site, disassemble and analyze the structural materials such as the roof and doors, and consult the technical documents to obtain the acoustic design parameters. Then use professional software such as 3ds_Max and SolidWorks to build the model, create a framework based on the measurement information, and set the material attributes and acoustic parameters. After completing the preliminary construction, use the actual vehicle and the acoustic measurement equipment for comparative testing, and adjust the model based on the measurement data to ensure that it accurately reflects the acoustic environment in the car. At the same time, obtain the device attribute information of the acoustic imager, read the technical manual to understand the microphone array layout, frequency response range and other parameters and imaging principles, technical characteristics, consult the third-party evaluation report and actual application cases, and fully grasp its performance in different scenarios, providing strong support for subsequent detection work.

[0017] Step S130, based on the acoustic detection influencing factor set and the equipment attribute information, the points of the vehicle interior space model are preliminarily planned to obtain the initial layout point information. Specifically, after completing the construction of the vehicle interior space model and obtaining relevant information, it is necessary to plan the points based on the acoustic detection influencing factor set and the equipment attribute information. First, integrate and analyze the two, clarify the relationship between factors such as coverage range and overlapping area and the imager's effective detection distance, resolution and other attributes, and determine the coverage area and overlapping ratio according to the imager performance. Then, acoustically analyze the vehicle interior space model, define the skylight area and its surroundings, analyze internal obstacles and sound wave propagation characteristics, and divide the focus and expand the monitoring area accordingly. After that, in the monitoring area, search for points with influencing factors and equipment attributes as constraints, and generate an optional layout point set containing three-dimensional coordinates, orientation and surrounding position relationships. Finally, set evaluation indicators such as noise source coverage and positioning accuracy, comprehensively evaluate the points according to the indicator weights, sort and filter by score, and determine the initial layout point information containing detailed information, laying the foundation for subsequent acoustic detection.

[0018] Step S140, the initial layout point information is simulated and optimized, and the layout point information of the acoustic imager is determined. Specifically, after obtaining the initial layout point information, in order to achieve the best detection effect, it is necessary to simulate and optimize its layout to determine the final point. First, select professional simulation software such as COMSOL_Multiphysics, import the vehicle interior space model and the initial layout point information, and set the simulation parameters covering the frequency range, noise source characteristics, etc. Then run the software to simulate noise propagation and analyze the detection effect of the imager at each initial point, such as positioning accuracy and image quality. Determine the optimization direction based on the detection effect, select a suitable algorithm such as a genetic algorithm, and set key parameters. Then use the optimization algorithm to adjust the point, and after multiple iterations and simulation verification after each iteration, the algorithm converges to the point combination with the best detection effect. Finally, comprehensively evaluate the optimization scheme of multiple factors, select the best scheme, and record and output the final acoustic imager layout point information containing three-dimensional coordinates, orientation, and relative position relationship with key components in detail, so as to provide guarantee for the precise detection of dynamic and static sound of the automobile sunroof.

[0019] In a possible implementation, the vehicle interior space model is preliminarily planned based on the acoustic detection influencing factor set and the device attribute information to obtain the initial layout point information. Step S130 further includes step S131, acoustic analysis marking of the vehicle interior space model, determining the skylight area information, the internal obstacle set and the sound wave propagation and reflection characteristics. Specifically, the acoustic analysis marking of the vehicle interior space model is an important basis for the acoustic detection scheme. First, in the high-precision model, professional software is used to accurately define the geometric shape and size of the skylight, such as the length and width of the rectangular skylight, the radius and arc length of the arc skylight, etc., and its position in the car is determined according to the three-dimensional coordinate system, and the movement trajectory of the skylight opening and closing and the position information of different states are clarified. Then, the model is scanned in all directions to identify internal obstacles such as seats and center consoles, and the layer management tool is used to classify and mark them, and the key features of each obstacle such as length, width, height, tilt angle, etc. are recorded in detail to build an information library. Finally, consult the acoustic properties data to obtain the acoustic parameters of different materials, such as metal reflection coefficient, soft material sound absorption coefficient, glass transmittance and reflectivity, etc. Use professional acoustic simulation software to set up simulation scenes, observe the propagation path, reflection angle, attenuation degree of sound waves in the car, and scattering around obstacles, fully understand the propagation and reflection characteristics of sound waves, and provide rich and accurate data support for subsequent work.

[0020] Step S132, based on the sunroof area information, the internal obstacle set and the sound wave propagation and reflection characteristics, the monitoring area is divided to obtain the acoustic monitoring area information. Specifically, after clarifying the relevant information, dividing the monitoring area and obtaining the acoustic monitoring area information is extremely critical for the dynamic and static detection of the automobile sunroof. First, the key monitoring area is determined with the sunroof area as the core, covering the 5-10 cm extension on both sides of the sunroof guide rail and the entire sealing strip area, while considering the 20-30 cm roof area around the sunroof when it is opened, because it is easily affected by the airflow and produces aerodynamic noise. Next, the extended monitoring area is determined in combination with the internal obstacles and the sound wave characteristics, and the area is set around obstacles such as the extension of 15-20 cm around the seats, 10-15 cm around the center console, and 20-30 cm behind, because the sound waves are disturbed here; the rear seats are also included in the area of ​​20 cm in front and behind, 15 cm on the left and right, and 30-50 cm from the edge of the sunroof on the roof, because it is affected by the propagation and diffusion of sound waves. Finally, the acoustic monitoring area information is integrated, and different colors or symbols are used in the model to mark key and extended areas, accurately define the boundary range, and record the main monitoring targets of each area and the expected noise type and intensity range, providing a solid foundation and clear guidance for subsequent detection work.

[0021] Step S133, using the acoustic detection influencing factor set and the equipment attribute information as constraint information, preliminary planning of points is performed within the acoustic monitoring area information to obtain a set of optional deployment points. Specifically, after mastering the acoustic detection influencing factor set, equipment attribute information and acoustic monitoring area information, preliminary planning of points is required to obtain a set of optional deployment points. First, the constraint information is clarified. In the acoustic detection influencing factor set, the coverage range must fully cover the potential noise sources in the monitoring area. The overlapping area is set at 20%-30% to improve the noise source positioning accuracy. The measurement accuracy requires a certain accuracy of noise intensity, source positioning and frequency measurement within a specific frequency, and the sensitivity must be able to capture weak noise. In terms of equipment attributes, microphone array layout, frequency response range, resolution, detection sensitivity, effective detection distance and angle range, etc., all affect the performance of the imager. Then search for points in the acoustic monitoring area and select them according to the effective detection range of the imager to ensure that the monitoring area and boundaries are covered; at the junction of adjacent monitoring areas, adjust the points so that the imager monitoring range overlaps by 20%-30%; in areas with high accuracy and sensitivity requirements, select points that can give full play to the advantages of the imager. Finally, record the three-dimensional coordinates, orientation, and relative position relationship with surrounding components for the generated optional layout point set, laying a solid foundation for subsequent work.

