Dynamic broadband vibration noise active prediction control method, system and device

Abstract

The invention relates to the field of vibration noise processing, and provides a dynamic broadband vibration noise active prediction control method, system and device, and the method comprises the steps: selecting and matching a target sub-band decomposition network corresponding to a sub-band signal of target vibration noise from a target sub-band decomposition network library; selecting and matching a target multi-step prediction model corresponding to the target sub-band decomposition network; predicting the sub-band signal of the target vibration noise, and determining predicted vibration noise; and performing active prediction control on the target vibration noise and the predicted vibration noise. The method and the device are used for solving the defects that in the prior art, the vibration and noise reduction technical effect on broadband vibration noise is poor, and effective vibration and noise reduction control on the broadband vibration noise cannot be carried out, the scheme of the invention can rapidly and accurately predict future data of the broadband vibration noise, real-time active control on the broadband vibration noise under non-causal time delay is realized, and the method and the device are suitable for large-scale popularization and application. A solution is provided for an existing complex method under the condition that hardware computing power is limited.

Description

Technical Field

[0001] The present invention relates to the technical field of vibration and noise processing, and in particular to a method, system and device for actively predicting and controlling dynamic broadband vibration and noise. Background Art

[0002] Vibration and noise sources are widespread and almost exist in all mechanical equipment and structures. Excessive vibration can cause excessive wear of machine components and reduce their service life. Therefore, controlling and reducing unnecessary vibration and noise is crucial for protecting human health, maintaining the ecological environment, and ensuring the reliability and efficiency of equipment.

[0003] Due to the complex spectral characteristics of vibration and noise, traditional vibration and noise reduction technologies such as passive sound insulation / vibration or simple feedback control systems are often difficult to effectively handle. There are also some active vibration and noise reduction methods and systems in the related technologies. These technologies basically generate vibration and noise reduction signals by the control system after the vibration and noise source generates vibration and noise. However, due to the fact that the actuator requires a certain execution time and generating the vibration and noise reduction signal requires a certain time, the time for the vibration and noise reduction signal to propagate to the vibration and noise reduction point is often longer than the time for the vibration and noise signal to propagate from the vibration and noise source to the noise reduction point, resulting in non-causal time delay and poor vibration and noise reduction effect, and unable to effectively control the vibration and noise reduction of vibration and noise. Summary of the Invention

[0004] The present invention provides a method, system and device for actively predicting and controlling dynamic broadband vibration and noise, so as to solve the defect that the vibration and noise reduction technology effect in the related technology is limited by causal time delay and cannot effectively reduce the noise of dynamic broadband vibration and noise under non-causal time delay. In the solution of the present application, the noise type corresponding to the target vibration and noise can be matched from the pre-constructed broadband vibration and noise library, the sub-band decomposition network corresponding to the vibration and noise can be matched from the pre-constructed sub-band decomposition network library, and the prediction model corresponding to the sub-band decomposition network can be matched from the pre-constructed multi-step prediction model library. When actively controlling the vibration and noise, the future data of the dynamic broadband noise can be predicted more quickly and accurately, and then its active prediction control can be realized.

[0005] The present invention provides a method for actively predicting and controlling dynamic broadband vibration and noise, including:

[0006] After adaptively dividing the target vibration and noise to be controlled into sub-bands, select and match the target sub-band decomposition network corresponding to the sub-band signal of the target vibration and noise from the pre-constructed target sub-band decomposition network library;

[0007] Select and match the target multi-step prediction model corresponding to the target sub-band decomposition network from the pre-constructed multi-step prediction model library;

[0008] Based on the target multi-step prediction model, predict the sub-band signals of the target vibration noise to determine the predicted vibration noise;

[0009] Based on the target vibration noise and the predicted vibration noise, generate a secondary vibration noise through an active control method to achieve active control of dynamic broadband vibration noise under non-causal time delay.

