A method and system for suppressing noise of a rear main reduction assembly of a vehicle

Abstract

The application discloses a kind of automobile rear main reduction assembly noise suppression method and system, belong to automobile parts noise control technical field.It includes: by trained recurrent neural network, noise data and vibration data are handled, obtain the noise prediction signal of next time;Recurrent neural network is deployed in FPGA chip, to accelerate the convolution operation of weight matrix;The real-time working condition data of automobile is acquired, and according to real-time working condition data and noise prediction signal, construct the real-time feature vector of the real-time characteristic of noise and working condition, according to real-time feature vector, utilize fuzzy rule base to carry out fuzzy reasoning, generate noise control strategy, to carry out noise suppression by corresponding noise reverse sound wave;Improved fuzzy C means clustering method is used to construct and dynamically update fuzzy rule base;Noise reverse sound wave is generated by acoustic system comprising not less than two sound field generating units.Can improve the stability of noise reduction result, solve the problem that existing rear main reduction assembly noise control effect is not ideal.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automobile part noise control, in particular to a rear main reduction assembly noise suppression method and system. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] With the rapid development of the automobile industry and the continuous improvement of people's living standards, people's requirements for automobile ride comfort are getting higher and higher, and NVH (Noise, Vibration, and Harshness) is one of the main factors to be considered.

[0004] During the working process of the automobile RDU (Rear Drive Unit, rear main reduction assembly), due to the existence of its own dynamic unbalance, the received unsteady excitation force and the transmission error in the meshing process, a kind of medium-high frequency noise, called gear whine, will be caused. The excitation force generated at the same time will cause torsional and bending vibration centered on the transmission shaft. This vibration centered on the transmission shaft is transmitted to the vehicle body through the suspension arm and the body connection point, thereby producing noise that can be perceived by the passengers. These noises not only affect the ride experience of the passengers in the vehicle, but also may have a negative impact on the overall vehicle.

[0005] Some traditional passive noise reduction measures are difficult to effectively suppress noise in a wide frequency band range, and at the same time, optimizing the gear design parameters and improving the vibration transmission path of the rear main reduction assembly face the problems of high cost, complex structure, etc. In order to solve the noise problem, the specific ways currently adopted by the mainstream are as follows:

[0006] (1) Installing sound insulation board or sound insulation cotton on the vehicle body floor; blocking and absorbing the noise path; (2) Pressing the torsional damper at the RDU flange to optimize the vibration and thereby reduce the noise; (3) Modifying the software strategy in the controller to disconnect and reduce the torque of the four-wheel drive system at the risk speed; (4) Down-regulating the suspension stiffness of the subframe RDU connection point to absorb the vibration through the bushing. However, the above-mentioned methods still have certain limitations, and the specific problems are as follows:

[0007] 1. The vehicle chassis is used in a harsh environment, and the chassis sound insulation product will fail due to stone impact and water immersion, while the sound insulation product in the vehicle compartment will compress the effective use space of the passengers, increase the production cost, and at the same time, the effect is limited when dealing with low-frequency noise. The wavelength of low-frequency noise is relatively long, which can easily bypass the barrier of traditional sound insulation materials and enter the vehicle.

[0008] 2. The elastic elements (such as springs) and damping elements (such as rubber) inside the torsional shock absorber will experience wear and aging during long-term use. For example, rubber parts may lose elasticity due to the harsh temperatures around the exhaust system, leading to a decline in performance and affecting the normal operation of the torsional shock absorber. Furthermore, the damping effect of the torsional shock absorber will gradually decrease with increasing usage time, potentially requiring periodic replacement, which increases operating costs and maintenance complexity.

[0009] 3. Impact on Four-Wheel Drive Performance: Disconnecting the four-wheel drive system will cause passengers inside the vehicle to experience a interruption in power output, followed by a delay in the recovery of power. Disconnecting the four-wheel drive system and reducing torque at high speeds will subject the transmission components to significant shocks. This sudden change in torque will alter the stress distribution between components such as the drive shaft and differential, easily leading to accelerated wear of the transmission components. Over time, this may shorten the lifespan of these components and increase maintenance costs.

[0010] 4. Adjusting the bushing stiffness of the subframe increases the dynamic envelope of the RDU, reduces the safety clearance with surrounding components, and increases the overall vehicle layout risk.