[0022] Step S134, set the threshold of the number of imagers, and optimize the layout effect of the optional layout point set based on the threshold of the number of imagers to obtain the initial layout point information. Specifically, after obtaining the optional layout point set, it is necessary to set the threshold of the number of imagers and optimize the layout effect accordingly to determine the initial layout point information. When setting the threshold, comprehensively consider the project's requirements for detection accuracy and comprehensiveness, as well as the cost constraints of purchasing, maintaining imagers and data processing, and refer to the detection effects of different numbers of imagers in the previous simulation experiments and the experience data of similar projects in the past. After determining the threshold, establish an evaluation index system including noise source coverage effect, positioning accuracy, and matching degree with equipment performance, and assign weights to each index according to the key needs of the project. If the focus is on positioning, the weight of the positioning accuracy index is increased. Then, calculate the comprehensive score of each point combination according to the formula, sort the point combinations by score on the premise of meeting the threshold of the number of imagers, select the ones with high scores for in-depth analysis and comparison, and determine the best combination. Finally, the initial layout point information is recorded in detail, including the imager's three-dimensional coordinates, orientation, and relative position relationship with key components, to provide a reliable basis for subsequent work.

[0023] In a possible implementation, the initial layout point information is simulated and optimized, and the acoustic imager layout point information is determined. Step S140 further includes step S141, using finite element simulation technology to import and run simulation of the vehicle interior space model and the initial layout point information to establish a vehicle acoustic simulation model. Specifically, to establish a vehicle acoustic simulation model using finite element simulation technology, it is necessary to first evaluate and screen from a variety of software, such as COMSOL Multiphysics and LMS_Virtual.Lab, and select according to project requirements and team capabilities. Then, the vehicle interior space model is imported after format conversion and preprocessing, and mechanical and acoustic parameters are set for each component according to the actual material. At the same time, the coordinates of the initial layout point information and the acoustic imager equipment parameters are accurately entered to determine the interactive relationship between the imager and the model. Subsequently, the model is meshed, fine meshes are set in key areas, appropriate unit types are selected, and acoustic analysis types, solvers and other parameters are set before running simulation. After completion, the simulation results are compared and verified with theoretical or actual test data. If there is any deviation, the model is adjusted. After multiple adjustments, it is ensured that the model can accurately reflect the acoustic environment inside the vehicle.

[0024] Step S142, using the vehicle acoustic simulation model to simulate the opening and closing of the car sunroof and monitor the noise propagation, and obtain noise propagation simulation information. Specifically, the following steps are required to use the vehicle acoustic simulation model to carry out the simulation of the opening and closing of the car sunroof and the monitoring of noise propagation to obtain the noise propagation simulation information. First, check and optimize the model to ensure that the geometric shape, size, position and material acoustic parameters of each component are accurate. Then, for the simulation of the opening and closing of the sunroof, accurately set the motion parameters of the translation or rotation sunroof, such as speed, stroke, angle, etc., and set the position, intensity and type of the noise source. After completing the parameter setting, start the simulation, reasonably set the time step and total duration, and the software calculates the phenomenon of sound wave propagation according to the acoustic principle during the simulation. At the same time, reasonably arrange the monitoring points in the car, covering the periphery of the sunroof, the area that is easily perceived by personnel and the key parts, and collect and record the sound pressure level, frequency component, phase and other data in real time. Finally, use data processing software to organize the data and convert it into a visual chart, based on the in-depth analysis of the propagation attenuation, reflection scattering and propagation differences between different materials and different working conditions, to provide a basis for vehicle acoustic optimization.

[0025] Step S143, based on the noise propagation simulation information, the effectiveness of the initial deployment point information is evaluated, and the noise source capture effectiveness parameters are determined. Specifically, after obtaining the noise propagation simulation information, the effectiveness of the initial deployment point information is evaluated and the noise source capture effectiveness parameters are determined, which is of great significance to the optimization of the acoustic imager deployment. First, an evaluation index system is established. By comparing the actual position of the simulated noise source with the estimated position of the imager, the positioning error is calculated using the Euclidean distance formula, and the positioning accuracy is measured by the statistical mean value and standard deviation; the error between the measured and simulated noise intensity of the monitoring point is calculated using the root mean square error formula to evaluate the accuracy of signal strength detection; the edge clarity and contrast of the noise source area are calculated with the help of the image recognition algorithm, and the acoustic imaging clarity index is constructed. Then, relevant data is extracted from the simulation information and sorted, and the effectiveness parameters are calculated according to the index system. Finally, the parameters are analyzed in depth, the performance of different points is compared, and the gap with the ideal value is found. If the index of a certain point is not good, the reasons such as the distance from the noise source or the complex surrounding acoustic environment are analyzed, and improvement measures such as adjusting the position or optimizing the parameters are proposed to improve the effectiveness of the initial deployment points and lay a solid foundation for determining the precise deployment points.

[0026] Step S144, the initial layout point information is iteratively simulated and optimized based on the noise source capture effectiveness parameter by the vehicle acoustic simulation model, and the layout point information of the acoustic imager is determined. Specifically, in order to determine the accurate layout point of the acoustic imager, the initial layout point information is iteratively optimized based on the noise source capture effectiveness parameter with the help of the vehicle acoustic simulation model. First, it is clear that the goal is to improve the noise source capture effectiveness, such as reducing the positioning error, the root mean square error, and improving the imaging clarity, while sorting out the constraints such as vehicle space restrictions and the upper limit of the number of imagers. After analysis, due to the complexity of the model and the need for multi-parameter optimization, the genetic algorithm is selected as the main method, and the performance is improved by combining the local search algorithm. Then, according to the initial layout information, the genetic algorithm is used to generate an initial population containing 50 point combinations to ensure that the constraints are met, and each combination is imported into the model, and the noise propagation information is simulated. The fitness value is calculated according to the effectiveness parameter, and then the population is subjected to genetic operations such as selection, crossover, and mutation to generate a new population, and the operation is repeated until the termination conditions such as the maximum number of iterations or the fitness value change threshold are met, and the optimal point combination is determined. Finally, actual vehicle tests or high-precision physical model simulation verification are used. If the requirements are met, the final layout points are determined. Otherwise, the reasons are analyzed, the model is improved, the algorithm is adjusted, and then re-optimized until the requirements are met.