[0010] According to the active prediction control method for dynamic broadband vibration noise provided by the present invention, before adaptively dividing the target vibration noise into sub-bands, it further includes:

[0011] Analyze the characteristics of the target vibration noise to determine the category of the target vibration noise;

[0012] Based on the category of the target vibration noise, select a matching corresponding type from a pre-constructed broadband vibration noise library, which is constructed by analyzing the characteristics of various broadband vibration noises in units of the category of the broadband vibration noise;

[0013] According to the matching type, select a target sub-band decomposition network library corresponding to the category of the target vibration noise from a number of pre-constructed sub-band decomposition network libraries.

[0014] According to the active prediction control method for dynamic broadband vibration noise provided by the present invention, the selection of a target sub-band decomposition network that matches the sub-band signal of the target vibration noise from the pre-constructed target sub-band decomposition network library includes:

[0015] According to the spectral characteristics of the target broadband vibration noise, with the energy at the spectral maximum point as the central value, achieve adaptive sub-band division, and based on the divided frequency bands of the sub-band signals, select a target sub-band decomposition network that matches the sub-band signal of the target vibration noise from the target sub-band decomposition network library;

[0016] Based on the maximum frequency band overlap as the criterion, perform adaptive division of the frequency bands and selection and matching of the sub-bands in the sub-band decomposition network library.

[0017] According to the active prediction control method for dynamic broadband vibration noise provided by the present invention, the construction process of the sub-band decomposition network library includes:

[0018] Convert the vibration noises of different categories in the vibration noise library into frequency domain signals through Fourier transform;

[0019] Decompose the frequency domain signals into sub-band signals of equal value segments;

[0020] After performing inverse Fourier transform on the sub-band signals, convert them into time domain signals;

[0021] The sub-band decomposition networks corresponding to different categories of broadband vibration noise are stored separately to construct a number of sub-band decomposition network libraries, and the vibration noise corresponding to each sub-band decomposition network library is of the same category.

[0022] According to the active prediction control method for dynamic broadband vibration noise provided by the present invention, the construction process of the multi-step prediction model library includes:

[0023] Using the constructed sub-band decomposition network to decompose different categories of broadband vibration noise into multiple sub-band signals, constructing different multi-step prediction models based on a neural network, selecting the prediction model with the smallest error when meeting the prediction steps, and storing it;

[0024] Performing multi-step prediction on the sub-band signals after decomposing different categories of broadband vibration noise to construct a multi-step prediction model library.

[0025] According to the active prediction control method for dynamic broadband vibration noise provided by the present invention, it further includes:

[0026] If there is no sub-band decomposition network in the target sub-band decomposition network library with the same frequency band as the sub-band signal, the target sub-band decomposition network library is supplemented based on the sub-band signal.

[0027] According to the active prediction control method for dynamic broadband vibration noise provided by the present invention, it further includes:

[0028] If the prediction accuracy of the selected and matched multi-step prediction model does not meet the requirements, the multi-step prediction model library is supplemented through model training.

[0029] According to the active prediction control method for dynamic broadband vibration noise provided by the present invention, the multi-step prediction models in the multi-step prediction model library are trained using the sub-band decomposition network library as a training set.

[0030] The present invention also provides a system for an active prediction control method for dynamic broadband vibration noise, including:

[0031] A decomposition network matching module, configured to, after adaptively dividing the target vibration noise to be controlled into sub-bands, select and match a target sub-band decomposition network corresponding to the sub-band signal of the target vibration noise from a pre-constructed target sub-band decomposition network library;

[0032] A prediction model matching module, configured to select and match a target multi-step prediction model corresponding to the target sub-band decomposition network from a pre-constructed multi-step prediction model library;

[0033] A vibration noise prediction module, configured to predict the sub-band signal of the target vibration noise based on the target multi-step prediction model to determine the predicted vibration noise;

[0034] An active predictive control module, which is used to generate secondary vibration noise through an active control strategy based on the target vibration noise and the predicted vibration noise, so as to achieve active control of dynamic broadband vibration noise in a non-causal time-delay environment.

