Method, device, equipment and storage medium for diagnosing abnormal sound of vehicle

Abstract

This application provides a method, apparatus, device, and storage medium for diagnosing abnormal noises in vehicles. The method includes: acquiring sound data from inside the vehicle, the vehicle's vibration frequency, vibration amplitude, and driving status data; determining the vibration level of the road condition based on the vibration frequency and amplitude; determining the abnormal noise data corresponding to the vibration level and driving status data from a preset abnormal noise database, the abnormal noise database including abnormal noise sounds from various vehicle components under different vibration levels and different driving states; performing similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise; and outputting the location of the abnormal noise. This solution accurately filters the abnormal noise database using vibration level and driving status data, and then performs targeted similarity matching, which can reduce the false alarm rate of abnormal noises.

Description

Technical Field

[0001] This application relates to the field of vehicle fault diagnosis, and in particular to a method, apparatus, device, and storage medium for diagnosing abnormal noises in vehicles. Background Technology

[0002] Unusual noises in automobiles are a key issue affecting user experience. With the development of electric vehicles, lightweighting, and intelligentization, the problem of unusual noises is becoming more complex, and identification technology is rapidly evolving from subjective human judgment to objective intelligent diagnosis.

[0003] In automobile manufacturing, 4S shop repairs, and daily user use, abnormal noise detection needs to quickly locate the source of the fault and provide accurate diagnostic results. Currently, the industry generally relies on manual listening or high-cost sensor equipment, which results in cumbersome and time-consuming testing processes that are greatly affected by the environment and personnel experience, easily leading to misdiagnosis. For example, if a user discovers an abnormal noise while driving, they need to go to a 4S shop where technicians will install sensors and conduct multi-condition tests, which is costly and inefficient.

[0004] Therefore, there is an urgent need for an automated method for detecting and diagnosing abnormal noises based on existing vehicle hardware, so as to achieve low-cost, high-efficiency, and high-precision identification and location of abnormal noises. Summary of the Invention

[0005] This application provides a method, apparatus, device, and storage medium for diagnosing abnormal noises in vehicles, in order to improve the speed and accuracy of abnormal noise detection.

[0006] In a first aspect, this application provides a method for diagnosing abnormal noises in a vehicle, the method comprising:

[0007] Acquire sound data inside the vehicle, the vehicle's vibration frequency, vibration amplitude, and driving status data;

[0008] Based on the vibration frequency and the vibration amplitude, the vibration level of the road condition is determined;

[0009] From the preset abnormal noise database, determine the abnormal noise data corresponding to the vibration level and the driving state data. The abnormal noise database includes the abnormal noise sounds of various vehicle components under different vibration levels and different driving states.

[0010] The sound data and the abnormal noise data are matched for similarity to determine the location of the abnormal noise;

[0011] Output the location where the abnormal noise occurs.

[0012] Furthermore, the step of performing similarity matching between the sound data and the abnormal noise database to determine the location of the abnormal noise includes:

[0013] Calculate the similarity between the sound data and each abnormal sound in the abnormal sound data to obtain multiple similarity values;

[0014] If the maximum value among the multiple similarity values ​​is greater than a preset threshold, then it is determined that the vehicle has abnormal noise;

[0015] The source of abnormal noises with a similarity value greater than a preset threshold is identified as the location of the abnormal noise.

[0016] Furthermore, determining the vibration level of the road condition based on the vibration frequency and the vibration amplitude includes:

[0017] The vibration level is determined by calculating the vibration frequency and vibration amplitude using a preset weighted summation.

[0018] Furthermore, the output of the abnormal noise generation location includes:

[0019] The locations where the abnormal noises occur are output in descending order of similarity scores.

[0020] Furthermore, the driving status data includes vehicle speed, acceleration, steering wheel angle, road surface condition, ambient temperature, and ambient humidity.

[0021] Furthermore, the step of performing similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise includes:

[0022] The sound data is input into a classification model pre-trained based on abnormal noise data to determine the location of the abnormal noise.

[0023] Furthermore, before performing similarity matching between the sound data and the abnormal noise data, the method further includes:

[0024] The audio data is filtered to remove background noise.

[0025] Furthermore, the filtering process for the audio data includes:

[0026] Obtain road surface tire noise data corresponding to the current road surface condition;

[0027] The sound data is filtered based on the road surface tire noise data.

[0028] Secondly, this application provides a device for diagnosing abnormal noises in a vehicle, the device comprising:

[0029] The acquisition module is used to acquire sound data inside the vehicle, the vehicle's vibration frequency, vibration amplitude, and driving status data.