[0011] Existing noise suppression methods for the rear main reduction gear do not take these influencing factors into account, resulting in unsatisfactory control effects and low control precision for the rear main reduction gear assembly, which can easily lead to passenger complaints. Summary of the Invention

[0012] To address the shortcomings of existing technologies, this invention provides a method, system, electronic device, computer-readable storage medium, and computer program product for noise suppression of a vehicle's rear main reduction assembly. This method accurately predicts noise trends and performs active noise reduction in conjunction with the vehicle's real-time operating conditions, thereby improving the noise suppression effect of the rear main reduction assembly.

[0013] In a first aspect, the present invention provides a method for noise suppression of a vehicle's rear main reduction gear assembly;

[0014] A method for noise suppression of a vehicle's rear main reduction assembly includes:

[0015] Acquire noise data in the cockpit and vibration data of the rear main reduction assembly, and calculate the difference between the noise data, vibration data and vehicle calibration data.

[0016] If the difference is greater than a preset threshold, the noise data and vibration data are processed by the trained recurrent neural network to obtain the noise prediction signal for the next moment; the recurrent neural network is deployed on an FPGA chip to accelerate the convolution operation of the weight matrix.

[0017] Real-time working condition data of the automobile is acquired, a real-time feature vector representing noise and working condition is constructed according to the real-time working condition data and a noise prediction signal, a noise control strategy is generated by fuzzy reasoning using a fuzzy rule base according to the real-time feature vector, so as to suppress noise by corresponding noise reverse sound waves;

[0018] The fuzzy rule base is constructed and dynamically updated by using an improved fuzzy C-means clustering method; the noise reverse sound waves are generated by a sound system comprising no less than two sound field generating units, the phase of the noise reverse sound waves is opposite to that of the noise prediction signal, and the amplitude of the noise reverse sound waves is dynamically adjusted according to the noise control strategy.

[0019] In some embodiments, before processing the noise data and the vibration data by the trained recurrent neural network, the method further comprises:

[0020] The noise data and the vibration data are sequentially processed by the FPGA chip to generate corresponding noise frequency domain signals.

[0021] In some embodiments, the recurrent neural network is trained with the minimization of the mean square error of the time-frequency features of the noiseless reference signal and the actual working condition vibration signal as the target.

[0022] In some embodiments, constructing a real-time feature vector representing noise and working condition according to real-time working condition data and a noise prediction signal comprises:

[0023] The noise prediction signal is subjected to time domain feature extraction to obtain a corresponding root mean square value; the noise prediction signal is sequentially subjected to band-pass filtering and time-frequency conversion to obtain a noise prediction frequency domain signal and perform feature extraction to determine a meshing frequency and a sideband amplitude; the real-time working condition data is subjected to normalization processing to obtain a standard speed and a standard torque;

[0024] The root mean square value, the meshing frequency, the sideband amplitude, the standard speed and the standard torque are combined to generate the real-time feature vector.

[0025] In some embodiments, the fuzzy reasoning using the fuzzy rule base according to the real-time feature vector to generate the noise control strategy specifically comprises: calculating a real-time control parameter by weighting the membership degrees of each rule in the fuzzy rule base corresponding to the real-time feature vector and outputting a corresponding noise control strategy.

[0026] In some embodiments, the fuzzy rule base is constructed and dynamically updated by using an improved fuzzy C-means clustering method, which comprises:

[0027] Vibration signals, noise signals and working condition parameters are acquired and preprocessed and subjected to feature extraction to obtain feature vectors;

[0028] The feature vectors are clustered by a fuzzy C-means clustering method, fuzzy rule nodes are divided, and the fuzzy rule nodes are stored in a fuzzy rule library together with corresponding fuzzy rules.

[0029] The clustering center and width of the membership function are dynamically adjusted according to the real-time feature vector.

[0030] In a second aspect, the present application provides a rear main reduction assembly noise suppression system for a vehicle.

[0031] The rear main reduction assembly noise suppression system for the vehicle comprises:

[0032] The comparison module is configured to obtain noise data in a cabin and vibration data of the rear main reduction assembly, and calculate difference values of the noise data and the vibration data from vehicle calibration data.

[0033] The noise suppression module is configured to, if the difference value is greater than a preset threshold, process the noise data and the vibration data by using a trained recurrent neural network to obtain a noise prediction signal at a next time point.

[0034] Real-time working condition data of the vehicle are obtained, real-time feature vectors representing noise and working conditions are constructed according to the real-time working condition data and the noise prediction signal, fuzzy inference is performed on the real-time feature vectors by using a fuzzy rule library to generate a noise control strategy, and noise suppression is performed on corresponding noise counter sound waves.