[0027] Step S200, according to the acoustic imager layout point information, the acoustic imager to be used is installed, and the installed acoustic imager to be used is started to collect the noise data in the process of opening and closing the car sunroof in real time and dynamically, and obtain the sunroof noise signal set and the sunroof noise image set. Specifically, when using the acoustic imager to collect the opening and closing noise data of the car sunroof, it is necessary to make preparations before installation, count the acoustic imager and accessories, tools, park the vehicle in a quiet place, turn off irrelevant equipment and clean the interior. Then, according to the layout point information, use a tape measure to accurately locate, and firmly install the imager through a bracket, screws and shock-absorbing pads. After installation, connect the device to the data acquisition terminal, turn on the power, and set the sampling frequency, image resolution, sensitivity and other parameters and storage path in the supporting software according to the detection requirements. Then, test the operation status of the sunroof, confirm that the imager is working properly, operate the sunroof according to the predetermined plan, and the imager collects noise signals and images in real time to generate a noise signal set and an image set. Finally, the data is preliminarily checked, sorted and organized by test conditions, and stored in reliable equipment to provide accurate data support for subsequent analysis.

[0028] Step S300, build a dual channel for acoustic data processing, and analyze and process the skylight noise signal set and the skylight noise image set based on the dual channel for acoustic data processing to obtain a skylight noise multidimensional feature set and a skylight noise spatial feature set. Specifically, to build a dual channel for acoustic data processing, you need to select an adaptation network first. The acoustic signal processing network selects RNN or LSTM to process the skylight noise signal, and adjusts the input layer receiving signal format during adaptation; the acoustic image processing network selects CNN architectures such as AlexNet, VGG or ResNet to adapt the input size and number of channels. Use TensorFlow or PyTorch to build an integrated framework, define parallel sub-models and set the data flow. When processing the skylight noise signal set, it is first denoised, normalized and pre-processed by feature engineering, and then input into the network. The LSTM and other layers extract time series features, and the fully connected layer integrates and outputs a multidimensional feature set. To process the sunroof noise image set, we first perform image enhancement, filtering, normalization, and possible segmentation preprocessing, and then input it into the network. The convolution layer and pooling layer extract features, and the fully connected layer outputs the spatial feature set. Finally, the multi-dimensional and spatial feature sets are spliced ​​and integrated, and the feature sets are input into the classification model and verified in combination with actual conditions to ensure that the feature sets are accurate and reliable, providing strong support for the study of automobile sunroof noise.

[0029] In a possible implementation, a dual channel for acoustic data processing is constructed, and the skylight noise signal set and the skylight noise image set are analyzed and processed synchronously based on the dual channel for acoustic data processing to obtain a skylight noise multidimensional feature set and a skylight noise spatial feature set, and step S300 further includes step S310, mining and collecting the skylight historical noise signal set and the skylight historical noise image set, filtering and preprocessing the skylight historical noise signal set and the skylight historical noise image set, and obtaining an available skylight noise signal set and an available skylight noise image set. Specifically, in order to obtain an available skylight noise data set, the skylight historical noise data is first collected from multiple channels such as automobile manufacturers, laboratory simulations, and road tests, and is sorted and summarized according to source, time, working conditions, etc., and detailed annotations are added. Then, abnormal signals and images are screened and data features are annotated. Subsequently, for the noise signal set, a low-pass, high-pass or band-pass filtering algorithm is selected according to the noise characteristics to remove interference; for the noise image set, a Gaussian and median filtering algorithm is used to smooth the image and remove noise points. After preprocessing, the effect is verified by comparing the spectrum, time domain waveform, and using indicators such as PSNR and SSIM. If it is not ideal, the algorithm is adjusted and reprocessed. Finally, the data is sorted by vehicle model, sunroof type, etc., metadata is added, and stored in a reliable device to provide high-quality data support for subsequent analysis.

[0030] Step S320, respectively, performs noise feature annotation training on the available skylight noise signal set and the available skylight noise image set to generate an acoustic signal processing network and an acoustic image processing network. Specifically, in order to generate the acoustic signal processing and acoustic image processing networks, noise feature annotation training should be carried out on the available skylight noise signal set and the image set respectively. In the early stage, the data is sorted and cleaned, the abnormal values ​​of the signal set and the images of poor quality in the image set are removed, and then the annotation rules are determined, such as the frequency, amplitude, and time domain characteristics of the signal, the position, shape, and visual intensity of the noise source of the image. When training the acoustic signal processing network, professionals use signal analysis software to annotate the signal set according to the rules, select appropriate neural network architectures such as RNN, LSTM, and CNN, divide the annotation set into proportions and input it into the network, set parameters such as the learning rate, train with the back propagation algorithm, use the validation set to prevent overfitting, and evaluate the network performance with the test set. In terms of acoustic image processing network training, professionals use annotation tools to annotate image sets according to rules, select CNN architectures such as AlexNet, VGG, and ResNet, divide the annotation sets and input them for training, set parameters such as the convolution kernel size, and train using the cross-entropy loss function and back-propagation algorithm. The network is monitored using the validation set and evaluated using the test set. Ultimately, two networks are successfully generated, laying a solid foundation for subsequent skylight noise analysis and processing.

[0031] Step S330, the acoustic signal processing network and the acoustic image processing network are integrated in parallel to build the acoustic data processing dual channel. Specifically, to build the acoustic data processing dual channel, it is necessary to first sort out the characteristics of the acoustic signal processing and image processing networks, clarify the noise types, frequency responses, image feature recognition capabilities and input requirements that each is good at processing, and evaluate the performance of both, and calculate the accuracy, recall rate of the acoustic signal processing network and the mAP value of the acoustic image processing network. Next, a unified input interface is designed to convert the acoustic signal into a similar image format, adjust the image size and number of channels, build a compatible output interface, and use a structure to integrate different output information. Subsequently, TensorFlow or PyTorch is selected to build an integrated framework, define parallel sub-models and control data flow, and reasonably allocate computing resources. Finally, by inputting a variety of acoustic data debugging functions, troubleshooting the reasons that do not meet expectations, and optimizing from the aspects of computational efficiency (such as model pruning, quantization) and accuracy (such as adjusting training parameters and increasing the amount of data), a dual channel is successfully built to lay the foundation for efficient processing of acoustic data.

