Active control method for on-line identification of secondary channel

Abstract

The invention relates to a secondary channel online identification active control method, and belongs to the technical field of vibration noise active control. In order to solve the problem that in the prior art, redundant vibration is introduced into active vibration, so that the control effect becomes poor, the invention provides a secondary channel online identification active control method. The method comprises the following steps: acquiring a reference signal and random noise; obtaining a first residual vibration signal; obtaining a second residual vibration signal; determining a weight coefficient of an active controller and an iteration step length of a modeling FIR low-pass filter; coupling of active control convergence and secondary channel identification of the vibration noise active control system is relieved, and real-time modeling is conducted on the secondary channel through a random noise generator, an identification signal needed by the secondary channel and an FIR low-pass filter; a power regulation factor is introduced to adjust the iteration step size and the random noise power in the active control link, coupling interference of active control and channel modeling is effectively relieved, and the control effect is good.

Description

Technical Field

[0001] The present invention relates to an active control method for online identification of a secondary channel, belonging to the technical field of active vibration and noise control. Background Art

[0002] When performing active vibration and noise control, an adaptive filtering algorithm is usually adopted. Due to the introduction of the secondary channel, it is necessary to identify it to enhance the convergence of the algorithm. In the case of complex working conditions and harsh application environments, an online identification method is adopted, that is, secondary channel identification and active control are carried out simultaneously. Although this algorithm structure ensures the real-time modeling of the secondary channel, it introduces redundant vibration to the active vibration, resulting in a poor control effect. Summary of the Invention

[0003] The purpose of the present invention is to solve the problem in the prior art that redundant vibration is introduced to the active vibration, resulting in a poor control effect, and to provide an active control method for online identification of a secondary channel, which can ensure the implementation modeling of the secondary channel without introducing redundant vibration and has a good control effect.

[0004] To solve the above problems, the present application is implemented through the following technical solutions: An active control method for online identification of a secondary channel, characterized in that it includes the following steps: Step 1, obtaining a reference signal and random noise: An acceleration sensor is added to the vibration and noise source to collect the acceleration signal, which is called the reference signal , and the reference signal has a sampling frequency of , and the reference signal passes through the primary channel to generate a vibration and noise signal ; A power-adjustable random noise generator and a modeling FIR low-pass filter of order M are added to the active vibration and noise control system , wherein the random noise generator is used to generate random noise and outputs the white noise signal required for identifying the secondary channel when the active control system is turned on, and the modeling FIR low-pass filter is used to simulate the secondary channel in real time; Step 2, obtaining the first residual vibration signal : Input the reference signal obtained in Step 1 into the active controller of the active vibration and noise control system, adjust the weight coefficient of the active controller, and output a vibration and noise suppression signal to cancel the noise signal , , After cancellation, the first residual vibration signal is output ; Step 3: Obtain the second residual vibration signal :The first residual vibration signal output from step 2 With the random noise signal in step 1 By estimating the secondary channel Output Superposition, get the second residual vibration signal ; , Where: is the first residual vibration signal, is the noise signal, is the vibration noise suppression signal, To identify white noise signals, It is represented by a transverse filter of length M, that is, , Represents a random noise signal After estimating the secondary channel The vibration signal generated; Step 4: Determine the active controller weight coefficient And modeling FIR low-pass filters Iteration step: The second residual vibration signal obtained in step 3 is used Acts as the true residual signal to participate in the active controller control coefficients And modeling FIR low-pass filters The iterative process determines the iterative step length according to the control effect of the vibration and noise active control system; Step 5, decoupling the active control convergence of the vibration noise active control system from the secondary channel identification; Step 6: When the secondary channel transfer characteristics change, the residual vibration Increase, and repeat steps 1-5 to re-identify the secondary channel.