[0035] The fuzzy rule library updates membership function parameters and control strategies online through self-organizing learning, and dynamically learns a mapping relationship between features and noise control by using a fuzzy clustering method; the noise counter sound waves are generated by a full-cabin loudspeaker and a headrest loudspeaker, a phase of the noise counter sound waves is opposite to that of the noise prediction signal, and an amplitude of the noise counter sound waves is dynamically adjusted according to the noise control strategy.

[0036] In a third aspect, the present application provides an electronic device.

[0037] The electronic device comprises a memory, a processor, and a computer program stored in the memory, and the processor executes the computer program to implement steps of the rear main reduction assembly noise suppression method for the vehicle.

[0038] In a fourth aspect, the present application provides a computer readable storage medium.

[0039] The computer readable storage medium stores a computer program / instruction, and the computer program / instruction is executed by a processor to implement steps of the rear main reduction assembly noise suppression method for the vehicle.

[0040] In a fifth aspect, the present application provides a computer program product.

[0041] A computer program product comprises computer programs / instructions that, when executed by a processor, implement the steps of the above-mentioned automobile rear main reduction assembly noise suppression method.

[0042] Compared with the prior art, the present application has the following advantages:

[0043] 1、The technical scheme provided by the present application models the time series of the noise signal through a recurrent neural network, which is deployed on an FPGA chip, and the high-speed parallel processing capability of the FPGA can accelerate the learning and analysis process of the RNN on the noise signal; when processing complex RDU non-stationary noise, the RNN can quickly learn the time series characteristics of the noise using the computing power of the FPGA, thereby more accurately predicting the trend of the noise and providing a more effective basis for active noise reduction.

[0044] 2、The technical scheme provided by the present application, the RNN can continuously adjust its model parameters according to the input noise signal, realizing adaptive RDU signal processing; combined with the reconfigurability of the FPGA, the entire system can more flexibly adapt to different noise environments; when the type or intensity of the environmental noise suddenly changes, the RNN-FPGA system can quickly adjust its noise reduction strategy, re-optimize the parameters of the digital filter or adjust the generated anti-noise signal to maintain good noise reduction effect.

[0045] 3、The technical scheme provided by the present application, the FPGA provides efficient computing resources at the hardware level, and the RNN provides powerful sequence data processing capability at the software algorithm level, and the combination of the two can improve the overall performance of the present application in terms of power consumption, computing speed and noise reduction effect; compared with using traditional noise reduction methods alone or relying solely on software-implemented RNN noise reduction, this combination can achieve faster and more accurate noise reduction operation at lower power consumption.

[0046] 4、The technical scheme provided by the present application generates noise reverse sound waves in cooperation with the full-cabin speaker and the headrest speaker, and the entire vehicle audio system can process the noise in the entire vehicle cabin space, macroscopically reducing the overall noise level in the vehicle, and the headrest speaker system can provide personalized noise reduction supplement for each seat area on the basis of the overall noise reduction of the vehicle.

[0047] 5、The technical scheme provided by the present application, when generating a noise control strategy, the real-time working condition data of the vehicle and the noise prediction data of the next moment are combined to adaptively update the fuzzy rule base to adapt to the changing real-time state of the vehicle, improve the accuracy of noise control, and ensure the noise reduction effect. BRIEF DESCRIPTION OF DRAWINGS

[0048] The accompanying drawings, which form a part of this specification, are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification. The embodiments of these drawings are of illustrative embodiments of the application and, as such, are not intended to limit or restrict the scope of the application to the specific embodi ments presented in the drawings.

[0049] Figure 1 A flowchart of the automobile rear main reduction assembly noise suppression method provided by the embodiment of the application is shown in the figure.

[0050] Figure 2 A logic diagram of the automobile rear main reduction assembly noise suppression method provided by the embodiment of the application is shown in the figure.

[0051] Figure 3 A structure diagram of the headrest provided by the embodiment of the application is shown in the figure.

[0052] Figure 4 A computing structure diagram provided by the embodiment of the application is shown in the figure.