[0032] In a possible implementation, the available skylight noise signal set and the available skylight noise image set are respectively trained for noise feature annotation to generate an acoustic signal processing network and an acoustic image processing network. Step S320 further includes step S321, respectively annotating the available skylight noise signal set and the available skylight noise image set for noise feature annotation to obtain a skylight noise signal sample set and a skylight noise image sample set. Specifically, in order to obtain a skylight noise signal sample set and an image sample set, it is necessary to carry out noise feature annotation work on the available skylight noise signal set and the image set, respectively. For the noise signal set, an acoustic professional annotation team is formed, equipped with signal analysis software such as MATLAB and LabVIEW and high-performance computers. According to the rules such as frequency range, amplitude, and time domain characteristics, the annotation personnel operate the software to measure and record the signal characteristics, and organize them into sample sets after review and reexamination. For the noise image set, we organize labelers who are familiar with image processing, select labeling tools such as LabelImg and configure them reasonably. According to the rules such as the location, shape, and visual intensity of the noise source, the labelers select, describe and quantify the noise information in the tool. After quality inspection and correction, an image sample set is formed, which lays a solid foundation for subsequent deep learning model training and acoustic analysis.

[0033] Step S322, use a deep neural network to perform feature recognition training on the skylight noise signal sample set to generate an initial signal processing network. Specifically, to generate the initial signal processing network, it is necessary to first select a suitable deep neural network architecture such as RNN and its variants or MLP based on the time series and other characteristics of the skylight noise signal, refer to successful cases in the field of acoustic signal processing. Next, the skylight noise signal sample set is normalized, and the data is enhanced by adding noise, adjusting the amplitude and time domain scale, and then divided into training set, validation set and test set at a ratio of 70%-80%, 10%-15%, and 10%-15%. After that, the number of input layer nodes is determined according to the signal feature dimension, and the hidden layer structure is set by experimentally trying different numbers of layers and nodes, and appropriate activation functions such as ReLU are selected, and the number of output layer nodes and activation functions are determined according to the task type. During training, select optimization algorithms such as Adam, set the learning rate, define loss functions such as cross entropy or mean square error according to the task, input the training set in batches for iterative training, and monitor the performance with the validation set. After training is completed, the model is evaluated using the test set. If the performance is poor, the network architecture, data preprocessing, hyperparameters, etc. are analyzed, and the architecture and parameters are adjusted and retrained until an initial signal processing network that meets the requirements is generated.

[0034] Step S323, use a convolutional neural network to perform feature recognition training on the skylight noise image sample set to generate an initial image processing network. Specifically, generate an initial signal processing network, first conduct a secondary review of the skylight noise signal sample set, correct labeling errors, process missing data, and build a computing environment with TensorFlow or PyTorch as the framework and GPU acceleration. According to the characteristics of the noise signal and the task requirements, select the LSTM, GRU or MLP architecture, determine the number of layers and nodes through experiments, and select activation functions such as ReLU. Normalize the sample set, extract and filter the time domain and frequency domain features, and divide the training, verification, and test sets into ratios of 70%-80%, 10%-15%, and 10%-15%. During training, select optimization algorithms such as Adam, set hyperparameters such as learning rate, batch size, and number of iterations, define the cross entropy or mean square error loss function according to the task, and monitor the performance with the verification set. After the training is completed, the test set is used for evaluation. If the performance meets the requirements, it is applied. Otherwise, the reasons for overfitting or underfitting are analyzed, the network complexity and hyperparameters are adjusted, and retraining and evaluation are performed until an initial signal processing network that meets the requirements is generated.

[0035] Step S324, cross-validation tuning is performed on the initial signal processing network and the initial image processing network respectively to generate the acoustic signal processing network and the acoustic image processing network. Specifically, to generate the acoustic signal and image processing network, the initial network needs to be cross-validated and tuned. In the preparation stage, the signal sample set and the image sample set are divided into data using the K-fold (K is usually 5 or 10) cross-validation method, and the evaluation indicators are determined according to the task characteristics, such as the signal network uses accuracy, recall rate, and F1 value to evaluate the classification performance, and the image network uses mAP to evaluate the target detection performance. When tuning the initial signal processing network, 4 subsets are used for training and 1 subset for verification in a cycle, and the K groups of indicator values ​​are analyzed. If overfitting, regularization terms or Dropout are added. If underfitting, the network complexity is increased and the hyperparameters are adjusted. The final network is determined based on the comprehensive performance. For the initial image processing network, the same operation is performed. If the mAP value is not good, the convolution kernel size can be adjusted, pruning or data enhancement can be performed, and the final network is determined according to the results. Finally, the two networks are comprehensively evaluated and their collaborative performance is jointly tested. Once they meet the standards, they are applied to actual acoustic data processing tasks.

[0036] Step S400, constructing a sunroof dynamic and static sound detection task set, mapping the sunroof noise multidimensional feature set to the sunroof noise spatial feature set based on the sunroof dynamic and static sound detection task set for fusion evaluation, and generating a car sunroof dynamic and static sound detection report. Specifically, to construct a sunroof dynamic and static sound detection task set and generate a detection report, it is necessary to first clarify the detection task types such as normal opening, specific working conditions and dynamic and static sound caused by faults, collect data from the manufacturer's test database, laboratory simulation and road test, and mark and organize them into a standard format according to the task type, noise intensity, etc. Next, analyze the connection between multidimensional and spatial feature sets, build mapping relationships by regression analysis, select weighted fusion or neural network fusion methods, evaluate the feature set fusion according to the task set, and judge the abnormal dynamic and static sound according to the threshold. Finally, design a report framework containing an introduction, a task set overview, etc., fill in the evaluation data, use charts for analysis, and after expert review and improvement, generate a comprehensive and accurate car sunroof dynamic and static sound detection report that can provide valuable information to relevant personnel.