[0030] The first determining module is used to determine the vibration level of the road condition based on the vibration frequency and the vibration amplitude.

[0031] The second determining module is used to determine the abnormal noise data corresponding to the vibration level and the driving state data from the preset abnormal noise library data. The abnormal noise library data includes the abnormal noise sounds of various vehicle components under different vibration levels and different driving states.

[0032] The third determining module is used to perform similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise.

[0033] The output module is used to output the location where the abnormal noise occurs.

[0034] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0035] The memory stores computer-executed instructions;

[0036] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0037] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0038] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0039] This application provides a method, apparatus, device, and storage medium for diagnosing abnormal noises in vehicles. The method includes: acquiring sound data from inside the vehicle, the vehicle's vibration frequency, vibration amplitude, and driving status data; determining the vibration level of the road condition based on the vibration frequency and amplitude; determining the abnormal noise data corresponding to the vibration level and driving status data from a preset abnormal noise database, the abnormal noise database including abnormal noise sounds from various vehicle components under different vibration levels and different driving states; performing similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise; and outputting the location of the abnormal noise. This solution accurately filters the abnormal noise database using vibration level and driving status data, and then performs targeted similarity matching, which can reduce the false alarm rate of abnormal noises. Attached Figure Description

[0040] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0041] Figure 1 A flowchart illustrating a method for diagnosing abnormal noises in a vehicle provided in this application. Figure 1 ;

[0042] Figure 2 A flowchart illustrating a method for diagnosing abnormal noises in a vehicle provided in this application. Figure 2 ;

[0043] Figure 3 This application provides a schematic diagram of a system architecture for identifying abnormal noises in vehicles.

[0044] Figure 4 A schematic diagram of the software architecture of the processing layer provided in this application;

[0045] Figure 5 This is a schematic diagram of the network architecture of the remote abnormal noise recognition system provided in this application;

[0046] Figure 6 A schematic diagram of the structure of a device for diagnosing abnormal noises in a vehicle, provided in this application;

[0047] Figure 7 A schematic diagram of the structure of the electronic device provided in this application.

[0048] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0049] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0050] Unusual noises in automobiles refer to unexpected sounds produced during vehicle operation that exceed the normal design range. These mainly include: humming: low-frequency noise caused by resonance or vibration transmission; squeaking: high-frequency sharp sounds caused by friction (such as rubber contacting plastic); and rattling: impact sounds caused by loose parts. Some unusual noises (such as those from the chassis or braking system) may indicate potential malfunctions, and ignoring them could lead to serious accidents.

[0051] Existing abnormal noise detection technologies are mainly divided into three categories:

[0052] Human experience method: This method relies on the engineer's auditory and tactile judgment, using a stethoscope or simulating the abnormal noise condition to locate it. This method is highly subjective, makes it difficult to quantify the characteristics of the abnormal noise, and has limited ability to identify complex abnormal noises (such as high-frequency friction sounds or low-frequency resonance sounds).

[0053] Sensor-assisted analysis: Vibration and sound signals are collected using devices such as accelerometers and microphone arrays, and then combined with spectrum analysis, coherence analysis, or acoustic imaging techniques to locate the source of abnormal noise. However, this method requires specialized equipment, is costly, and has a complex setup, requiring vehicle lifting and multiple sensor placements, which limits its widespread adoption in 4S stores and by individual customers.

[0054] Preliminary explorations driven by AI: Some technologies attempt to classify abnormal noise signals through machine learning models (such as SVM and random forest) or deep learning algorithms (such as CNN), but due to limitations in data annotation quality, insufficient feature extraction dimensions, and weak multimodal information fusion capabilities, there are still problems such as low positioning accuracy and poor generalization ability in practical applications.

[0055] The aforementioned technical solutions generally suffer from the following drawbacks: The testing process is complex, requiring specialized equipment and relying on technical personnel; users often find it difficult to complete independently. Positioning accuracy is insufficient, only able to roughly locate the area of ​​abnormal noise (such as tires or engine compartment), unable to pinpoint specific components. There is a conflict between cost and widespread adoption; high-cost equipment limits the technology's application in small and medium-sized 4S shops and among users, leading to a long-term reliance on manual experience to handle abnormal noise issues. Subjectivity and the risk of misjudgment: The lack of objective quantitative standards makes the solutions susceptible to environmental interference (such as background noise) and differences in personnel experience, resulting in unstable diagnostic results.