[0053] Figure 5 An architecture diagram of the automobile rear main reduction assembly noise suppression system provided by the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0054] It should be noted that the following detailed description is merely exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0055] It is to be understood that the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. It should be noted that the terms "comprises," "comprising," "includes," "including," and the like can be used herein. Similarly, the terms "comprises", "comprising", "includes", "including" and the like are not intended to exclude many embodiments from the technical scope of the present application. Furthermore, the terms "first", "second", third", "fourth" and the like in the description and in the claims, if any, are used for distinguishing between similar elements and not necessarily for describing a sequential or chronological order. It is to be understood that the terms so used are interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of use in either orientation.

[0056] The embodiments in the present application and the features in the embodiments can be combined with each other under the condition of no conflict.

[0057] Embodiment One

[0058] The existing noise control method of the automobile rear main reduction assembly mainly focuses on the improvement of the automobile structure, affects the service life and maintenance cost of the automobile parts, and the noise control effect is not ideal, which affects the riding experience of the passengers, therefore, the application provides a noise suppression method of automobile rear main reduction assembly, the recurrent neural network is deployed in the FPGA chip, the superior computing performance is used to improve the data processing efficiency, and the real-time working condition parameters of the automobile are combined to perform adaptive updating, and the accuracy of noise suppression is improved.

[0059] Next, in combination with Figures 1-4 , a noise suppression method of automobile rear main reduction assembly disclosed in the embodiment is described in detail. The noise suppression method of automobile rear main reduction assembly comprises the following steps:

[0060] S1, in the four-wheel drive mode of the automobile, noise data in the cabin and vibration data of the rear main reduction assembly are acquired, the difference between the noise data and the vibration data and the vehicle calibration data is calculated, if the difference is greater than a preset threshold, S2 is executed, otherwise, the noise data in the cabin and the vibration data of the rear main reduction assembly are continuously acquired.

[0061] Here, the noise data in the cabin is collected by a microphone, which can be represented as a noise signal sequence X N , the vibration data of the rear main reduction assembly is collected by a vibration sensor installed on the rear main reduction assembly, which can be represented as a vibration load spectrum Y N , and the vibration load spectrum can determine the frequency and amplitude information of the noise source in the noise analysis process.

[0062] The above-mentioned sensor should have high sensitivity and high fidelity to ensure that the noise signal generated by the rear main reduction assembly can be accurately captured.

[0063] Specifically, the noise signal sequence X N and the vibration load spectrum Y N are subjected to vector subtraction with the calibration signal sequence provided by the host factory, to obtain the absolute value K of the signal difference, and when K is greater than a given noise threshold, S2 is executed.

[0064] In the embodiment, the noise threshold is 50dB.

[0065] Further, before S2 is executed, whether to execute noise control is prompted by a pop-up window displayed on the touch screen, if the user does not intervene, S2 is automatically executed.

[0066] S2, the noise data and the vibration data are processed by the trained recurrent neural network to obtain the noise prediction signal at the next moment.

[0067] Further, before performing S2, the noise data and the vibration data are subjected to digital signal processing by the FPGA chip. Specifically, first, the XY N The processing is sequentially performed to convert the signal from a high point number to a multiple low point number, and output a noise frequency domain signal sequence XY T .

[0068] It should be noted that the entire Fourier transform can be composed of base-2 and base-4 Fourier transforms, the 2k FFT (Fast Fourier Transform) can be realized by 5 base-4 and 1 base-2 transforms, the 4k FFT transform can be realized by 6 base-4 transforms, and the 8k FFT can be realized by 6 base-4 and 1 base-2 transforms, and the butterfly operation unit is a base-2 / 4 module.

[0069] In the embodiment, the RAM of the FPGA chip is used to store input data, intermediate results in the operation process, and data after the operation is completed, and the ROM is used to store a rotation factor table; the algorithm and the database can be upgraded through the NB-LOT module integrated in the FPGA chip.

[0070] The control module of the FPGA chip generates control timing and address signals through state machine logic to coordinate the operation process of each submodule in the FPGA, and specifically includes:

[0071] 1. Timing control: define the phased clock beats of subsequent FFT, RNN prediction, and fuzzy inference.

[0072] 2. Address management: dynamically allocate read-write addresses of the RAM according to the operation stage (such as storing original data addresses 0x0000-0x0FFF, intermediate result addresses 0x1000-0x1FFF), and trigger the reading of the rotation factor table in the ROM.

[0073] 3. Result output: write the finally generated reverse sound wave control instruction into a specified RAM area for calling by a loudspeaker driving module.