[0037] In a possible implementation, a sunroof dynamic and static sound detection task set is constructed, and based on the sunroof dynamic and static sound detection task set, the sunroof noise multidimensional feature set is mapped to the sunroof noise spatial feature set for fusion evaluation, and a car sunroof dynamic and static sound detection report is generated. Step S400 further includes step S410, and a sunroof acoustic distribution space is constructed according to the sunroof noise spatial feature set. Specifically, to construct the sunroof acoustic distribution space, first widely collect multi-source sunroof noise spatial feature set data such as automobile manufacturer tests, laboratory simulations and road tests, sort out and remove defects, and communicate with relevant parties to clarify the construction goals and requirements such as accuracy and dimension. Then, determine the three-dimensional or two-dimensional space dimension according to the characteristics and requirements, select a Cartesian coordinate system with a fixed point of the sunroof as the origin to build the system, and divide the space grid of appropriate size according to accuracy. Subsequently, formulate rules to map the characteristics such as the position and intensity of the noise source to the grid unit, fill and quantize each unit so that it has a clear acoustic characteristic value, and use visualization software to present it intuitively. Finally, choose comparison or simulation method for verification, analyze the cause of the error and correct it, and continue to optimize and update according to technological development and changes in demand, so as to construct a practical skylight acoustic distribution space.

[0038] Step S420, maps the multidimensional feature set of the skylight noise to the skylight acoustic distribution space according to the spatial distribution information for feature matching and fusion, and obtains the skylight noise fusion feature set. Specifically, to obtain the skylight noise fusion feature set, first review the multidimensional feature set and the acoustic distribution space data, pre-process the numerical features, and clarify the mapping rules and matching standards based on the physical characteristics and requirements. Then, according to the rules, locate the multidimensional features to the acoustic distribution space area according to the noise source position, calculate the matching degree according to the standard and filter the feature pairs. Subsequently, select fusion methods such as weighted average, PCA or deep learning according to the noise characteristics, operate on the filtered feature pairs, organize the fused features and store them into sets, and add information. Finally, determine the accuracy and other verification indicators to evaluate the fused feature set. If there is a problem, analyze the mapping rules and other reasons, and optimize them in a targeted manner to complete the construction of the feature set.

[0039] Step S430, based on the sunroof dynamic and static sound detection task set, the sunroof noise fusion feature set is sequentially analyzed and evaluated to generate a car sunroof dynamic and static sound detection report. Specifically, to generate a car sunroof dynamic and static sound detection report, first clarify the details of the sunroof dynamic and static sound detection task set and the corresponding evaluation criteria, and adapt and pre-process the sunroof noise fusion feature set data. Subsequently, according to the order of the task set, relevant features are extracted for each task, and statistical analysis and machine learning algorithm comparison standards are used for evaluation, and the actual value of the task, comparison situation, evaluation conclusion and other results are recorded in detail. Finally, a report framework covering the introduction, task overview and other parts is designed, and the evaluation data is filled in and displayed with charts. After being reviewed and improved by experts, errors are corrected and information is supplemented to form a high-quality detection report to provide valuable reference for relevant personnel.

[0040] In a possible implementation, the skylight noise multidimensional feature set is mapped to the skylight acoustic distribution space according to the spatial distribution information for feature matching and fusion, and a skylight noise fusion feature set is obtained. Step S420 further includes step S421, mapping the skylight noise multidimensional feature set to the skylight acoustic distribution space according to the spatial distribution information for feature matching, and obtaining a skylight noise matching feature set. Specifically, firstly, the skylight noise multidimensional feature set (including time domain, frequency domain, energy and other features) and the skylight acoustic distribution space are combed and analyzed, the data is unified and preprocessed, and the outliers and normalized numerical features are removed. Then, mapping rules are formulated, based on the location of the noise source, the corresponding spatial area is inferred according to the coordinates or signal attenuation; according to the frequency characteristics, the high-frequency noise features are mapped to a small area near the noise source, and the low-frequency noise features are mapped to a larger range area; combined with other features such as energy for comprehensive mapping, such as high energy corresponding to a larger area. Finally, the feature values ​​are extracted according to the rules, the matching degree with the spatial area is calculated using methods such as Euclidean distance, and a threshold is set for screening, and the successfully matched information is organized into a skylight noise matching feature set.

[0041] Step S422, weighted fusion is performed on each matching signal feature and spatial feature in the skylight noise matching feature set to obtain the skylight noise fusion feature set. Specifically, the role of matching signals and spatial features in describing the characteristics of skylight noise is deeply analyzed, and the weight of each feature is determined based on actual application requirements and the experience of acoustic experts. For example, the weight of amplitude features is higher when focusing on noise intensity judgment. Subsequently, each group of matching signal features (such as amplitude, frequency, energy, etc.) in the matching feature set is weighted and summed according to the weight, and the spatial features (such as noise source position coordinates, regional area, shape parameters, etc., the coordinates are normalized first) are also operated in the same way. Next, the weighted signal and spatial feature values ​​are spliced ​​into a fused feature vector, and finally all fused feature vectors are sorted into a set in order. This set combines two types of feature information and can provide strong support for subsequent noise analysis and other tasks.

[0042] In a possible implementation, based on the sunroof dynamic and static sound detection task set, the sunroof noise fusion feature set is analyzed and evaluated in sequence to generate a sunroof dynamic and static sound detection report for the automobile. Step S430 further includes step S431, based on each detection task in the sunroof dynamic and static sound detection task set, the sunroof noise fusion feature set is analyzed and evaluated in sequence to obtain a sunroof noise detection task information set. Specifically, the sunroof dynamic and static sound detection task set is comprehensively sorted out to clarify the conditions and objectives of each task, and the noise fusion feature set is familiar at the same time. Check and adapt the data format, pre-process the data, remove outliers, and normalize the features. Then, according to the order of the task set, extract relevant data from the fusion feature set for each task, compare it with the preset standard, and use statistical analysis and machine learning algorithms for comprehensive evaluation. Finally, a structured table is designed to record the evaluation information of each task, including the task name, actual value of the feature, standard range, conclusion, abnormal type and cause speculation, etc. in detail, and the table is summarized to form an information set to provide key data for subsequent analysis and report generation.