[0056] In view of this, this application acquires vehicle vibration and driving data based on existing in-vehicle hardware, then specifically identifies and matches corresponding abnormal noise data from a constructed abnormal noise library, and finally outputs diagnostic results through the vehicle's infotainment system or cloud platform. This technical solution aims for low cost, high efficiency, and high accuracy, solving the problems of reliance on manual experience, high equipment costs, low positioning accuracy, and poor model generalization ability in traditional abnormal noise detection.

[0057] The following describes the technical solution of this application and how it solves the aforementioned technical problems using a vehicle-mounted system as the execution entity, through specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0058] Figure 1 A flowchart illustrating a method for diagnosing abnormal noises in a vehicle provided in this application. Figure 1 ,like Figure 1 As shown, the method includes:

[0059] S101. Acquire sound data inside the vehicle, vehicle vibration frequency, vibration amplitude, and driving status data.

[0060] In one specific implementation, in-vehicle sound data is collected by three high-sensitivity microphone arrays installed at the driver's head position, the center of the dashboard, and the rear seats; for example, the sampling rate is 16kHz and the bit depth is 16bit.

[0061] Vibration frequency and amplitude are obtained through vehicle body acceleration sensors and gyroscopes; for example, the peak value or RMS effective value can be directly obtained from the acceleration signal as the vibration amplitude through a three-axis MEMS accelerometer; the frequency corresponding to the peak value can be obtained by performing a fast Fourier transform or power spectral density analysis on the time-domain acceleration signal.

[0062] Driving status data is read in real time via the vehicle's CAN bus, including instantaneous vehicle speed, acceleration, steering wheel angle, and may also include road conditions, ambient temperature, and ambient humidity identified by cameras.

[0063] S102. Determine the vibration classification of road conditions based on vibration frequency and vibration amplitude.

[0064] Even the same road surface can exhibit different vibration feedback depending on its usage. For example, on a concrete road, in addition to the gaps between concrete blocks, there may also be potholes. Therefore, simply classifying the road surface can be inaccurate. This solution employs vibration classification to avoid the drawbacks of relying solely on road surface classification. There is no limit to the number of vibration classification levels; it can be three, four, five, or more.

[0065] In one implementation, vibration frequency and vibration amplitude can be weighted and fused into a vibration classification.

[0066] In one implementation, vibration frequency and vibration amplitude are respectively assigned to different vibration levels, such as vibration frequency level 2 - vibration amplitude level 3.

[0067] For example, vibration frequencies are divided into low-frequency (<50Hz), mid-frequency (50~120Hz), and high-frequency (>120Hz), and vibration amplitudes are divided into weak (<0.5g), moderate (0.5~2g), and strong (>2g). Then, according to their respective weights (e.g., frequency weight 0.6, amplitude weight 0.4), a weighted sum is obtained to obtain a comprehensive vibration score, which is finally mapped to a vibration classification of 1 to 5, with level 1 representing the most stable road condition and level 5 representing the most severe road condition. For example, a comprehensive score of 0~20 is classified as level 1, and 81~100 is classified as level 5.

[0068] S103. Determine the abnormal noise data corresponding to the vibration level and driving status data from the preset abnormal noise library data. The abnormal noise library data includes the abnormal noise sounds of various vehicle parts under different vibration levels and different driving statuses.

[0069] In this step, the abnormal noise library is pre-built through manual annotation based on real-vehicle road tests under various standard road conditions, different driving states, and different ambient temperatures. The library contains abnormal noise samples from multiple vehicle components, such as the suspension system, chassis connectors, engine compartment, door seals, air conditioning compressor, and braking system. Each sample is simultaneously labeled with its corresponding vibration level, driving state parameters, and time-frequency feature vector. Driving state reflects the vehicle's driving behavior, such as acceleration, braking, and turning; abnormal noises may differ under different driving states within the same road condition.

[0070] During matching, the corresponding subset in the database is directly indexed based on the current vibration level and driving status data to narrow down the matching range.

[0071] S104. Perform similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise.

[0072] In one specific implementation, the sound data is first bandpass filtered (cutoff frequency 20Hz~8kHz) to remove background noise, and then its 13-dimensional MFCC feature vector is extracted. Then, the time similarity and cosine similarity of the feature vector are calculated using the Dynamic Time Warping (DTW) algorithm, respectively, to obtain multiple similarity values. If the maximum similarity value is greater than the preset threshold of 0.85, it is determined that there is an abnormal noise, and the component label corresponding to the source of the abnormal noise with a similarity greater than the threshold is identified as the location of the abnormal noise.

[0073] Other algorithms can be used to calculate similarity, and this application does not impose any restrictions.

[0074] In this embodiment, it is also possible to selectively switch to a pre-trained lightweight CNN classification model, which directly outputs the label with the highest probability of the part.

