Method for realizing active noise reduction of ventilation system of transformer substation

Abstract

The invention relates to a method for realizing active noise reduction of a transformer substation ventilation system. Firstly, noise of a transformer substation ventilation system is collected; then establishing a basic LMS adaptive noise cancellation model; and finally, establishing an adaptive noise cancellation model based on the variable step size LMS. Based on the principle of sound wave interference counteracting, the active noise reduction system utilizes a self-adaptive filtering algorithm to change the structure and output of the system according to noise and emit anti-noise sound waves, so that the anti-noise sound waves and noise source sound waves are subjected to interference counteracting in the space, the problem of fan noise of the transformer substation is effectively solved, the noise pollution treatment level of the transformer substation is improved, and the service life of the transformer substation is prolonged. Technical support is provided for safe and stable operation of the transformer substation, and effective control measures are provided for noise of a built urban transformer substation ventilation system.

Description

Technical Field

[0001] The present invention belongs to the technical field of noise reduction of electric power ventilation systems, and in particular relates to a method for realizing active noise reduction of a ventilation system in a substation. Background Art

[0002] In recent years, with the continuous expansion of urban boundaries and the continuous construction and development of transmission and distribution networks, some substations have been forced to be built near or even within residential areas. The resulting noise problem of urban substations has become increasingly prominent. The high-power exhaust fans within the stations, running 24 hours a day, generate power-frequency noise, which is the main source of noise in urban substations. In the dead of night, it easily contrasts with the surrounding background sound, affecting surrounding residents and causing noise nuisance. Currently, the noise of urban substation ventilation systems has become a significant source of noise pollution in urban life, and is increasingly attracting public attention and becoming a focus of environmental complaints. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a method for achieving active noise reduction in a transformer substation ventilation system.

[0004] The technical solution adopted by the present invention is: a method for realizing active noise reduction of a substation ventilation system, comprising the following steps:

[0005] S1: Collect noise from the substation ventilation system;

[0006] S2: Establish a basic LMS adaptive noise cancellation model;

[0007] S3: Establish a variable step size LMS adaptive noise cancellation model.

[0008] Specifically, in step S2, the iterative formula of the basic LMS algorithm is:

[0009] e(n)=d(n)-X T (n)W N (n)

[0010] W N (n+1)=W N (n)+2μe(n)X N (n)

[0011] Where W N (n) is the weight vector of the adaptive filter at time n; μ is a constant that controls the convergence speed, called the step factor; the LMS algorithm convergence condition is 0<μ<1 / λ max ,λ max is the largest eigenvalue of the autocorrelation matrix of the input signal.

[0012] In step S3, the specific iterative formula of the improved LMS algorithm based on variable step size is:

[0013]

[0014] W'(n)=W(n)+2μe(n)X(n)

[0015] y'(n)=[W'(n)] T X(n)

[0016] e'(n)=d(n)-y'(n)

[0017] W(n+1)=W'(n)+2μe'(n)X(n)

[0018] Substituting in the equation, we get:

[0019] W(n+1)=W(n)+2μ{d(n)-[W(n)+2μe(n)X(n)] T X(n)}X(n)=W(n)+4μ[1-μ||X(n)|| 2 ]e(n)X(n)

[0020] It can be seen that the improved LMS algorithm based on variable step size is equivalent to replacing the step size factor μ of the basic LMS algorithm with μ'=2μ[1-μ||X(n)|| 2 ], it can be seen from the expression that μ' is a quadratic parabolic function about μ. When μ is small, μ'>μ, which makes the convergence speed faster; when μ is large, μ'<μ, which makes the convergence speed slower. As μ increases, the equivalent step length μ' first increases and then decreases. In order to take into account both the convergence speed and the steady-state offset signal, the maximum value of μ' can be taken.

[0021] because Maximum value Then the weight coefficient update vector corresponding to the new algorithm is:

[0022]

[0023] The control convergence parameter in the basic LMS algorithm is a fixed constant μ, which leads to slow convergence. The step size factor of the improved LMS algorithm based on variable step size is This value is variable, and the value is adjusted in real time based on the size of the sampling power at time n, that is, the size of the step factor. In order to compromise, the maximum value of the step factor adjustment process can be taken. This ensures both the convergence speed and the steady-state offset error requirements.