[0005] Furthermore, the weight coefficient adjustment of the active controller in step 2 is based on the filtered least mean square algorithm, which specifically includes the following steps: Step 2.1: Reference signal Perform convolution calculation, that is: , Where: is the reference signal, is the active controller coefficient vector, is the output signal; Step 2.2: The output signal obtained in step 2.1 and identification of white noise signals Superpose and jointly enter the secondary channel to output a vibration noise suppression signal , , In the formula: is the vibration signal generated after passing through the secondary channel ; denotes the secondary channel, and is the transfer characteristic of the entire process from the output signal to the first residual vibration signal , which is represented by a transverse filter of length M, i.e., ; Furthermore, the iterative process of the active controller control coefficient in step 4 is as follows: The iterative formula for the control coefficient is: , In the formula: is the iteration step size, is the reference signal for the filtered correction signal of the secondary channel, , In the formula: is the reference signal, which is represented by a transverse filter of length M, i.e., ; is the residual vibration, is the identification error; according to the control effect that in the initial stage of the vibration noise active control system, the residual vibration is the main factor to accelerate the convergence speed, and in the convergence stage, the identification error is the main factor to achieve the precise search of the system, determine the iteration step size as: ; In the formula: is the correction coefficient to adjust the iteration step size, usually in the range of 0.9 - 1.2; is the exponential calculation of e; is the power adjustment factor; Furthermore, the iterative process of the modeled FIR low-pass filter in step 4 is as follows: The iterative formula for the modeled FIR low-pass filter is: , In the formula: is the amplification factor; According to the power of white noise required in the active vibration and noise control system, the amplification factor is set as follows: , where: is a coefficient related to the convergence value of the first vibration residual error e(n), usually in the range of 0.0001 - 0.01; The setting of can perform segmented adjustment on the white noise power; The control process of the active vibration and noise control system is as follows: Initial stage: To ensure the accelerated convergence of the residual vibration, white noise does not participate in the initial stage, and white noise with a smaller power should be added; Convergence stage: As the residual vibration of the active control becomes smaller, the noise power should be gradually increased so that the modeling coefficient approaches the true value; End of convergence stage: When the active control and channel identification are stable, the noise power should be gradually reduced to zero, so as to reduce the influence of secondary channel identification on the active control, and make further reduced, and the control effect reaches the optimal; Furthermore, in step 5, the power adjustment factor is used to decouple the active control convergence and secondary channel identification of the active vibration and noise control system, and the power adjustment factor is: , where: represents the power of the second residual vibration signal at time n, represents the power of the random noise signal at time n, represents the power of the first residual vibration signal at time n, represents the power of the reference signal at time n; In the initial stage, the power of the random noise is small and the active control has not converged, so that ; Along with the active control process, the algorithm converges, , and is obtained. The power of the random noise decreases, , , and at this time , and is obtained. gradually decreases. At the same time, to ensure that the random noise is not submerged by the reference vibration signal, it is multiplied by the coefficient ; Furthermore, the values of , , , change relatively slowly and are determined by the method of single exponential smoothing prediction: , , , , wherein: respectively represent the power of the second residual vibration signal at the nth moment and the (n + 1)th moment, respectively represent the power of the random noise signal at the nth moment and the (n + 1)th moment, respectively represent the power of the first residual vibration signal at the nth moment and the (n + 1)th moment, respectively represent the power of the reference signal at the nth moment and the (n + 1)th moment, is the forgetting factor, .

[0006] The secondary channel online identification active control method of the present application outputs the identification signal required for the secondary channel and the FIR low-pass filter by adding a random noise generator to perform real-time modeling on the secondary channel; Introduce a power adjustment factor to adjust the iteration step of the active control link and the magnitude of the random noise power, effectively eliminating the coupling interference between active control and channel modeling; in the time-frequency domain, the comparison of the vibration signal suppression effect before and after shows the effective suppression of vibration by this method through the comparison of the initial vibration signal and the residual vibration signal; when the initial vibration signal changes, the residual vibration signal also achieves fast and accurate convergence, verifying the robustness of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 is the flow chart of the present application; Figure 2 is the amplitude diagram of each order before the change of the secondary channel transfer characteristics; Figure 3 is the amplitude diagram of each order after the change of the secondary channel transfer characteristics; Figure 4 is the time-domain effect comparison diagram of the vibration signal before and after suppression; Figure 5 is the frequency-domain effect comparison diagram of the vibration signal before and after suppression; Figure 6 is the flow chart of Embodiment 2; Figure 7 is Figure 6 the working principle diagram of. DETAILED DESCRIPTION OF THE INVENTION

[0008] The following refers to the accompanying drawings to give the specific implementation manner of the present invention to further illustrate the composition of the present invention.