[0035] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements any one of the above-mentioned dynamic broadband vibration noise active predictive control methods.

[0036] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements any one of the above-mentioned dynamic broadband vibration noise active predictive control methods.

[0037] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements any one of the above-mentioned dynamic broadband vibration noise active predictive control methods.

[0038] In the dynamic broadband vibration noise active predictive control method provided by the present invention, a sub-band decomposition network corresponding to the vibration noise can be matched from a pre-constructed sub-band decomposition network library, and a prediction model corresponding to the sub-band decomposition network can be matched from a pre-constructed multi-step prediction model library. When controlling vibration noise, it is faster, and on the premise of ensuring real-time performance, the adaptability of the active noise reduction system in a non-causal time-delay environment is improved, the influence of time-varying parameters is alleviated, the performance of active noise reduction of vibration noise is optimized, and the calculation efficiency of active control of vibration noise can also be improved through fast and accurate matching. Description of the Drawings

[0039] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0040] Figure 1 is one of the flow diagrams of the dynamic broadband vibration noise active predictive control method provided by the embodiments of the present invention;

[0041] Figure 2 is the second of the flow diagrams of the dynamic broadband vibration noise active predictive control method provided by the embodiments of the present invention;

[0042] Figure 3 is the structural diagram of the dynamic broadband vibration noise active predictive control method system provided by the embodiments of the present invention;

[0043] Figure 4 It is a schematic structural diagram of a dynamic broadband vibration noise active prediction and control device provided by an embodiment of the present invention. Specific embodiments

[0044] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0045] Figure 1 It is one of the schematic flowcharts of a dynamic broadband vibration noise active prediction and control method provided by an embodiment of the present invention.

[0046] Figure 2 It is another schematic flowchart of a dynamic broadband vibration noise active prediction and control method provided by an embodiment of the present invention.

[0047] As Figure 1 and Figure 2 shown, this embodiment provides a dynamic broadband vibration noise active prediction and control method, including:

[0048] Step 101, after adaptively dividing the target vibration noise into sub-bands, select a target sub-band decomposition network that matches the sub-band signal of the target vibration noise from a pre-constructed target sub-band decomposition network library;

[0049] In implementation, sensors can be set at the vibration noise source to collect the vibration noise generated by the vibration noise source in real time. This vibration noise can be any type of broadband vibration noise, such as gas dynamic vibration noise, mechanical vibration noise, and electromagnetic vibration noise, etc.

[0050] Sub-band decomposition is a technology that decomposes a signal into multiple sub-bands so that each sub-band contains different frequency components of the signal. In practical applications, the obtained target vibration noise can be Fourier-transformed to obtain the frequency-domain representation of the vibration noise, and then the frequency-domain representation is divided into several frequency bands. Each frequency band represents a certain range of frequencies. After that, the frequency-domain representation of each frequency band is inverse Fourier-transformed, and the sub-band signal can be obtained. In this way, the process of decomposing the target vibration noise into sub-bands is completed.

[0051] Step 102, select a target multi-step prediction model that matches the target sub-band decomposition network from a pre-constructed multi-step prediction model library;

[0052] Step 103: Based on the target multi-step prediction model, predict the sub-band signals of the target vibration noise to determine the predicted vibration noise;

[0053] Step 104: Based on the target vibration noise and the predicted vibration noise, generate a secondary vibration noise through an active control method to achieve active control of dynamic broadband vibration noise under non-causal time delay.

[0054] In implementation, the gain of the predicted vibration noise can be regulated, which can improve the control accuracy. Moreover, independent controllers can be designed for the target vibration noise and the predicted vibration noise respectively, which can adjust the output more precisely and improve the overall performance.