[0024] The beneficial effects of the present invention are as follows: The present invention first collects noise from the substation ventilation system; then establishes a basic LMS adaptive noise cancellation model; and finally establishes a variable-step-size LMS adaptive noise cancellation model. Based on the principle of acoustic wave interference cancellation, the present invention uses an active noise reduction system that adapts its structure and output according to the noise, emitting anti-noise sound waves that interfere and cancel with the noise source sound waves in space. This effectively resolves the problem of substation fan noise, improves the level of noise pollution control in substations, provides technical support for the safe and stable operation of substations, and offers effective control measures for noise in substation ventilation systems in existing urban areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flow chart of the steps in the present invention;

[0026] Figure 2 This is a schematic diagram of the principle of adaptive noise cancellation in the present invention;

[0027] Figure 3 Schematic diagram of a noisy signal in the present invention;

[0028] Figure 4 Schematic diagram of the expected signal in the present invention;

[0029] Figure 5 Schematic diagram of the signal after denoising by the traditional basic algorithm in the present invention;

[0030] Figure 6 Schematic diagram of the improved LMS algorithm signal based on variable step size in the present invention;

[0031] Figure 7 Schematic diagram of mean square error of the traditional basic algorithm in the present invention;

[0032] Figure 8 Schematic diagram of the mean square error of the improved LMS algorithm based on variable step size in the present invention. DETAILED DESCRIPTION

[0033] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of the present invention and are described in detail in conjunction with the embodiments below.

[0034] like Figure 1-8 As shown, the present invention includes the following steps:

[0035] S1: Collect noise from the substation ventilation system;

[0036] S2: Establish a basic LMS adaptive noise cancellation model; the specific process is as follows:

[0037] In step S2, the iterative formula of the basic LMS algorithm is:

[0038] e(n)=d(n)-X T (n)W N (n)

[0039] W N (n+1)=W N (n)+2μe(n)X N (n)

[0040] Where W N (n) is the weight vector of the adaptive filter at time n; μ is a constant that controls the convergence speed, called the step factor; the LMS algorithm convergence condition is 0<μ<1 / λ max ,λ max is the largest eigenvalue of the autocorrelation matrix of the input signal.

[0041] (1) Analysis of parameter μ. The conditions for the algorithm to converge are:

[0042] 0<μ<1 / λ max

[0043] Among them, the diagonal elements λ i is the characteristic value of the autocorrelation function of the sampled signal, λ min is the minimum eigenvalue, λ max is the maximum eigenvalue.

[0044] (2) Convergence time of the LMS algorithm. In the sense of mean square error convergence, it can be seen that its corresponding time constant is:

[0045] τ max =1 / (4μλ min )

[0046] It can be seen that, λ min The smaller τ max The larger the value, the slower the convergence; from 0<μ<1 / λ max It can be seen that λ max The larger μ max The smaller it is, the slower the convergence. This means that for noise cancellation, the LMS algorithm converges faster when the signal spectrum is wider. Therefore, to speed up the convergence of the algorithm, it is hoped that the μ value will be larger.

[0047] (3) The offset coefficient of the LMS algorithm. The offset coefficient of the LMS is defined as the ratio of the excess mean square error to the minimum mean square error:

[0048] M = MSE 超量 / ξ min =μ·tr[R]=μ·L·P in

[0049] Where: L is the filter order; P in is the signal input power. It can be seen that the larger μ is, the larger M is. Therefore, to reduce the stability offset error, it is desirable to keep μ as small as possible.