[0074] The FPGA chip has high parallel processing capability and low delay characteristics, and it can quickly process the input noise signal; the FPGA can easily realize various digital filter structures, such as FIR (Finite Impulse Response filter) and IIR (Infinite Impulse Response filter), so as to customize the frequency characteristics of the RDU. The digital signal noise reduction in the FPGA is a key technology to optimize the signal quality and improve the stability of the system. Through effective noise reduction, the signal processing accuracy can be improved, the error can be reduced, and the characteristics of the noise signal can be more accurately identified, so as to generate a more suitable anti-noise signal to offset the original noise, improve the stability of the noise reduction effect, and avoid the case that the noise reduction effect fluctuates greatly; and the FPGA can allocate resources and customize according to the specific signal processing requirements. In this application, the signals of multiple channels need to be processed (such as environmental noise collected by multiple microphones), and the resources of the FPGA, such as logic gates and storage units, can be flexibly allocated to meet the processing requirements of different channel signals and improve the overall noise reduction efficiency.

[0075] In this embodiment, the FPGA chip can be Xilinx Zynq-7000 series, and the logic unit is greater than or equal to 85k.

[0076] As an implementation manner, S2 is specifically: converting the noise frequency domain signal sequence XY T Inputting the recurrent neural network RNN for processing, converting the noise frequency domain signal sequence XY T According to the time expansion, the neuron structure is obtained, and the vector matrix XY K is output through the tanh activation function and the sigmoid activation function. K Inverse Fourier transform is performed on the vector matrix XY K , the signal is converted to the time domain again, and the final noise prediction result is output to provide accurate basic data for the future time period for subsequent noise suppression.

[0077] In addition, the feature values of the input data are captured through the processing of the recurrent neural network, so that the output result is more lightweight.

[0078] Specifically, in combination with Figure 4 , the weight directly from the input layer to the hidden layer is represented by U, which abstracts the original input as the input of the hidden layer; the weight W from the hidden layer to the hidden layer is responsible for scheduling memory and network memory control; the weight V from the hidden layer to the output layer is used to abstract the hidden layer twice and is used as the final output.

[0079] Figure 4 In S tis the "memory" at time t; f() represents a non-linear transformation function, such as an activation function like tanh, S t represents a neuron, which has two outputs; o t represents the output at time t (the output result after the neuron is activated), for example, if it is to predict the next signal, it may be the probability of the sigmoid (softmax) output belonging to each signal value, O t = softmax(VS t ), Softmax represents a normalized exponential function, each element is between 0 and 1, and the sum is 1.

[0080] Further, when training the recurrent neural network, first, the recurrent neural network is selected, the number of layers of the network, the number of neurons in each layer and other hyperparameters are determined, and the model structure is constructed; then, according to the previously collected and preprocessed data, the minimum mean square error of the time-frequency features of the noise-free reference signal and the actual working condition vibration signal is taken as the training target, and the recurrent neural network is trained to optimize the recurrent neural network parameters.

[0081] Further, in order to avoid the problem of gradient loss caused by too large gradient, the gradient of the memory gradient W is cut off by algorithm to prevent signal explosion.

[0082] In this embodiment, the algorithm can be value truncation and norm truncation, for example, when training the recurrent neural network, if the gradient exceeds a certain threshold, it is limited within the threshold to avoid excessive parameter update.

[0083] S3, obtaining real-time working condition data of the automobile, constructing a real-time feature vector representing noise and working condition according to the real-time working condition data and the noise prediction signal, and generating a noise control strategy by fuzzy reasoning using a fuzzy rule base according to the real-time feature vector, so as to suppress noise by corresponding noise counter sound waves. Specifically, it includes:

[0084] S301, time domain feature extraction is performed on the noise prediction signal to obtain the corresponding root mean square value; band pass filtering is performed on the noise prediction signal, and time-frequency conversion is performed on the band pass filtered noise prediction signal by FFT to obtain noise prediction frequency domain signal and perform feature extraction to determine the meshing frequency f m and the sideband amplitude A side ; the real-time working condition data is normalized to obtain the standard speed and the standard torque; the root mean square value, the meshing frequency, the sideband amplitude, the standard speed and the standard torque are combined to generate a real-time feature vector

[0085] Wherein, the meshing frequency is expressed as:

[0086] f m= f1·z1= f2·z2= (n1 / 60)·z1= (n2 / 60)·z2;

[0087] In the formula, f1 represents the rotation frequency of the driving gear, f2 represents the rotation frequency of the driven gear, z1 represents the number of teeth of the driving gear, z2 represents the number of teeth of the driven gear, n1 represents the rotation speed of the driving gear, and n2 represents the rotation speed of the driven gear.