[0055] In practical applications, the vibration noise prediction model can perform multi-step prediction on the target vibration noise. Specifically, it can predict the target vibration noise for a relatively long period of time, and then data processing can be performed on the predicted vibration noise. In this way, it is equivalent to making preparations for the vibration noise in advance. Compared with the existing method of generating an anti-noise signal only after the vibration noise source generates a vibration noise signal, predicting the vibration noise can facilitate the subsequent generation of an anti-noise signal in advance, that is, the secondary vibration noise in Step 104. After generating the secondary vibration noise based on the predicted vibration noise, noise reduction processing can be performed on it in a timely manner when the predicted vibration noise reaches the noise reduction point.

[0056] In this step, the primary vibration noise is obtained by combining the target vibration noise and the predicted vibration noise. The primary vibration noise can be a complete audio signal obtained by combining the target vibration noise and the predicted vibration noise.

[0057] In implementation, the amplitude of the secondary vibration noise can be the same as that of the primary vibration noise. In this way, new vibration noise pollution can be avoided after noise reduction control of the primary vibration noise.

[0058] In practical applications, a secondary vibration noise can be generated through a parallel active vibration noise control strategy.

[0059] In practical applications, the time for each band of the primary vibration noise to propagate to the noise reduction point can be calculated based on the distance between the vibration noise source and the noise reduction point and the speed of sound. By controlling the transmission time of the secondary vibration noise, the corresponding band of the secondary vibration noise can be made to propagate to the position of the noise reduction point at the same time, so that the corresponding bands of the primary vibration noise and the secondary vibration noise are superimposed at the same position to form destructive interference, achieving the purpose of eliminating vibration noise.

[0060] In the dynamic broadband vibration noise active prediction control method provided in this embodiment, a sub-band decomposition network corresponding to the vibration noise can be matched from a pre-constructed sub-band decomposition network library, and a prediction model corresponding to the sub-band decomposition network can be matched from a pre-constructed multi-step prediction model library. When controlling vibration noise, it is faster, and on the premise of ensuring real-time performance, the adaptability of the active noise reduction system in a non-causal time-delay environment is improved, the influence of time-varying parameters is alleviated, the performance of active vibration noise reduction is optimized, and accurate matching can also improve the vibration noise control effect.

[0061] In an exemplary embodiment, before performing adaptive sub-band division on the target vibration noise, it further includes:

[0062] Analyze the vibration noise characteristics of the target vibration noise to determine the category of the target vibration noise;

[0063] Based on the category of the target vibration noise, select and match the corresponding type from a pre-constructed broadband vibration noise library. The broadband vibration noise library is constructed by analyzing the characteristics of various broadband vibration noises in units of the categories of broadband vibration noises;

[0064] According to the matching type, select a target sub-band decomposition network library corresponding to the category of the target vibration noise from several pre-constructed sub-band decomposition network libraries.

[0065] In practical applications, since there are significant differences in the frequencies, amplitudes, etc. of different categories of vibration noises, a sub-band decomposition network library can be pre-constructed for each type of vibration noise. When in use, a suitable target sub-band decomposition network library can be selected based on the category of the target vibration noise to be controlled. The signals stored in the target sub-band decomposition network library have a high similarity to the target vibration noise. In this way, the computational amount during vibration noise prediction can be effectively reduced, and the prediction efficiency can be improved.

[0066] In implementation, what can be stored in each sub-band decomposition network library are several sub-band signals of different frequency bands of the corresponding type of vibration noise.

[0067] In an exemplary embodiment, the step of selecting and matching a target sub-band decomposition network corresponding to the sub-band signal of the target vibration noise from the pre-constructed target sub-band decomposition network library includes:

[0068] According to the spectral characteristics of the target broadband vibration noise, with the energy of the spectral maximum point as the central value, perform adaptive sub-band division, and based on the divided frequency bands of the sub-band signal, select and match a target sub-band decomposition network corresponding to the sub-band signal of the target vibration noise from the target sub-band decomposition network library;

[0069] Based on the maximum frequency band overlap, perform adaptive division of the frequency band and select and match the sub-bands in the sub-band decomposition network library.