[0050] S3: Establish a variable step-size LMS adaptive noise cancellation model; the specific process is as follows:

[0051] In step S3, the specific iterative formula of the improved LMS algorithm based on variable step size is:

[0052]

[0053] W'(n)=W(n)+2μe(n)X(n)

[0054] y'(n)=[W'(n)] T X(n)

[0055] e'(n)=d(n)-y'(n)

[0056] W(n+1)=W'(n)+2μe'(n)X(n)

[0057] Substituting in the equation, we get:

[0058] W(n+1)=W(n)+2μ{d(n)-[W(n)+2μe(n)X(n)] T X(n)}X(n)=W(n)+4μ[1-μ||X(n)|| 2 ]e(n)X(n)

[0059] It can be seen that the improved LMS algorithm based on variable step size is equivalent to replacing the step size factor μ of the basic LMS algorithm with μ'=2μ[1-μ||X(n|)| 2 ], it can be seen from the expression that μ' is a quadratic parabolic function about μ. When μ is small, μ'>μ, which makes the convergence speed faster; when μ is large, μ'<μ, which makes the convergence speed slower. As μ increases, the equivalent step length μ' first increases and then decreases. In order to take into account both the convergence speed and the steady-state offset signal, the maximum value of μ' can be taken.

[0060] because Maximum value Then the weight coefficient update vector corresponding to the new algorithm is:

[0061]

[0062] The control convergence parameter in the basic LMS algorithm is a fixed constant μ, which leads to very slow convergence; the step size factor of the improved LMS algorithm based on variable step size is μ'=2μ[1-μ||X(n)|| 2 ], the value is variable, and the value is adjusted in real time according to the size of the sampling power at time n, that is, the size of the step factor. In order to compromise, the maximum value of the step factor adjustment process can be μ'=2μ[1-μ||X(n)|| 2 ], which not only ensures the convergence speed but also meets the steady-state offset error requirements.

[0063] Figure 2 The schematic diagram of adaptive noise cancellation is shown in Figure 2. The original signal d(n) includes the signal s(n) and the noise N(n). x(n) is the reference noise input. N1(n) is related to N(n) but not to s(n).

[0064] The principle of adaptive noise cancellation is as follows:

[0065] e(n)=d(n)-y(n)=s(n)+N(n)-y(n)

[0066] e 2 (n)=s 2 (n)+[N(n)-y(n)] 2 +2s(n)[N(n)-y(n)]

[0067] Taking the mathematical expectation on both sides of the above formula, we get:

[0068] E[e 2 (n)]=E[s 2 (n)]+E[(N(n)-y(n)) 2 ]+2E[s(n)·(N(n)-y(n))]

[0069] Since s(n) is not related to N(n) and N1(n), then:

[0070] E[e 2 (n)]=E[s 2 (n)]+E[(N(n)-y(n)) 2 ]

[0071] When the filter parameters are adjusted so that E[e 2 (n)] is minimized, E[s 2 (n)] is not affected, so that

[0072] E min [e 2 (n)]=E[s 2 (n)]+E min [(N(n)-y(n))2 ]

[0073] When E min [e 2 (n)] is minimum, E min [(N(n)-y(n)) 2 ] is also minimum, the output y(n) of the adaptive filter is the best estimate of N(n), and the system output is

[0074] e(n)=d(n)-y(n)=s(n)+N(n)-y(n)

[0075] In this way, e(n) will approach the useful signal s(n). Ideally, y(n) = N(n), e(n) = s(n).

[0076] Specific simulation and result analysis

[0077] The improved LMS algorithm based on variable step size is simulated on the Matlab platform. The simulation conditions are as follows: ① The adaptive filter order L = 20, the number of sampling points is 500, and the fixed step size factor μ = 0.01; ② The noisy signal d(n) = sin(0.04Πt) + 0.5·sin(0.1Πt) + randn; the expected signal x(t) = sin(0.04Πt), as shown in the following example. Figure 3 and Figure 4 As shown. The improved LMS algorithm based on variable step size and the fixed step size LMS algorithm are respectively called for simulation, and the influence of μ on the convergence speed and the amount of misalignment is observed and recorded. μ=1 / (2||X(n)|| 2 ),μ=0.01.

[0078] The simulation results are as follows Figures 3 to 8 shown.

[0079] Figure 4 is a useful or desired signal, i.e. a sine wave signal, Figure 3 is the input signal, that is, the signal contaminated by noise.