[0075] S105, Location of abnormal noise.

[0076] In this step, the location of the abnormal noise is displayed on the vehicle's dashboard central control screen. For example, the abnormal noise is located in the left front suspension system, with a similarity of 0.92.

[0077] Optionally, the top three possible sources of abnormal noise can be listed in descending order of similarity scores.

[0078] In addition, maintenance suggestions can be output by combining the fault severity mapping table, such as "stop and repair immediately" or "it is recommended to go to the 4S store for inspection as soon as possible". The driver can also be notified through the voice broadcast module or mobile APP.

[0079] This application provides a method for diagnosing abnormal noises in vehicles. The method includes: acquiring sound data from inside the vehicle, the vehicle's vibration frequency, vibration amplitude, and driving status data; determining the vibration level of the road condition based on the vibration frequency and amplitude; identifying abnormal noise data corresponding to the vibration level and driving status data from a pre-set abnormal noise database, the database including abnormal noise sounds from various vehicle components under different vibration levels and driving states; performing similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise; and outputting the location of the abnormal noise. This solution accurately filters the abnormal noise database using vibration level and driving status data, and then performs targeted similarity matching, which can reduce the false alarm rate for abnormal noises.

[0080] Based on the above embodiment 1, the similarity matching process will be described below.

[0081] Figure 2 A flowchart illustrating a method for diagnosing abnormal noises in a vehicle provided in this application. Figure 2 ,like Figure 2 As shown, it includes the following steps:

[0082] S1041. Calculate the similarity between the sound data and the abnormal noise data for each abnormal noise sound, and obtain multiple similarity values.

[0083] Specifically, the audio signal is first divided into multiple frames, each with a selectable frame length of 25ms and a frame shift of 10ms. A Hamming window or other window function is applied to each frame to reduce boundary effects and ensure a smoother signal.

[0084] MFCC features are extracted from each frame of the signal. A commonly used feature dimension is 13, which can effectively capture the time and frequency domain information of the sound. For each segment of sound data, a 13-dimensional MFCC feature vector is calculated through the above steps, forming a digital representation of the sound signal.

[0085] Each abnormal noise in the abnormal noise data also undergoes the same framing, windowing, and MFCC feature extraction process to obtain a set of corresponding feature vectors. These feature vectors are stored in the abnormal noise library and are already bound to the corresponding vibration classification and driving status data.

[0086] For each segment of real-time acquired sound data, the similarity is calculated by comparing it with the feature vector of each abnormal sound in the abnormal sound database. The similarity calculation algorithm can be selected from commonly used algorithms, such as:

[0087] Dynamic Time Warping (DTW) algorithm: used to calculate the similarity of time series data (such as sound signals), and can align signals of different lengths on the time axis and calculate similarity.

[0088] Calculate cosine similarity: Calculate the angular distance between two feature vectors to reflect their similarity.

[0089] After the calculation is completed, a similarity value will be obtained between each real-time sound data and each abnormal sound in the abnormal sound database. In the end, multiple similarity values ​​will be obtained, each corresponding to a different abnormal sound sample.

[0090] S1042. If the maximum value among multiple similarity values ​​is greater than a preset threshold, then it is determined that the vehicle has abnormal noise.

[0091] Set a preset threshold, such as 0.85, which serves as the threshold for determining whether an abnormal noise has occurred. This threshold can be adjusted based on experimental data and actual testing. A higher threshold results in higher detection accuracy; a lower threshold increases detection sensitivity.

[0092] The system compares multiple similarity scores calculated each time and takes the maximum value. This maximum similarity value is then compared to a preset threshold. If the maximum similarity value is greater than or equal to the preset threshold, the vehicle is considered to have made an abnormal noise. If the maximum similarity value is less than the preset threshold, the vehicle is considered not to have made an abnormal noise.

[0093] This judgment mechanism ensures that only situations highly similar to abnormal sounds are identified as abnormal sounds, thus avoiding false alarms.

[0094] S1043. The source of abnormal noise with a similarity value greater than a preset threshold is determined as the location of the abnormal noise.

[0095] All abnormal noise samples with a similarity greater than a preset threshold are used as candidate abnormal noise locations. If the similarity with multiple components in the abnormal noise database is greater than the threshold, then these components are all possible locations where abnormal noise is occurring.

[0096] Each noise sample is associated with a specific component, such as the suspension system, door seals, or engine compartment. In this step, the specific component causing the noise is identified by matching it with the corresponding component tags in the noise database.

[0097] In this embodiment, by matching sound data with a similarity value greater than a threshold with the corresponding component tags in the abnormal noise database, the source of abnormal noise can be effectively located. The system can accurately identify the specific location where the abnormal noise occurs.