[0070] In an exemplary embodiment, the construction process of the sub-band decomposition network library includes:

[0071] Convert different categories of vibration noises in the vibration noise library into frequency-domain signals through Fourier transform;

[0072] Decompose the frequency-domain signals into sub-band signals of equal value segments;

[0073] After performing inverse Fourier transform on the sub-band signals, convert them into time-domain signals;

[0074] Store the sub-band decomposition networks corresponding to different categories of broadband vibration noises respectively, and construct several sub-band decomposition network libraries. Each sub-band decomposition network library corresponds to the same category of vibration noise.

[0075] In an exemplary embodiment, the construction process of the multi-step prediction model library includes:

[0076] Use the constructed sub-band decomposition network to decompose different categories of broadband vibration noises into multiple sub-band signals, construct different multi-step prediction models based on neural networks, select the prediction model with the smallest error when meeting the prediction steps, and store it;

[0077] Perform multi-step prediction on the sub-band signals decomposed from different categories of broadband vibration noises, and construct a multi-step prediction model library.

[0078] In an exemplary embodiment, it further includes:

[0079] If there is no sub-band decomposition network in the target sub-band decomposition network library with the same frequency band as the sub-band signal, supplement the target sub-band decomposition network library based on the sub-band signal.

[0080] In an exemplary embodiment, it further includes:

[0081] If the prediction accuracy of the selected and matched multi-step prediction model cannot meet the requirements, supplement the multi-step prediction model library through model training.

[0082] In an exemplary embodiment, the multi-step prediction models in the multi-step prediction model library are trained using the sub-band decomposition network library as the training set.

[0083] In implementation, for each different sub-band decomposition network in the sub-band decomposition network library, that is, for each sub-band signal, a corresponding vibration noise prediction model can be established and trained specifically. The advantage of using each sub-band decomposition network in the sub-band decomposition network library as the training set is that it is convenient to directly determine a targeted vibration noise prediction model based on the target sub-band decomposition network. Specifically, in the solution of this application, after obtaining the target vibration noise to be controlled, the characteristics of the target vibration noise can be analyzed to determine the type of vibration noise to which the target vibration noise belongs. Then, based on the type of vibration noise, the target sub-band decomposition network library corresponding to the target vibration noise can be determined. After that, the target vibration noise is decomposed into several sub-band signals, and based on these sub-band signals, the target sub-band decomposition network with a relatively high degree of overlap is selected from the target sub-band decomposition network library. Since the prediction model is trained based on the target sub-band decomposition network, after selecting the target sub-band decomposition network, the corresponding prediction model can be further determined. Based on this prediction model, the sub-band signals corresponding to the target vibration noise can be predicted more accurately to obtain the predicted vibration noise corresponding to the target vibration noise.

[0084] In an exemplary embodiment, it further includes:

[0085] If the adaptive sub-band division is centered on the spectral maximum value for sub-band division, select the sub-band decomposition network with more overlapping parts in the matching frequency band.

[0086] In practical applications, if the frequency bands of two sub-band signals are the same, it can indicate that the similarity of the two sub-band signals is relatively high. Therefore, in implementation, if the frequency band of the target vibration signal overlaps with the network in the library, sub-bands with a relatively high degree of overlap are selected for subsequent prediction.

[0087] The dynamic broadband vibration noise active prediction control method provided by this application at least further has the following functions:

[0088] First: It can effectively reduce noise in a non-causal time-delay environment for vibration noise. By decomposing and then predicting the vibration noise in sub-bands, the influence of time-varying parameters on the active noise reduction system is reduced. At the same time, predicting first and then controlling also significantly improves the noise reduction effect and enhances the flexibility and adaptability of the system.

[0089] Second: By establishing an offline sub-band decomposition network database and a multi-step prediction model database, fast response of online prediction is achieved. During online prediction, only the characteristics of the target vibration noise need to be analyzed and matched with the vibration noise in the library to achieve fast multi-step prediction of the vibration noise, improving the accuracy and efficiency of active noise reduction.