[0009] Embodiment 1. A secondary channel online identification active control method, the specific process is as Figure 1As shown, it includes the following steps: Step 1, obtain a reference signal and random noise: Add an acceleration sensor to the vibration noise source to collect the acceleration signal, which is called the reference signal , the reference signal has a sampling frequency of , the reference signal passes through the primary channel to generate a vibration noise signal ; Add a power-adjustable random noise generator and a modeling FIR low-pass filter of order M to the vibration noise active control system , where the random noise generator is used to generate random noise and outputs the white noise signal required for identifying the secondary channel while the active control system is turned on , and the modeling FIR low-pass filter is used to simulate the secondary channel in real time; Step 2, obtain the first residual vibration signal : Input the reference signal obtained in Step 1 into the active controller of the vibration noise active control system, adjust the weight coefficients of the active controller, and output a vibration noise suppression signal to cancel the noise signal , , and output the first residual vibration signal after cancellation ; Step 3, obtain the second residual vibration signal : Add the first residual vibration signal output in Step 2 to the random noise signal in Step 1 through the estimated secondary channel and output to obtain the second residual vibration signal ; , In the formula: is the first residual vibration signal, is the noise signal, is the vibration noise suppression signal, is the identified white noise signal, is represented by a transversal filter of length M, that is , represents the vibration signal generated by the random noise signal passing through the estimated secondary channel ; Step 4, determine the control coefficient of the active controller And modeling FIR low-pass filters Iteration step: The second residual vibration signal obtained in step 3 is used Acts as the true residual signal to participate in the active controller control coefficients And modeling FIR low-pass filters The iterative process determines the iterative step length according to the control effect of the vibration and noise active control system; Step 5, decoupling the active control convergence of the vibration noise active control system from the secondary channel identification; Introducing Power Regulation Factor , so that the active control link and the secondary channel modeling are effectively separated. At the beginning, the active control link is accelerated to converge first, and the secondary channel is not modeled. As the active control gradually converges, the step size is reduced and fine-tuning begins. At the same time, the output power of the random noise signal is increased to accelerate the modeling of the secondary channel, thereby effectively decoupling the active control convergence and channel identification; The weight coefficient adjustment of the active controller in step 2 is based on the filtered least mean square algorithm, which specifically includes the following steps: Step 2.1: Reference signal Perform convolution calculation, that is: , Where: is the reference signal, is an active controller, is the output signal; Step 2.2: The output signal obtained in step 2.1 and identification of white noise signals Superposition and common entry into the secondary channel , output vibration noise suppression signal , , Where: for Vibration signal generated through the secondary channel; Indicates the secondary channel, which is the output signal To the first residual vibration signal The transfer characteristic of the whole process is represented by a transverse filter with a length of M, that is, ; Among them, the control coefficient of the active controller in step 4 is The iteration process is as follows: Control coefficient The iteration formula is: , Where: is the iteration step size; is the reference signal Perform filtering correction of the secondary channel signal, , In the formula: is the reference signal, is represented by a transversal filter of length M, that is ; is the residual vibration, is the identification error; According to the control effect of the vibration noise active control system that the residual vibration is the main part in the initial stage to accelerate the convergence speed, and the identification error is the main part in the convergence stage to achieve the precise search of the system, determine the iteration step size is: ; In the formula: is the correction coefficient to adjust the iteration step size in the range of 0.9 - 1.2; is the exponential calculation of e; is the power adjustment factor; Initial stage: Residual vibration is the main part, and a larger iteration step size needs to be set to accelerate the convergence speed, the influence of the identification error is not considered; Convergence stage: As the residual vibration signal decreases, the identification error is the main part, the influence of the residual vibration signal is not considered, adjust the iteration step size to a smaller value, precisely search for the optimal value of the system, and complete the active control; Among them, the iterative process of modeling the FIR low - pass filter in step 4 is as follows: The iterative formula for modeling the FIR low - pass filter is: , In the formula: is the amplification coefficient; According to the power of the white noise required in the vibration noise active control system, set the amplification coefficient as follows: , In the formula: is the coefficient related to the convergence value of the first vibration residual error e(n), usually in the range of 0.0001 - 0.01; The setting of can adjust the white noise power in a segmented manner; The control process of the vibration noise active control system is as follows: Initial stage: To ensure the accelerated convergence of residual vibration, white noise is not involved in the initial stage, and white noise with a relatively small power should be added. Convergence stage: As the residual vibration of active control is relatively small, the noise power should be gradually increased to make the modeling coefficient approach the true value. Final stage of convergence: When the active control and channel identification are stable, the noise power should be gradually reduced to zero, so as to reduce the influence of secondary channel identification on active control, making further reduced and the control effect reach the optimal. Among them, in step 5, the power adjustment factor is used to decouple the active control convergence and secondary channel identification of the vibration noise active control system. The power adjustment factor is: , where: represents the power of the second residual vibration signal at time n, represents the power of the random noise signal at time n, represents the power of the first residual vibration signal at time n, represents the power of the reference signal at time n; In the initial stage, the random noise power is relatively small and the active control has not converged, making ; Along with the active control process, the algorithm converges, , obtaining , the random noise power decreases, , , at this time , obtaining , gradually decreases. At the same time, to ensure that the random noise is not submerged by the reference vibration signal, it is multiplied by the coefficient ; Among them, the values of the , , , change relatively slowly and are determined by the method of single exponential smoothing prediction: , , , , where: respectively represent the power of the second residual vibration signal at time n and (n + 1), respectively represent the power of the random noise signal at time n and (n + 1), respectively represent the power of the first residual vibration signal at time n and (n + 1), respectively represent the power of the reference signal at time n and (n + 1), is the forgetting factor, .