[0098] In some embodiments, if the similarity values ​​of multiple candidate components are greater than a threshold, they can be sorted according to the size of the similarity values, and the components with the highest similarity can be selected as the final output of the abnormal noise location.

[0099] Based on Example 1 above, in addition to algorithmic calculation, similarity matching can also be performed directly using a large model. An example is described below.

[0100] First, a classification model is trained by collecting and labeling a large amount of known vehicle noise data, such as recording the vehicle part corresponding to each sound. This data covers various types of noises and parts, such as the engine, tires, suspension system, and transmission. Based on the characteristics of the sound data, appropriate machine learning algorithms are selected, such as Deep Neural Networks (DNNs), Convolutional Neural Networks (CNNs), and Support Vector Machines (SVMs). The labeled noise dataset is used to train the classification model. Through backpropagation and optimization algorithms, a model capable of determining the location of the noise based on the input sound features is trained. During vehicle inspection or use, the collected sound data is fed into the trained classification model. The model calculates the similarity between the input sound features and each category (i.e., part), and outputs the most likely location of the noise. For example, if the sound data features best match the engine part, the model will determine that the noise originates in the engine part.

[0101] In some embodiments, the above model similarity matching steps are integrated into an automated detection system, enabling the vehicle to monitor, identify, and locate abnormal noise sources in real time during operation. The system may also incorporate data from other sensors (such as vibration and temperature) to further improve the accuracy and robustness of abnormal noise location.

[0102] Based on Example 1, the sound data used for similarity matching is either the original collected sound data or sound data filtered using a bandpass filter. However, in some scenarios, road surface noise can affect sound matching. To avoid background noise interfering with the abnormal noise analysis, the road surface noise needs to be filtered. First, the current road surface condition, i.e., the type of road surface (cement road, asphalt road, cobblestone road, etc.), is acquired using a camera. Then, the road surface noise data corresponding to the current road surface condition is acquired. This tire noise is pre-set. By filtering the sound data, low-frequency noise related to tire noise can be effectively removed, thereby improving the accuracy of subsequent abnormal noise analysis.

[0103] Tire noise is the most significant background noise during vehicle operation, especially under different road conditions, and it can affect the accuracy of the entire sound data. Removing this irrelevant noise ensures the clarity of subsequent data analysis. After eliminating tire noise and other background noise, the system can more accurately capture abnormal noise signals caused by the vehicle itself, enhancing the sensitivity and reliability of abnormal noise diagnosis. By acquiring tire noise data corresponding to road conditions in real time, the system can adaptively adjust the filtering strategy according to different road conditions, avoiding distortion or missed diagnoses caused by simple filtering methods.

[0104] In some embodiments, tire noise levels vary at different vehicle speeds, and the tire noise data to be eliminated needs to be determined based on vehicle speed during filtering. Specifically, a dynamic filtering model can be established based on the vehicle's actual driving conditions (such as vehicle speed and acceleration) and road surface information, allowing the filtering strategy to be adjusted in real time to ensure optimal noise elimination under various operating conditions. It should be noted that in this processing method, the abnormal noise data in the abnormal noise database is also the data for which tire noise has been pre-eliminated.

[0105] When matching abnormal noise data from the abnormal noise database based on driving status data, in addition to vehicle speed, acceleration, and steering wheel angle, road conditions, ambient temperature, ambient humidity, and tire pressure can also be considered; all of these parameters will affect the abnormal noise data.

[0106] Building upon Example 1, some embodiments integrate vibration frequency and amplitude to determine the vibration level. In one implementation, the weights of vibration frequency and amplitude are statically set. While this method is simple, it may not accurately reflect the vehicle's true condition. For example, during low-speed urban driving, the contact frequency between the wheels and the road surface is low, and the vibration amplitude may be large. However, static weights cannot flexibly reflect this, leading to diagnostic errors. Conversely, during high-speed driving, the vibration frequency may be high, but the amplitude is small. In this case, static weights may be biased towards amplitude, failing to accurately distinguish abnormalities in vehicle components. To improve the accuracy of abnormal noise detection, the weights of vibration frequency and amplitude can be dynamically determined. The weight allocation can be automatically adjusted based on different driving states (such as acceleration, braking, turning, etc.) and road conditions (such as flat, bumpy), ensuring that the vibration score more closely matches actual driving conditions.

[0107] The steps to dynamically determine weights:

[0108] The vehicle's vibration frequency and amplitude data are collected in real time by sensors such as accelerometers and gyroscopes. At the same time, the vehicle's driving status information (such as vehicle speed, acceleration, braking status, etc.) and road condition information (such as road type, road surface smoothness, tire noise, etc.) are also obtained.