[0088] The root mean square value is represented as:

[0089]

[0090] The band-pass filtering is represented as:

[0091] x(t) = Butterworth(x(t), f low ,f high );

[0092] The band-pass filtering retains the characteristic frequency band (such as 500 kHz-2 kHz) of the meshing of the main and driven gears.

[0093] The standard rotation speed is represented as:

[0094]

[0095] The standard torque is represented as:

[0096]

[0097] Here, in the spectrum diagram, the sideband is a frequency component that is centered on the meshing frequency and symmetrically distributed on both sides, and the sideband amplitude can be directly read by existing software. In addition, the relative ratio of the sideband amplitude to the meshing frequency amplitude can be used as an index to judge the severity of the gear fault. As the gear fault develops, the sideband amplitude may increase relative to the meshing frequency amplitude.

[0098] S302, according to the membership degree u ij of each rule in the fuzzy rule base corresponding to the real-time feature vector, the real-time control parameter is weighted and calculated, the corresponding phase compensation and damping gain are selected in the fuzzy rule base according to the real-time control parameter, as the corresponding noise control strategy, input to the full-cabin loudspeaker and headrest loudspeaker to generate noise reverse sound waves, the phase of the noise reverse sound waves is opposite to the noise prediction signal, and the amplitude of the noise reverse sound waves is dynamically adjusted according to the reasoning result.

[0099] In noise control, the phase of the noise anti-wave is opposite to the noise prediction signal, which is the basis for achieving destructive interference. The degree of phase compensation will indirectly affect the amplitude adjustment. For example, if the phase compensation is not accurate, in order to achieve better noise cancellation effect, the amplitude may need to be adjusted. Assuming that there is an ideal cancellation point, when the phase deviates from this ideal value, the amplitude of the anti-wave may need to be increased or decreased to compensate for the lack of cancellation effect.

[0100] wherein the real-time control parameter is represented as:

[0101]

[0102] Further, before performing S3, further comprising: constructing the fuzzy rule base by using an improved fuzzy C-means clustering method, and dynamically updating the fuzzy rule base by using real-time data; the specific process is as follows:

[0103] Step 1, obtaining original vibration signals, original noise signals and working condition parameters and performing pretreatment and feature extraction to obtain a plurality of characteristic vector sequences representing noise and working conditions

[0104]

[0105] The data processing method here is the same as S301, which will not be repeated here.

[0106] Step 2, using the fuzzy C-means method to cluster the characteristic vector sequence F N and divide the fuzzy rule node R j .

[0107] wherein the fuzzy rule node R j The corresponding membership function is represented as:

[0108]

[0109] In the formula, c j represents the clustering center, m represents the fuzzy index, and C represents the number of rules.

[0110] Further, in each fuzzy inference, the clustering center c j and the width σ j of the membership function are updated according to the real-time characteristic vector.

[0111] By adjusting the mathematical representation of the fuzzy rule base in real time, the robustness and accuracy of noise suppression are improved; the essence is to update the fuzzy rule base through online self-organizing learning to meet the high-precision noise reduction under complex working conditions.

[0112] For example, the updated clustering center is represented as:

[0113]

[0114] The updated width is expressed as:

[0115]

[0116] In the formula, η represents a learning rate, and controls the convergence speed.

[0117] Based on this, the clustering center and the width can be continuously optimized according to new data, so that the fuzzy rule base is adaptively adjusted according to the change of the vehicle state, and is adapted to different working conditions.

[0118] Step 3: Each clustering center corresponds to a fuzzy rule, which is stored in the fuzzy rule base and is expressed as:

[0119] IFF belongs to R j , THEN control parameter is a phase compensation, K j is a damping gain.

[0120] In summary, the way of generating the noise suppression strategy through fuzzy reasoning in this step is adaptive, and through self-organizing learning, it can cope with complex working conditions (such as sudden speed change and load change). According to the vibration, noise and working condition data collected in real time, the control strategy is dynamically adjusted, and the noise of the rear main reduction assembly is significantly reduced; the fuzzy reasoning calculation is low, which meets the real-time demand of noise control during the driving process of the four-wheel drive vehicle; and the fuzzy rule base is stored in the form of "IF-THEN", which is convenient for engineers to debug and optimize.