[0090] The dynamic broadband vibration and noise active prediction control system provided by the present invention will be described below. The dynamic broadband vibration and noise active prediction control system described below can be referred to in correspondence with the dynamic broadband vibration and noise active prediction control method described above.

[0091] Figure 3 It is a schematic structural diagram of the dynamic broadband vibration and noise active prediction control system provided by an embodiment of the present invention.

[0092] As Figure 3 shown, the dynamic broadband vibration and noise active prediction control system provided in this embodiment includes:

[0093] A decomposition network matching module 301, which is used to perform adaptive sub-band division on the target vibration and noise to be controlled, and then select and match a target sub-band decomposition network corresponding to the sub-band signal of the target vibration and noise from a pre-constructed target sub-band decomposition network library;

[0094] A prediction model matching module 302, which is used to select and match a target multi-step prediction model corresponding to the target sub-band decomposition network from a pre-constructed multi-step prediction model library;

[0095] A vibration and noise prediction module 303, which is used to perform multi-step prediction on the sub-band signal of the target vibration and noise based on the target multi-step prediction model to determine the predicted vibration and noise;

[0096] An active prediction control module 304, which is used to generate secondary vibration and noise through an active control method based on the target vibration and noise and the predicted vibration and noise, so as to achieve active control of dynamic broadband vibration and noise under non-causal time delay.

[0097] In an exemplary embodiment, it further includes a characteristic analysis module, and the characteristic analysis module is used for:

[0098] Performing vibration and noise characteristic analysis on the target vibration and noise to determine the category of the target vibration and noise;

[0099] Based on the category of the target vibration and noise, selecting and matching the corresponding type from a pre-constructed broadband vibration and noise library, and the broadband vibration and noise library is constructed by performing characteristic analysis on various broadband vibration and noises and taking the category to which the broadband vibration and noise belongs as a unit;

[0100] According to the matching type, selecting a target sub-band decomposition network library corresponding to the category of the target vibration and noise from several pre-constructed sub-band decomposition network libraries.

[0101] In an exemplary embodiment, the characteristic analysis module is further used for:

[0102] According to the spectral characteristics of the target broadband vibration noise, with the energy at the maximum point of the spectrum as the central value, adaptive sub-band division is realized, and based on the divided frequency bands of the sub-band signals, a target sub-band decomposition network corresponding to the sub-band signal of the target vibration noise is selected from the target sub-band decomposition network library;

[0103] Based on the maximum frequency band overlap as the criterion, adaptive division of frequency bands and selection and matching of sub-bands in the sub-band decomposition network library are carried out.

[0104] In an exemplary embodiment, the characteristic analysis module is further configured to: convert the vibration noise in the vibration noise library into a frequency-domain signal through Fourier transform;

[0105] Decompose the frequency-domain signal into sub-band signals of equal value segments;

[0106] After performing inverse Fourier transform on the sub-band signal, convert it into a time-domain signal;

[0107] The sub-band decomposition networks corresponding to different categories of broadband vibration noise are stored separately to construct several sub-band decomposition network libraries, and the vibration noise corresponding to each sub-band decomposition network library is of the same category.

[0108] In an exemplary embodiment, the characteristic analysis module is further configured to: select a sub-band decomposition network with the same frequency band as the sub-band signal from the target sub-band decomposition network library.

[0109] In an exemplary embodiment, the characteristic analysis module is further configured to: if there is no sub-band decomposition network with the same frequency band as the sub-band signal in the target sub-band decomposition network library, supplement the target sub-band decomposition network library based on the sub-band signal.

[0110] In an exemplary embodiment, the characteristic analysis module is further configured to: if there is no sub-band decomposition network with the same frequency band as the sub-band signal in the target sub-band decomposition network library, select a sub-band decomposition network with more overlapping frequency band parts.