[0043] Step S432, record and integrate the sunroof noise detection task information set to generate the car sunroof dynamic and static sound detection report. Specifically, when generating the car sunroof dynamic and static sound detection report, first build a framework containing core sections such as introduction and task details based on the detection requirements and specifications, and plan the internal structure of each section. Then, extract data such as task name and actual value of features from the sunroof noise detection task information set, organize them according to the framework requirements, and write text descriptions of each section. Subsequently, invite acoustic experts and automotive engineers to review and proofread, check accuracy, completeness and logic, make visual charts such as bar charts and line charts and insert them into the report. Finally, improve the report according to the review opinions and chart conditions, adjust the content order and structure, optimize the coordination between charts and text, and generate a final report that can provide a strong reference for relevant parties.

[0044] The embodiment of the present application obtains the internal space structure information of the target vehicle, determines the optimal installation position based on the acoustic imager layout point analysis, collects the noise data of the car sunroof in real time during the opening and closing process, and obtains the sunroof noise signal set and image set; builds a dual-channel acoustic data processing, synchronously analyzes the noise signal and the image, and extracts the multi-dimensional feature set and spatial feature set of the sunroof noise; constructs a sunroof dynamic and static sound detection task set, maps the multi-dimensional feature set to the spatial feature set for fusion evaluation, and finally generates a car sunroof dynamic and static sound detection report, so as to realize the accurate detection and quantitative analysis of dynamic noise. Through the optimization of the acoustic imager layout and dual-channel data processing, the real-time spatial detection and quantitative evaluation of the dynamic noise in the process of opening and closing the car sunroof are realized, and the technical effect of improving the detection accuracy and efficiency is achieved.

[0045] In the above, refer to Figure 1The following describes in detail the intelligent detection method of the car sunroof dynamic and static sound by integrating acoustic imaging according to an embodiment of the present invention. Figure 2 The following describes an intelligent detection system for automobile sunroof movement and stillness integrating acoustic imaging according to an embodiment of the present invention.

[0046] The intelligent detection system for the dynamic and static noise of the automobile sunroof integrated with acoustic imaging according to the embodiment of the present invention is used to solve the technical problem that the existing technology cannot perform real-time, high-precision spatial detection and quantitative evaluation of the dynamic noise during the opening and closing process of the automobile sunroof. Through the optimization of the layout of the acoustic imager and dual-channel data processing, the real-time spatial detection and quantitative evaluation of the dynamic noise during the opening and closing process of the automobile sunroof is realized, achieving the technical effect of improving the detection accuracy and efficiency. The intelligent detection system for the dynamic and static noise of the automobile sunroof integrated with acoustic imaging includes: a layout point analysis module 10, a real-time dynamic acquisition module 20, a feature set acquisition module 30, and a fusion evaluation module 40.

[0047] The layout point analysis module 10 is used to obtain the internal space structure information of the target vehicle and the acoustic imager to be used, and perform layout point analysis on the acoustic imager to be used based on the internal space structure information to determine the layout point information of the acoustic imager.

[0048] The real-time dynamic acquisition module 20 is used to install the acoustic imager to be used according to the acoustic imager layout point information, and start the installed acoustic imager to be used to perform real-time dynamic acquisition of noise data during the opening and closing process of the car sunroof, so as to obtain a sunroof noise signal set and a sunroof noise image set.

[0049] The feature set acquisition module 30 is used to build a dual channel for acoustic data processing, and based on the dual channel for acoustic data processing, the skylight noise signal set and the skylight noise image set are synchronously analyzed and processed to obtain a multi-dimensional feature set of skylight noise and a spatial feature set of skylight noise.

[0050] The fusion evaluation module 40 is used to construct a sunroof dynamic and static sound detection task set, and based on the sunroof dynamic and static sound detection task set, the sunroof noise multidimensional feature set is mapped to the sunroof noise spatial feature set for fusion evaluation to generate a car sunroof dynamic and static sound detection report.

[0051] The specific configuration of the deployment point analysis module 10 will be described in detail below. As described above, the internal space structure information of the target vehicle and the acoustic imager to be used are obtained, and the deployment point analysis of the acoustic imager to be used is performed based on the internal space structure information to determine the deployment point information of the acoustic imager. The deployment point analysis module 10 further includes: a detection target acquisition unit, the detection target acquisition unit is used to obtain the dynamic and static detection target of the car sunroof, extract acoustic influencing factors based on the dynamic and static detection target of the car sunroof, and determine the acoustic detection influencing factor set, the acoustic detection influencing factor set includes coverage, overlapping area, measurement accuracy and sensitivity; equipment attribute information acquisition The device attribute information acquisition unit is used to generate a vehicle interior space model according to the interior space construction information, and acquire the device attribute information of the acoustic imager to be applied; the point preliminary planning unit is used to perform preliminary point planning on the vehicle interior space model based on the acoustic detection influencing factor set and the device attribute information to obtain initial layout point information; the acoustic imager layout point information determination unit is used to perform simulation and layout optimization on the initial layout point information to determine the acoustic imager layout point information.

[0052] Among them, based on the acoustic detection influencing factor set and the equipment attribute information, the vehicle interior space model is preliminarily planned for points to obtain initial layout point information, and the point preliminary planning unit further includes: an acoustic analysis marking subunit, the acoustic analysis marking subunit is used to perform acoustic analysis marking on the vehicle interior space model, determine the skylight area information, the internal obstacle set and the sound wave propagation and reflection characteristics; a monitoring area division subunit, the monitoring area division subunit is used to divide the monitoring area based on the skylight area information, the internal obstacle set and the sound wave propagation and reflection characteristics, and obtain the acoustic monitoring area information; a constraint information determination subunit, the constraint information determination subunit is used to use the acoustic detection influencing factor set and the equipment attribute information as constraint information, perform preliminary point planning within the acoustic monitoring area information, and obtain an optional layout point set; an imager number threshold setting subunit, the imager number threshold setting subunit is used to set the imager number threshold, and optimize the layout effect of the optional layout point set based on the imager number threshold to obtain the initial layout point information.