[0109] Then, the weights are adjusted based on the driving conditions and road surface conditions. The impact of vibration frequency and amplitude varies under different driving conditions (such as acceleration, deceleration, and constant speed driving). For example, when the vehicle accelerates, the vibration frequency may be more prominent, while during braking, the vibration amplitude may become the primary diagnostic criterion. The system adjusts the weights of frequency and amplitude based on real-time driving data. For instance, during acceleration, the frequency weight increases, and the amplitude weight relatively decreases; during braking, the amplitude weight increases, and the frequency weight decreases; during smooth driving, the weights of frequency and amplitude are relatively balanced; under different road surface conditions, such as on bumpy roads, the amplitude has a greater impact, while on flat roads, frequency changes are more significant, and the system adjusts the weights according to road conditions.

[0110] During system operation, machine learning models can be used to continuously optimize weight adjustment rules based on historical data and real-time diagnostic results. The weight allocation of frequency and amplitude is automatically learned and adjusted according to the correlation between historical fault data and sensor data. For example, by analyzing the correlation between frequency and amplitude and different types of faults (such as suspension system faults, braking system faults, etc.), the optimal weight values ​​are automatically adjusted.

[0111] Based on the comprehensive vibration score calculated using dynamic weights, the system maps the vibration score to a vibration level of 1-5 and issues real-time warnings to the driver via the vehicle's dashboard or a mobile application. By adjusting the weights in real time, the system can accurately identify vehicle anomalies under different operating conditions, improving the accuracy of abnormal noise detection.

[0112] After each vibration score, the system compares and analyzes the results with subsequent actual repairs to assess the accuracy of the diagnosis. If the diagnostic results do not match, the system will automatically adjust the weight settings based on feedback information. For example, when the system detects that a certain type of fault frequently occurs in a specific vibration frequency range, it will correspondingly increase the weight of that frequency range and decrease the weight of other frequency ranges.

[0113] This embodiment, by dynamically adjusting the weights of vibration frequency and vibration amplitude, enables more accurate analysis of vibration data based on different driving conditions and road surface conditions, thereby improving the accuracy and robustness of the vehicle abnormal noise diagnosis system.

[0114] The following is a specific example.

[0115] Step 1. Application selection.

[0116] Access the abnormal noise diagnostic system through the car's infotainment system menu via app or browser. Activate the abnormal noise detection mode.

[0117] Step 2. Road test and relevant signals.

[0118] Click the recording interface in the app or webpage. The vehicle's recording software will then start recording sound. Simultaneously, the camera detects road surface conditions, such as concrete, cobblestone, manhole cover, or asphalt. The chassis control domain detects vehicle speed, acceleration, and height changes.

[0119] Step 3. Signal processing.

[0120] The system generates road surface and sound waveforms. The app processes the collected sound signals via the vehicle's infotainment chip or a browser connected to the internet.

[0121] Step 4. Classify according to different vibration frequencies and amplitudes.

[0122] The vehicle speed, acceleration, vehicle height sensor readings, steering torque, steering angle, and electromagnetic damper activation signals are obtained via CAN signals. These signals are then used to confirm vehicle acceleration, deceleration, steering, road surface conditions, and MEMS accelerometer data. The vibration frequency and amplitude are determined based on the collected MEMS accelerometer data.

[0123] 5. Remove background noise.

[0124] Use filtering or audio trimming functions to remove speech or unusual background noise.

[0125] 6. Identify whether any abnormal noises are occurring.

[0126] The system uses an app to process the collected sound signals via the vehicle's infotainment chip or a web browser. The signals are then compared to a database of abnormal noises using machine learning or neural network algorithms. This process identifies potential fault locations and generates corresponding descriptions and similarity scores. For example, VIN:091001-Cement Road-Steering Gear Noise - 98%, VIN:091001-Cobblestone Road-Action Cabin Noise - 82%.

[0127] 7. The type of abnormal noise is displayed on the vehicle's infotainment system.

[0128] The system identifies potential fault locations, generates corresponding text and similarity scores, and displays these on the instrument panel via the vehicle's CAN signal. Examples include: VIN:091001-Cement Road-Steering Gear Noise-98%, VIN:091001-Cobblestone Road-Engine Cabin Noise-82%.

[0129] 8. Cloud-based statistics.

[0130] By processing data in the cloud, we can retrieve a list of vehicles with abnormal noises and their information for each day or a fixed time period, and then perform statistical analysis.