[0010] Experimental verification was carried out on the above method, and vibration and noise control simulation was carried out on the MATLAB / Simulink platform. The reference signal is synthesized by sine signals with frequencies of 25 Hz and 30 Hz, amplitudes of 1, and zero phase. A random noise generator generates random Gaussian white noise with a mean of zero and a variance of 0.001 for online identification and active control of the secondary channel. The real secondary channel is simulated by a 200-order transverse FIR filter, and its amplitude characteristics are as Figure 2 shown.

[0011] The set parameters are a sampling rate of 10,000 Hz and an identification filter length of 300 orders, is 1, is 0.001. At the same time, in order to verify the robustness of the algorithm, it is set that the transfer characteristics of the secondary channel change at the 5th second, and its amplitude characteristics are as Figure 3 shown. The settings of each parameter are listed in Table 1.

[0012] Table 1 Simulation parameter settings table , After the settings are completed, the simulation is started, and a comparison diagram of the effects of the method on suppressing vibration signals before and after is obtained, as Figures 4 - 5 shown, Figure 4 is the time-domain effect diagram, Figure 5 is the frequency-domain effect diagram. Through the comparison of the initial vibration signal and the residual vibration signal, in the time domain, the convergence speed is very fast, and the amplitude drops by 95%; in the frequency domain, the vibration line spectra at 25 Hz and 30 Hz are reduced by 30 dB and 32 dB respectively, indicating the effective suppression of vibration by the algorithm. At the 5th second, the transfer characteristics of the secondary channel change, and the amplitude of the residual vibration signal converges rapidly to 90% at the 6th second and to 95% at the 7th second. This shows that identification can be quickly achieved again when the secondary channel changes, verifying the robustness of the algorithm.

[0013] Example 2. An online identification and active control method for the secondary channel, as Figure 6 shown, its steps are basically the same as those of Example 1, and the difference is that: it further includes step 6. When the transfer characteristics of the secondary channel change, the residual vibration increases, and steps 1-5 are repeated to re-identify the secondary channel.

[0014] The working principle of this embodiment is as Figure 7 shown, an additional random noise generator outputs the identification signal required for the secondary channel and the FIR low-pass filter to perform real-time modeling on the secondary channel; At the 5th second, the reference signal remains unchanged while the transfer characteristics of the secondary channel change, and its amplitude characteristics are as Figure 3 shown.

[0015] Introduce a power adjustment factor to adjust the iteration step size of the active control link and the magnitude of the random noise power, effectively eliminating the coupling interference between active control and channel modeling.

[0016] The specific embodiments described herein are merely illustrative of the spirit of the present invention. Those skilled in the art to which the present invention pertains may make various modifications or supplements to the described specific embodiments or use similar methods for substitution, but will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. An online identification active control method for a secondary channel, characterized in that: It includes the following steps: Step 1, obtain a reference signal and random noise; Step 2: Obtain the first residual vibration signal : Input the reference signal obtained in Step 1 into the active controller of the vibration and noise active control system, adjust the weight coefficients of the active controller, and output a vibration and noise suppression signal to cancel the noise signal , Output the first residual vibration signal after cancellation ; Step 3: Obtain the second residual vibration signal : The first residual vibration signal output in Step 2 and the random noise signal in Step 1 are superimposed through estimating the secondary channel output to obtain the second residual vibration signal ; ， Wherein: is the first residual vibration signal, is the noise signal, is the vibration noise suppression signal, is the identified white noise signal, which is also represented by a transversal filter with a length of M, i.e., , Represents a random noise signal After estimating the secondary channel The generated vibration signal; Step 4: Determine the active controller weight coefficient and model the FIR low-pass filter Iteration step: Use the second residual vibration signal obtained in Step 3 as the true residual signal to participate in the iteration process of the active controller control coefficient and model the FIR low-pass filter According to the control effect of the vibration noise active control system, determine the iteration step; Step 5, decouple the active control convergence and secondary channel identification of the vibration noise active control system.