[0053] Among them, the initial layout point information is simulated and optimized, and the layout point information of the acoustic imager is determined. The acoustic imager layout point information determination unit further includes: an operation simulation subunit, the operation simulation subunit is used to use finite element simulation technology to import the vehicle interior space model and the initial layout point information for operation simulation, and establish a vehicle acoustic simulation model; a propagation monitoring subunit, the propagation monitoring subunit is used to use the vehicle acoustic simulation model to simulate the opening and closing of the car sunroof and monitor noise propagation to obtain noise propagation simulation information; an effectiveness evaluation subunit, the effectiveness evaluation subunit is used to evaluate the effectiveness of the initial layout point information based on the noise propagation simulation information, and determine the noise source capture effectiveness parameters; an iterative simulation optimization subunit, the iterative simulation optimization subunit is used to iteratively simulate and optimize the initial layout point information based on the noise source capture effectiveness parameters through the vehicle acoustic simulation model to determine the acoustic imager layout point information.

[0054] The specific configuration of the feature set acquisition module 30 will be described in detail below. As described above, a dual channel for acoustic data processing is constructed, and the skylight noise signal set and the skylight noise image set are analyzed and processed synchronously based on the dual channel for acoustic data processing to obtain a multi-dimensional feature set of skylight noise and a spatial feature set of skylight noise. The feature set acquisition module 30 further includes: a filtering preprocessing unit, which is used to mine and collect the skylight historical noise signal set and the skylight historical noise image set, and filter and preprocess the skylight historical noise signal set and the skylight historical noise image set to obtain an available skylight noise signal set and an available skylight noise image set; a processing network generation unit, which is used to perform noise feature annotation training on the available skylight noise signal set and the available skylight noise image set, respectively, to generate an acoustic signal processing network and an acoustic image processing network; a parallel integration unit, which is used to integrate the acoustic signal processing network and the acoustic image processing network in parallel to construct the dual channel for acoustic data processing.

[0055] Among them, the available skylight noise signal set and the available skylight noise image set are respectively trained for noise feature annotation to generate an acoustic signal processing network and an acoustic image processing network, and the processing network generation unit further includes: a sample set acquisition subunit, the sample set acquisition subunit is used to respectively perform noise feature annotation on the available skylight noise signal set and the available skylight noise image set to obtain a skylight noise signal sample set and a skylight noise image sample set; an initial signal processing network generation subunit, the initial signal processing network generation subunit is used to use a deep neural network to perform feature recognition training on the skylight noise signal sample set to generate an initial signal processing network; an initial image processing network generation subunit, the initial image processing network generation subunit is used to use a convolutional neural network to perform feature recognition training on the skylight noise image sample set to generate an initial image processing network; a cross-validation tuning subunit, the cross-validation tuning subunit is used to respectively perform cross-validation tuning on the initial signal processing network and the initial image processing network to generate the acoustic signal processing network and the acoustic image processing network.

[0056] The specific configuration of the fusion evaluation module 40 will be described in detail below. As described above, a sunroof dynamic and static sound detection task set is constructed, and based on the sunroof dynamic and static sound detection task set, the sunroof noise multidimensional feature set is mapped to the sunroof noise spatial feature set for fusion evaluation, and a car sunroof dynamic and static sound detection report is generated. The fusion evaluation module 40 further includes: a sunroof acoustic distribution space construction unit, which is used to construct a sunroof acoustic distribution space according to the sunroof noise spatial feature set; a feature matching fusion unit, which is used to map the sunroof noise multidimensional feature set to the sunroof acoustic distribution space according to the spatial distribution information for feature matching fusion to obtain a sunroof noise fusion feature set; and a silence detection report generation unit, which is used to perform noise analysis and evaluation on the sunroof noise fusion feature set in turn based on the sunroof dynamic and static sound detection task set to generate a car sunroof dynamic and static sound detection report.

[0057] Among them, the multidimensional feature set of skylight noise is mapped to the skylight acoustic distribution space according to the spatial distribution information for feature matching and fusion to obtain the skylight noise fusion feature set, and the feature matching fusion unit further includes: a distribution information mapping subunit, the distribution information mapping subunit is used to map the multidimensional feature set of skylight noise to the skylight acoustic distribution space according to the spatial distribution information for feature matching to obtain the skylight noise matching feature set; a weighted fusion subunit, the weighted fusion subunit is used to weightedly fuse each matching signal feature and spatial feature in the skylight noise matching feature set to obtain the skylight noise fusion feature set.

[0058] Among them, based on the sunroof dynamic and static sound detection task set, the sunroof noise fusion feature set is analyzed and evaluated in sequence to generate a car sunroof dynamic and static sound detection report, and the silent detection report generation unit further includes: a sunroof noise detection task information set acquisition subunit, the sunroof noise detection task information set acquisition subunit is used to perform noise analysis and evaluation on the sunroof noise fusion feature set in sequence based on each detection task in the sunroof dynamic and static sound detection task set to obtain a sunroof noise detection task information set; a record integration subunit, the record integration subunit is used to record and integrate the sunroof noise detection task information set to generate the car sunroof dynamic and static sound detection report.

[0059] The intelligent detection system for automobile sunroof movement and stillness integrated with acoustic imaging provided in an embodiment of the present invention can execute the intelligent detection method for automobile sunroof movement and stillness integrated with acoustic imaging provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0060] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0061] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. An intelligent detection method for automobile sunroof dynamic and static noises integrating acoustic imaging, characterized in that: The method comprises: Acquiring the internal space structure information of the target vehicle and the acoustic imager to be used, and performing a layout point analysis on the acoustic imager to be used based on the internal space structure information to determine the layout point information of the acoustic imager; The acoustic imager to be used is installed according to the acoustic imager arrangement point information, and the installed acoustic imager to be used is started to perform real-time dynamic collection of noise data during the opening and closing process of the automobile sunroof to obtain a sunroof noise signal set and a sunroof noise image set; Building a dual channel for acoustic data processing, and synchronously analyzing and processing the skylight noise signal set and the skylight noise image set based on the dual channel for acoustic data processing to obtain a skylight noise multidimensional feature set and a skylight noise spatial feature set; A sunroof movement and sound detection task set is constructed, and based on the sunroof movement and sound detection task set, the sunroof noise multidimensional feature set is mapped to the sunroof noise spatial feature set for fusion evaluation to generate a car sunroof movement and sound detection report.