[0131] The system architecture for vehicle noise recognition is as follows: Figure 3 As shown, it includes an input layer, a processing layer, and an output layer;

[0132] Input layer:

[0133] 1. Car Microphone. Car microphones are typically small, black / gray, round / square components with small sound-collecting holes. Common locations include: the front dome light area, near the steering wheel, the center console multimedia area, and the rear roof / headrests. These microphones are used to collect sound signals during road tests.

[0134] 2. Vehicle-mounted cameras. These include front-view cameras, rear-view cameras, and side-view cameras.

[0135] The system uses various cameras to determine the road conditions and type where the vehicle is located. It then processes the sound signal to match the type of sound it is producing.

[0136] 3. Vehicle motion status signal.

[0137] Environmental and road conditions. Sensors, either built into the vehicle or added to it, can detect signals such as temperature and humidity when abnormal noises occur. Road conditions are identified using the vehicle's forward-facing camera and AI detection algorithms (such as YOLO).

[0138] Control signals input by the driver, such as steering torque, accelerator pedal, and brake pedal opening signals, are obtained through CAN signals and other means.

[0139] And vehicle status signals, including vehicle speed, acceleration, and Z-axis vibration acceleration of the vehicle body.

[0140] Multimodal integrated verification records the environmental conditions and vehicle status-related signals when the abnormal vehicle noise occurred. The data is shown in Table 1.

[0141] Table 1. Collected Data

[0142]

[0143] Classification and Recognition: By identifying the vibration amplitude and frequency of the vehicle, different road conditions are matched (e.g., vibration level: Level 1, Level 2, Level 3).

[0144] Figure 4 This is a schematic diagram of the software architecture of the processing layer. (Processing layer architecture reference) Figure 4 .

[0145] Option 1: Based on the vehicle's MCU and EPPROM. Use machine learning or neural network methods to process and identify abnormal noise signals.

[0146] Option 2: Connect to the internet via a communication module. This involves the architecture of the in-vehicle application app or browser with the cloud. Figure 5 as follows:

[0147] In this embodiment, the neural network model used needs to collect, filter, perform FFT transformation, group and train the abnormal noise videos of each brand of vehicle, and generate abnormal noise audio-abnormal noise type library labels.

[0148] During model diagnostics: The acquired video or audio is compared with library files. Methods such as cosine similarity, clustering, support vector machines, and neural networks are used to identify relevant similarities.

[0149] Output layer:

[0150] Based on the road test audio, the backend processing layer processes and calculates to output whether there are any abnormal noises. If there are no abnormalities, it displays: "Sound level OK," along with the corresponding audio graph and spectrum. If there are abnormal noises, it displays the type of abnormal noise and its spectrum.

[0151] In summary, the proposed solution has the following advantages:

[0152] 1. Ease of use for noise identification methods. Operation is simple, requiring only an updated in-vehicle app or browser. It can quickly and easily identify intermittent abnormal noises and can be applied to vehicle assembly line production and 4S dealership repairs.

[0153] 2. Objective and quantifiable sound characteristics. This design processes and learns from normal and abnormal sound characteristic signals to generate anomaly detection boundaries. This prevents misjudgments due to insufficient experience or memory errors.

[0154] Meanwhile, by iteratively analyzing the sound signals of normal and abnormal vehicles, the model's accuracy can be stabilized and gradually improved. Through iterative analysis of features such as amplitude and spectral group values ​​of different sound characteristics, and by updating the judgment criteria via OTA updates to the app or cloud-based updates to the abnormal sound library file, the model's accuracy can become increasingly precise.

[0155] 3. Low cost. The data collection equipment utilizes a vehicle-mounted microphone, in-vehicle infotainment system, and instrument cluster display. Compared to devices such as accelerometers and decibel meters, it is lower in cost and more convenient to collect data.

[0156] Figure 6 A schematic diagram of the structure of a device for diagnosing abnormal noises in a vehicle, as provided in this application, is shown below. Figure 6 As shown, the vehicle abnormal noise diagnosis device 40 provided in this embodiment includes:

[0157] The acquisition module 401 is used to acquire sound data inside the vehicle, vibration frequency, vibration amplitude, and driving status data of the vehicle.

[0158] The first determining module 402 is used to determine the vibration level of the road condition based on the vibration frequency and vibration amplitude.

[0159] The second determining module 403 is used to determine the abnormal noise data corresponding to the vibration level and driving state data from the preset abnormal noise library data. The abnormal noise library data includes the abnormal noise sounds of various vehicle parts under different vibration levels and different driving states.

[0160] The third determining module 404 is used to perform similarity matching between sound data and abnormal noise data to determine the location of the abnormal noise.