[0080] Depend on Figure 5 and Figure 6 By comparison, it can be seen that the denoising result of the improved algorithm is very close to the expected signal and is much better than the traditional basic algorithm. Figure 7 and Figure 8 By comparison, it can be seen that the improved LMS algorithm based on variable step size converges very quickly, and begins to converge after about 200 iterations, while the convergence speed of the basic LMS algorithm is much slower, and begins to converge after about 600 iterations. On the offset signal, the offset of the new algorithm is δ=μ(n)·L·P in , since μ(n)=1 / (2||X(n)||2 ), so δ1=1 / (2||X(n)|| 2 )·L·P in , after reaching steady state, the offset is a constant value and does not change with the step size; the offset of the traditional basic algorithm is δ2=μ·L·P in ,This value is affected by the step size. When the traditional basic algorithm enters the steady state, the size of each offset error is calculated and the average is taken. The result is about 20.158%. When the improved LMS algorithm based on variable step size enters the steady state, the size of each offset is calculated and the average is taken. The value is 15.163%. It can be seen that the steady-state offset of the improved LMS algorithm based on variable step size is not large and is smaller than that of the traditional algorithm.

[0081] To improve the performance of the LMS algorithm, this paper analyzes existing basic LMS algorithms and makes several improvements. The relationship between μ, convergence rate, and steady-state misalignment is discussed. Based on this, an improved LMS algorithm based on a variable step size is studied, establishing a new nonlinear relationship between the step size factor μ, the input signal, and the number of iterations, n. Theoretical analysis shows that compared to the traditional basic LMS algorithm, this algorithm has a faster convergence rate, smaller steady-state error, and a minimal increase in computational complexity. Simulation experiments using the improved LMS algorithm based on a variable step size demonstrate that the adaptive noise canceller based on the improved LMS algorithm can effectively resist noise interference and exhibits excellent noise cancellation capabilities for noisy signals.

[0082] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and it is intended that all variations within the meaning and scope of the appended claims be encompassed. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for achieving active noise reduction in a substation ventilation system, characterized in that: The steps include: S1: Collect noise from the substation ventilation system; S2: Establish a basic LMS adaptive noise cancellation model; S3: Establish a variable step size LMS adaptive noise cancellation model.

2. The method for achieving active noise reduction in a substation ventilation system according to claim 1, characterized in that: In step S2, the iterative formula of the basic LMS algorithm is: e(n)=d(n)-X T (n)W N (n) W N (n+1)=W N (n)+2μe(n)X N (n) Where W N (n) is the weight vector of the adaptive filter at time n; μ is a constant that controls the convergence speed, called the step factor; The convergence condition of the LMS algorithm is 0<μ<1 / λ max ,λ max is the largest eigenvalue of the autocorrelation matrix of the input signal.

3. The method for realizing active noise reduction of a substation ventilation system according to claim 1, characterized in that: In step S3, the specific iterative formula of the improved LMS algorithm based on variable step size is: W'(n)=W(n)+2μe(n)X(n) y'(n)=[W'(n)] T X(n) e'(n)=d(n)-y'(n) W(n+1)=W'(n)+2μe'(n)X(n) Substituting in the equation, we get: W(n+1)=W(n)+2μ{d(n)-[W(n)+2μe(n)X(n)] T X(n)}X(n)=W(n)+4µ[1-µ||X(n)|| 2 ]e(n)X(n) It can be seen that the improved LMS algorithm based on variable step size is equivalent to replacing the step size factor μ of the basic LMS algorithm with μ'=2μ[1-μ||X(n)|| 2 ], it can be seen from the expression that μ' is a quadratic parabolic function about μ. When μ is small, μ'>μ, which makes the convergence speed faster; when μ is large, μ'<μ, which makes the convergence speed slower. As μ increases, the equivalent step length μ' first increases and then decreases. In order to take into account both the convergence speed and the steady-state offset signal, the maximum value of μ' can be taken. because Maximum value Then the weight coefficient update vector corresponding to the new algorithm is: The control convergence parameter in the basic LMS algorithm is a fixed constant μ, which leads to very slow convergence; the step size factor of the improved LMS algorithm based on variable step size is μ'=2μ[1-μ||X(n)|| 2 ], the value is variable, and the value is adjusted in real time according to the size of the sampling power at time n, that is, the size of the step factor. In order to compromise, the maximum value of the step factor adjustment process can be μ'=2μ[1-μ||X(n)|| 2 ], which not only ensures the convergence speed but also meets the steady-state offset error requirements.

Images