2. The intelligent detection method for automobile sunroof dynamic and static noises integrating acoustic imaging as claimed in claim 1 is characterized in that: The step of determining the acoustic imager deployment point information includes: Acquire a vehicle sunroof dynamic and static sound detection target, extract acoustic influencing factors based on the vehicle sunroof dynamic and static sound detection target, and determine an acoustic detection influencing factor set, wherein the acoustic detection influencing factor set includes coverage, overlapping area, measurement accuracy, and sensitivity; Generate a vehicle interior space model according to the interior space structure information, and simultaneously obtain device attribute information of the acoustic imager to be applied; Performing preliminary point planning on the vehicle interior space model based on the acoustic detection influencing factor set and the equipment attribute information to obtain initial layout point information; The initial layout point information is simulated and optimized to determine the layout point information of the acoustic imager.

3. The intelligent detection method for automobile sunroof dynamic and static noises fused with acoustic imaging as claimed in claim 2, characterized in that: The obtaining of initial deployment point information includes: Performing acoustic analysis and marking on the vehicle interior space model to determine the skylight area information, the internal obstacle set, and the sound wave propagation and reflection characteristics; Based on the skylight area information, the internal obstacle set and the sound wave propagation and reflection characteristics, the monitoring area is divided to obtain the acoustic monitoring area information; Using the acoustic detection influencing factor set and the equipment attribute information as constraint information, preliminary planning of points is performed within the acoustic monitoring area information to obtain a set of optional deployment points; A threshold value for the number of imagers is set, and based on the threshold value for the number of imagers, the optional deployment point set is optimized for deployment effects to obtain the initial deployment point information.

4. The intelligent detection method for automobile sunroof dynamic and static sound integrating acoustic imaging as claimed in claim 3 is characterized in that: The step of determining the acoustic imager deployment point information includes: Using finite element simulation technology to import the vehicle interior space model and the initial layout point information for simulation, and establish a vehicle acoustic simulation model; The vehicle acoustic simulation model is used to simulate the opening and closing of the automobile sunroof and monitor noise propagation to obtain noise propagation simulation information; Based on the noise propagation simulation information, the effectiveness of the initial deployment point information is evaluated to determine the noise source capture effectiveness parameter; The initial layout point information is iteratively simulated and optimized based on the noise source capture effectiveness parameter by using the vehicle acoustic simulation model to determine the layout point information of the acoustic imager.

5. The intelligent detection method for automobile sunroof dynamic and static noises fused with acoustic imaging as claimed in claim 1, characterized in that: The dual-channel acoustic data processing system includes: Mining and collecting skylight historical noise signal sets and skylight historical noise image sets, performing filtering preprocessing on the skylight historical noise signal sets and skylight historical noise image sets, and obtaining available skylight noise signal sets and available skylight noise image sets; Performing noise feature labeling training on the available skylight noise signal set and the available skylight noise image set respectively to generate an acoustic signal processing network and an acoustic image processing network; The acoustic signal processing network and the acoustic image processing network are integrated in parallel to build the acoustic data processing dual channel.

6. The intelligent detection method for automobile sunroof dynamic and static noises fused with acoustic imaging as claimed in claim 5, characterized in that: The generating of the acoustic signal processing network and the acoustic image processing network comprises: Respectively annotating the available skylight noise signal set and the available skylight noise image set with noise features to obtain a skylight noise signal sample set and a skylight noise image sample set; Using a deep neural network to perform feature recognition training on the skylight noise signal sample set to generate an initial signal processing network; Using a convolutional neural network to perform feature recognition training on the skylight noise image sample set to generate an initial image processing network; The initial signal processing network and the initial image processing network are cross-validated and tuned respectively to generate the acoustic signal processing network and the acoustic image processing network.

7. The intelligent detection method for automobile sunroof dynamic and static noises fused with acoustic imaging as claimed in claim 1, characterized in that: The generating of the automobile sunroof dynamic and static sound detection report comprises: Constructing a skylight acoustic distribution space according to the skylight noise spatial feature set; Mapping the skylight noise multidimensional feature set to the skylight acoustic distribution space according to the spatial distribution information to perform feature matching and fusion to obtain a skylight noise fusion feature set; Based on the sunroof dynamic and static sound detection task set, noise analysis and evaluation are performed on the sunroof noise fusion feature set in sequence to generate a car sunroof dynamic and static sound detection report.

8. The intelligent detection method for automobile sunroof dynamic and static noises fused with acoustic imaging as claimed in claim 7, characterized in that: The step of obtaining a skylight noise fusion feature set comprises: Mapping the skylight noise multidimensional feature set to the skylight acoustic distribution space according to the spatial distribution information to perform feature matching to obtain a skylight noise matching feature set; The matching signal features and spatial features in the skylight noise matching feature set are weightedly fused to obtain the skylight noise fusion feature set.

9. The intelligent detection method for automobile sunroof dynamic and static noises fused with acoustic imaging as claimed in claim 7, characterized in that: The generating of the automobile sunroof dynamic and static sound detection report comprises: Based on each detection task in the sunroof dynamic and static detection task set, noise analysis and evaluation are performed on the sunroof noise fusion feature set in turn to obtain a sunroof noise detection task information set; The sunroof noise detection task information set is recorded and integrated to generate the automobile sunroof dynamic and static noise detection report.

10. The intelligent detection system for automobile sunroof movement and stillness combined with acoustic imaging is characterized by: The system is used to implement the intelligent detection method for automobile sunroof movement and stillness by integrating acoustic imaging according to any one of claims 1 to 9, and the system comprises: A layout point analysis module, the layout point analysis module is used to obtain the internal space structure information of the target vehicle and the acoustic imager to be applied, and perform layout point analysis on the acoustic imager to be applied based on the internal space structure information to determine the layout point information of the acoustic imager; A real-time dynamic acquisition module, wherein the real-time dynamic acquisition module is used to install the acoustic imager to be used according to the acoustic imager layout point information, and start the installed acoustic imager to be used to perform real-time dynamic acquisition of noise data during the opening and closing process of the automobile sunroof, so as to obtain a sunroof noise signal set and a sunroof noise image set; A feature set acquisition module, wherein the feature set acquisition module is used to build a dual channel for acoustic data processing, and based on the dual channel for acoustic data processing, the skylight noise signal set and the skylight noise image set are analyzed and processed synchronously to obtain a skylight noise multidimensional feature set and a skylight noise spatial feature set; A fusion evaluation module is used to construct a sunroof dynamic and static sound detection task set, and based on the sunroof dynamic and static sound detection task set, the multi-dimensional feature set of sunroof noise is mapped to the sunroof noise spatial feature set for fusion evaluation to generate a car sunroof dynamic and static sound detection report.

Images