[0161] Output module 405 is used to output the location where the abnormal noise occurs.

[0162] Optionally, the third determining module 404 is specifically used for:

[0163] Calculate the similarity between the sound data and the abnormal noise data for each abnormal noise sound, and obtain multiple similarity values;

[0164] If the maximum value among multiple similarity values ​​is greater than a preset threshold, then the vehicle is determined to have abnormal noise.

[0165] The source of abnormal noises with a similarity value greater than a preset threshold is identified as the location of the abnormal noise.

[0166] Optionally, the first determining module 402 is specifically used for:

[0167] The vibration level is determined by calculating the weighted sum of the vibration frequency and vibration amplitude using preset weights.

[0168] Optionally, output module 405 is specifically used for:

[0169] The locations of the abnormal noises are output in descending order of similarity scores.

[0170] Optionally, driving status data includes vehicle speed, acceleration, steering wheel angle, road surface condition, ambient temperature, and ambient humidity.

[0171] Optionally, the third determining module 404 is specifically used for:

[0172] The sound data is input into a classification model that has been pre-trained based on abnormal noise data, and the location of the abnormal noise is determined by the classification model.

[0173] Optionally, the device also includes a preprocessing module for:

[0174] The audio data is filtered to remove background noise.

[0175] Furthermore, the preprocessing module is specifically used for:

[0176] Obtain road surface tire noise data corresponding to the current road surface condition;

[0177] The sound data is filtered based on the road surface tire noise data.

[0178] The apparatus provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0179] Figure 7 A schematic diagram of the structure of the electronic device provided in this application. Figure 7 As shown, the electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus 504.

[0180] In a specific implementation, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to perform the above-described method.

[0181] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0182] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0183] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0184] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0185] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0186] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0187] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0188] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0189] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0190] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0191] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0192] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0193] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0194] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method of diagnosing an abnormal sound of a vehicle, characterized by, The method includes: Acquire sound data inside the vehicle, the vehicle's vibration frequency, vibration amplitude, and driving status data; Based on the vibration frequency and the vibration amplitude, the vibration level of the road condition is determined; From the preset abnormal noise database, determine the abnormal noise data corresponding to the vibration level and the driving state data. The abnormal noise database includes the abnormal noise sounds of various vehicle components under different vibration levels and different driving states. The sound data and the abnormal noise data are matched for similarity to determine the location of the abnormal noise; Output the location where the abnormal noise occurs.

2. The method of claim 1, wherein, The step of performing similarity matching between the sound data and the abnormal noise database to determine the location of the abnormal noise includes: Calculate the similarity between the sound data and each abnormal sound in the abnormal sound data to obtain multiple similarity values; If the maximum value among the multiple similarity values ​​is greater than a preset threshold, then it is determined that the vehicle has abnormal noise; The source of abnormal noises with a similarity value greater than a preset threshold is identified as the location of the abnormal noise.

3. The method of claim 2, wherein, The step of determining the vibration level of road conditions based on the vibration frequency and the vibration amplitude includes: The vibration level is determined by calculating the vibration frequency and vibration amplitude using a preset weighted summation.

4. The method according to any one of claims 1 to 3, characterized in that, The driving status data includes vehicle speed, acceleration, steering wheel angle, road surface condition, ambient temperature, and ambient humidity.

5. The method of claim 1, wherein, The step of performing similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise includes: The sound data is input into a classification model pre-trained based on abnormal noise data to determine the location of the abnormal noise.

6. The method according to any one of claims 1 to 3, characterized in that, Before performing similarity matching between the sound data and the abnormal noise data, the method further includes: The audio data is filtered to remove background noise.

7. The method of claim 6, wherein, The filtering process for the audio data includes: Obtain road surface tire noise data corresponding to the current road surface condition; The sound data is filtered based on the road surface tire noise data.

8. An apparatus for diagnosing an abnormal sound of a vehicle, characterized by comprising: The device includes: The acquisition module is used to acquire sound data inside the vehicle, the vehicle's vibration frequency, vibration amplitude, and driving status data. The first determining module is used to determine the vibration level of the road condition based on the vibration frequency and the vibration amplitude. The second determining module is used to determine the abnormal noise data corresponding to the vibration level and the driving state data from the preset abnormal noise library data. The abnormal noise library data includes the abnormal noise sounds of various vehicle components under different vibration levels and different driving states. The third determining module is used to perform similarity matching between the sound data and the abnormal noise data to determine the location of the abnormal noise. The output module is used to output the location where the abnormal noise occurs.

9. An electronic device, comprising: include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

Images