[0111] The specific implementation method of the dynamic broadband vibration noise active prediction control system provided in this embodiment can be implemented with reference to the above embodiment, and will not be elaborated here.

[0112] Figure 4 It is a schematic structural diagram of the dynamic broadband vibration noise active prediction control device provided by an embodiment of the present invention.

[0113] As Figure 4 shown, this embodiment further provides a dynamic broadband vibration noise active prediction control device, including:

[0114] A sensor for collecting vibration noise signals;

[0115] A conditioning circuit module for filtering, amplifying, and A / D conversion;

[0116] A first processor responsible for analyzing the characteristics of the target vibration noise, selecting and matching sub-band signals, and multi-step prediction;

[0117] A memory for storing the established sub-band decomposition network library, multi-step prediction model library, and multi-step prediction data;

[0118] A second processor for gain regulation and active control of signals; A drive circuit for D / A, filtering, and power amplification;

[0119] An actuator for outputting secondary vibration noise to achieve the purpose of noise reduction.

[0120] In practical applications, as Figure 4 shown, the dynamic broadband vibration noise active prediction and control device provided in this embodiment can use dual-core control. The multi-step predictor and the active noise reduction controller are connected through a memory for data exchange. Among them, sensors (microphones, accelerometers, etc.) are responsible for collecting vibration noise signals and transmitting the collected signals to the multi-step predictor. The multi-step predictor decomposes and multi-step predicts the vibration noise by retrieving the pre-established sub-band decomposition network library and multi-step prediction model library. After the prediction is completed, the data is transmitted to the active prediction controller through the memory. The controller generates secondary vibration noise, which is then output through the actuator (speaker, shock absorber, etc.), thereby achieving the purpose of vibration and noise reduction.

[0121] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the dynamic broadband vibration noise active prediction and control method provided by the above-mentioned various methods. The method includes:

[0122] After adaptively dividing the target vibration noise to be controlled into sub-bands, select and match a target sub-band decomposition network corresponding to the sub-band signal of the target vibration noise from a pre-constructed target sub-band decomposition network library;

[0123] Select and match a target multi-step prediction model corresponding to the target sub-band decomposition network from a pre-constructed multi-step prediction model library;

[0124] Based on the target multi-step prediction model, predict the sub-band signal of the target vibration noise to determine the predicted vibration noise;

[0125] Generate secondary vibration noise based on the target vibration noise and the predicted vibration noise through an active control method to achieve active control of dynamic broadband vibration noise under non-causal time delay.

[0126] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the dynamic broadband vibration noise active prediction control method provided by the above-mentioned various methods. The method includes:

[0127] After adaptively dividing the target vibration noise to be controlled into subbands, select a target subband decomposition network that matches the subband signal of the target vibration noise from a pre-constructed target subband decomposition network library;

[0128] Select a target multi-step prediction model that matches the target subband decomposition network from a pre-constructed multi-step prediction model library;

[0129] Based on the target multi-step prediction model, predict the subband signal of the target vibration noise to determine the predicted vibration noise;

[0130] Generate a secondary vibration noise through an active control method based on the target vibration noise and the predicted vibration noise, so as to achieve active control of dynamic broadband vibration noise under non-causal time delay.

[0131] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0132] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course also by hardware. Based on this understanding, the above technical solutions, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiments.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A dynamic broadband vibration noise active prediction control method, characterized in that: include: Adaptively divide the target vibration noise to be controlled into sub-bands, and select a target sub-band decomposition network that matches the sub-band signal of the target vibration noise from a pre-built target sub-band decomposition network library; Selecting and matching a target multi-step prediction model corresponding to the target sub-band decomposition network from a pre-built multi-step prediction model library; Based on the target multi-step prediction model, predicting the subband signal of the target vibration noise to determine the predicted vibration noise; Based on the target vibration noise and the predicted vibration noise, secondary vibration noise is generated through an active control method to achieve active control of vibration noise under non-causal time delay.