2. The online identification active control method for a secondary channel according to claim 1, wherein: It further includes Step 6, when the transfer characteristic of the secondary channel changes, repeat Steps 1-5 to re-identify the secondary channel.

3. A method for online identification and active control of a secondary channel according to claim 1 or 2, characterized in that: The specific steps of Step 1 are as follows: An acceleration sensor is added to the vibration noise source to collect acceleration signals, which are called reference signals , the reference signal has a sampling frequency of , the reference signal passes through the primary channel to generate a vibration noise signal ; Add a power-adjustable random noise generator and a modeling FIR low-pass filter of order M to the active vibration and noise control system , where the random noise generator is used to generate random noise and output the identification white noise signal required for the secondary channel while the active control system is turned on , and the modeling FIR low-pass filter is used to perform real-time simulation on the secondary channel .

4. The active control method for online identification of a secondary channel according to claim 3, wherein: In step 2, the weight coefficient of the active controller is adjusted according to the filtered least mean square algorithm, which specifically includes the following steps: Step 2.1, perform convolution calculation on the reference signal That is: , Wherein: is a reference signal, is an active controller, is an output signal; Step 2.2: The output signal obtained in Step 2.1 is superimposed with the identified white noise signal and they jointly enter the secondary channel to output a vibration noise suppression signal , , Wherein: is the vibration signal generated through the secondary channel; represents the secondary channel and is the output signal to the first residual vibration signal for the entire process of transfer characteristics, represented by a transverse filter of length M, that is .

5. The online identification active control method for a secondary channel according to claim 3, wherein: The iteration process of the control coefficient of the active controller in step 4 is as follows: Control coefficient The iterative formula is as follows: ， In the formula: is the iteration step size; is a reference signal performs filtering and correction of the secondary channel signal ， Wherein: is a reference signal, is represented by a transversal filter of length M, that is ； is the residual vibration, is the identification error; Based on the control effect that in the initial stage of the vibration noise active control system, the residual vibration is dominant and the convergence speed is accelerated, and in the convergence stage, the identification error is dominant and the precise search of the system is realized, determine Iterative step size is: ; Where: is the correction coefficient, used to adjust the iteration step size, usually in the range of 0.9 - 1.2; is the exponential calculation of e; is the power adjustment factor.

6. The online identification active control method for a secondary channel according to claim 3, characterized in that: Modeling the FIR low-pass filter in step 4 The iterative process is as follows: Modeling FIR Low-Pass Filter Iterative formula: , In the formula: is the amplification factor; According to the power of the white noise required in the active vibration and noise control system, the amplification factor is set as follows: , Wherein: is a coefficient related to the convergence value of the first vibration residual error e(n), within the range of 0.0001 - 0.

01.

7. An on-line identification active control method for a secondary channel according to claim 3, characterized in that: The power adjustment factor is adopted in the step 5 Decouple the active control convergence and secondary path identification of the vibration noise active control system, and the power adjustment factor : , Wherein: represents the power of the second residual vibration signal at the nth moment, represents the power of the random noise signal at the nth moment, represents the power of the first residual vibration signal at the nth moment, represents the power of the reference signal at the nth moment.

8. The active control method for online identification of a secondary channel according to claim 7, characterized in that: In the said step 5, the , , , values are determined by the method of single exponential smoothing prediction: ， ， ， ， Wherein: respectively represent the power of the second residual vibration signal at the nth moment and the (n + 1)th moment, respectively represent the power of the random noise signal at the nth moment and the (n + 1)th moment, respectively represent the power of the first residual vibration signal at the nth moment and the (n + 1)th moment, respectively represent the power of the reference signal at the nth moment and the (n + 1)th moment, is the forgetting factor, .

Images