[0121] Further, in some embodiments, a dial button can be additionally installed on the side of the seat of the vehicle, and the user can divide the noise reduction depth into three grades according to the noise reduction demand of the user through the additionally installed dial button and the program OTA (Over The Air air download technology), and the noise reduction depth is 40%, 70% and 100% of the maximum noise reduction depth, respectively.

[0122] Experiments prove that after the automobile rear main reduction assembly noise suppression method described in this embodiment is processed, the 1 / 3 octave sound pressure level attenuation is ≥15dB(A), the speech intelligibility index is improved by ≥0.25, the 20-2000Hz frequency band transmission loss is ≥18dB, and the phase delay is ≤0.15ms.

[0123] Embodiment two

[0124] In combination Figure 5 , the embodiment discloses an automobile rear main reduction assembly noise suppression system, comprising:

[0125] The comparison module is configured to: acquire noise data in the cabin and vibration data of the rear main reduction assembly, and calculate the difference between the noise data and the vibration data and the vehicle calibration data.

[0126] The noise suppression module is configured to, if the difference is greater than a preset threshold, process the noise data and the vibration data by using the trained recurrent neural network to obtain a noise prediction signal at the next moment.

[0127] Real-time working condition data of the automobile is obtained, and a real-time feature vector representing noise and working condition is constructed according to the real-time working condition data and the noise prediction signal; fuzzy inference is performed on the real-time feature vector by using a fuzzy rule base to generate a noise control strategy, so that noise suppression is performed on the corresponding noise reverse sound wave.

[0128] The fuzzy rule base updates membership function parameters and control strategies online through self-organizing learning, and dynamically learns the mapping relationship between features and noise control by using a fuzzy clustering method; the noise reverse sound wave is generated by a full-cabin loudspeaker and a headrest loudspeaker, the phase of the noise reverse sound wave is opposite to that of the noise prediction signal, and the amplitude of the noise reverse sound wave is dynamically adjusted according to the noise control strategy.

[0129] It should be noted that the above comparison module and noise suppression module correspond to the steps in Embodiment One, and the above modules and the corresponding steps have the same examples and application scenarios, but are not limited to the content disclosed in Embodiment One. It should be noted that the above modules as part of the system can be executed in a computer system such as a set of computer executable instructions.

[0130] Embodiment Three

[0131] Embodiment Three of the present application provides an electronic device, which includes a memory and a processor, and computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, the steps of the above automobile rear main reduction assembly noise suppression method are completed.

[0132] Embodiment Four

[0133] Embodiment Four of the present application provides a computer readable storage medium for storing computer instructions, which are executed by a processor to complete the steps of the above automobile rear main reduction assembly noise suppression method.

[0134] Embodiment Five

[0135] Embodiment Five of the present application provides a computer program product, which includes a computer program / instruction, which is executed by a processor to implement the steps of the above automobile rear main reduction assembly noise suppression method.

[0136] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in the flowchart block or blocks

[0137] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in the flowchart block or blocks

[0138] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in the flowchart block or blocks

[0139] The above description of the various embodiments can have emphasized certain aspects of the various embodiments, which description can have been incomplete in not describing some portions of the various embodiments. Descriptions of the various embodiments in the above can be supplemented with relevant descriptions from other embodiments.

[0140] The above only describes preferred embodiments of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method of suppressing noise in an automotive rear mainshaft assembly, the method comprising: The method comprises the following steps: acquiring noise data in the cabin and vibration data of the rear main reduction assembly, and calculating the difference between the noise data and the vibration data and vehicle calibration data; if the difference is greater than a preset threshold, processing the noise data and the vibration data through a trained recurrent neural network to obtain a noise prediction signal at the next moment; the recurrent neural network is deployed on an FPGA chip to accelerate convolution operation of a weight matrix; acquiring real-time working condition data of the vehicle, constructing a real-time feature vector representing noise and working conditions according to the real-time working condition data and the noise prediction signal, and generating a noise control strategy by fuzzy reasoning using a fuzzy rule base according to the real-time feature vector, so as to suppress noise by corresponding noise counter sound waves; constructing a real-time feature vector representing noise and working conditions according to real-time working condition data and a noise prediction signal comprises: performing time domain feature extraction on the noise prediction signal to obtain a corresponding root mean square value; sequentially performing band-pass filtering and time-frequency conversion on the noise prediction signal to obtain a noise prediction frequency domain signal and perform feature extraction to determine meshing frequency and sideband amplitude; performing normalization processing on the real-time working condition data to obtain a standard speed and a standard torque; combining the root mean square value, the meshing frequency, the sideband amplitude, the standard speed and the standard torque to generate the real-time feature vector; when generating the noise control strategy, the real-time working condition data of the vehicle and the noise prediction signal at the next moment are combined to perform adaptive updating of the fuzzy rule base; wherein the fuzzy rule base is constructed and dynamically updated using an improved fuzzy C-means clustering method; the noise counter sound waves are generated by an audio system comprising no less than two sound field generating units, the phase of the noise counter sound waves is opposite to that of the noise prediction signal, and the amplitude of the noise counter sound waves is dynamically adjusted according to the noise control strategy.