2. The dynamic broadband vibration noise active prediction control method according to claim 1 is characterized in that: Before the target vibration noise is adaptively divided into sub-bands, the method further comprises: Performing a vibration noise characteristic analysis on the target vibration noise to determine the category of the target vibration noise; Based on the category of the target vibration noise, a matching corresponding type is selected from a pre-built broadband vibration noise library, wherein the broadband vibration noise library is constructed by analyzing the characteristics of a plurality of broadband vibration noises based on the category to which the broadband vibration noise belongs; According to the matching type, a target sub-band decomposition network library corresponding to the category of the target vibration noise is selected from a plurality of pre-constructed sub-band decomposition network libraries.

3. The dynamic broadband vibration noise active prediction control method according to claim 2 is characterized in that: The step of selecting a target subband decomposition network corresponding to the subband signal of the target vibration noise from a pre-built target subband decomposition network library comprises: According to the spectrum characteristics of the target broadband vibration noise, the energy of the spectrum maximum point is taken as the center value to realize adaptive sub-band division, and based on the divided frequency band of the sub-band signal, a target sub-band decomposition network corresponding to the sub-band signal of the target vibration noise is selected from the target sub-band decomposition network library; Based on the maximum frequency band overlap, the frequency bands are adaptively divided and matched with the subband selection in the subband decomposition network library.

4. The method for active prediction and control of dynamic broadband vibration noise according to claim 3, characterized in that: The construction process of the subband decomposition network library includes: Convert different types of vibration noise in the vibration noise library into frequency domain signals through Fourier transform; Decomposing the frequency domain signal into sub-band signals of equal value segments; Performing inverse Fourier transform on the subband signal to convert it into a time domain signal; The sub-band decomposition networks corresponding to different types of broadband vibration noise are stored respectively, and a plurality of sub-band decomposition network libraries are constructed.

5. The method for active prediction and control of dynamic broadband vibration noise according to claim 3, characterized in that: The construction process of the multi-step prediction model library includes: The constructed subband decomposition network is used to decompose different types of broadband vibration noise into multiple subband signals. Different multi-step prediction models are constructed based on the neural network. The prediction model with the smallest error when meeting the prediction step number is selected and stored. The sub-band signals after decomposition of different types of broadband vibration noise are predicted in multiple steps to build a multi-step prediction model library.

6. The method for active prediction and control of dynamic broadband vibration noise according to claim 5, characterized in that: Also includes: If there is no subband decomposition network having the same frequency band as the subband signal in the target subband decomposition network library, the target subband decomposition network library is supplemented based on the subband signal.

7. The method for active prediction and control of dynamic broadband vibration noise according to claim 5, characterized in that: Also includes: If the prediction accuracy of the selected matching multi-step prediction model cannot meet the requirements, the multi-step prediction model library is supplemented through model training.

8. The method for active prediction and control of dynamic broadband vibration noise according to claim 4, characterized in that: The multi-step prediction models in the multi-step prediction model library are trained using the sub-band decomposition network library as a training set.

9. Dynamic broadband vibration noise active prediction control system, characterized in that: include: A decomposition network matching module is used to select a target subband decomposition network that matches a subband signal of the target vibration noise from a pre-built target subband decomposition network library after adaptively dividing the target vibration noise to be controlled into subbands; A prediction model matching module, used to select and match a target multi-step prediction model corresponding to the target sub-band decomposition network from a pre-built multi-step prediction model library; A vibration noise prediction module, used to perform multi-step prediction on the sub-band signal of the target vibration noise based on the target multi-step prediction model to determine the predicted vibration noise; The active prediction control module is used to generate secondary vibration noise through an active control method based on the target vibration noise and the predicted vibration noise, so as to realize active control of broadband vibration noise under non-causal time delay.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the dynamic broadband vibration noise active prediction control method as described in any one of claims 1-8 is implemented.

Images