2. The automotive rear mainshaft assembly noise suppression method of claim 1, wherein, Before processing the noise data and the vibration data through the trained recurrent neural network, the method further comprises: processing the noise data and the vibration data through the FPGA chip to generate corresponding noise frequency domain signals.

3. The automotive rear mainshaft assembly noise suppression method of claim 1, wherein, The recurrent neural network is trained to minimize the mean square error of the time-frequency features of the noise reference signal and the actual working condition vibration signal.

4. The automotive rear mainshaft assembly noise suppression method of claim 1, wherein, The fuzzy reasoning using the fuzzy rule base to generate the noise control strategy according to the real-time feature vector comprises: calculating a real-time control parameter by weighting the membership degrees of each rule in the fuzzy rule base corresponding to the real-time feature vector, and outputting the corresponding noise control strategy.

5. The automotive rear mainshaft assembly noise suppression method of claim 1, wherein, The fuzzy rule base is constructed and dynamically updated using an improved fuzzy C-means clustering method, which comprises: acquiring vibration signals, noise signals and working condition parameters, and performing preprocessing and feature extraction to obtain feature vectors; clustering the feature vectors by the fuzzy C-means clustering method, dividing fuzzy rule nodes, and storing the fuzzy rule nodes and corresponding fuzzy rules in the fuzzy rule base; wherein the clustering centers and widths of the membership functions are dynamically adjusted according to real-time feature vectors.

6. An automotive rear mainshaft assembly noise suppression system, characterized by, The method comprises the following steps: a comparison module configured to acquire noise data in the cabin and vibration data of the rear main reduction assembly, and calculate the difference between the noise data and the vibration data and vehicle calibration data; The noise suppression module is configured to: if the difference is greater than a preset threshold, process the noise data and the vibration data by using the trained recurrent neural network to obtain a noise prediction signal at the next moment; The recurrent neural network is deployed on an FPGA chip to accelerate convolution operation of a weight matrix. Real-time working condition data of the automobile is obtained, a real-time feature vector representing noise and working condition is constructed according to the real-time working condition data and the noise prediction signal, fuzzy reasoning is performed on the real-time feature vector by using a fuzzy rule base to generate a noise control strategy, and noise is suppressed by using a corresponding noise counter sound wave. The real-time feature vector representing noise and working condition is constructed according to the real-time working condition data and the noise prediction signal, including: time-domain feature extraction is performed on the noise prediction signal to obtain a corresponding root mean square value; band-pass filtering and time-frequency conversion are sequentially performed on the noise prediction signal to obtain a noise prediction frequency domain signal and perform feature extraction to determine an engagement frequency and a sideband amplitude value; the real-time working condition data is normalized to obtain a standard speed and a standard torque; and the root mean square value, the engagement frequency, the sideband amplitude value, the standard speed and the standard torque are combined to generate the real-time feature vector. When the noise control strategy is generated, adaptive updating of the fuzzy rule base is performed in combination with the real-time working condition data of the automobile and the noise prediction signal at the next moment. The fuzzy rule base is constructed and dynamically updated by using an improved fuzzy C-means clustering method; the noise counter sound wave is generated by using a sound system including not less than two sound field generating units, a phase of the noise counter sound wave is opposite to that of the noise prediction signal, and an amplitude of the noise counter sound wave is dynamically adjusted according to the noise control strategy.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the electronic device to perform the method of any one of claims 1-6. The processor executes the computer program to implement the steps of the automobile rear main reduction assembly noise suppression method according to any one of claims 1-5.

8. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, The computer program / instruction is executed by the processor to implement the steps of the automobile rear main reduction assembly noise suppression method according to any one of claims 1-5.

9. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instruction is executed by the processor to implement the steps of the automobile rear main reduction assembly noise suppression method according to any one of claims 1-5.

Images