Electric energy quality signal noise reduction and optimization detection method and system and application thereof
By optimizing variational mode decomposition parameters and multi-chip collaborative architecture using particle swarm optimization algorithm, the problem of noise interference in power quality detection is solved, and high-precision signal denoising and harmonic analysis are achieved.
Patent Information
- Application Number
- CN202510809105.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-17
- Publication Date
- 2025-11-14
AI Technical Summary
Existing technologies struggle to balance real-time performance and accuracy in power quality detection under complex noise environments. Traditional methods such as Fourier transform and wavelet denoising are ineffective under high-frequency noise and electromagnetic interference.
The penalty factor and number of modes in variational mode decomposition are dynamically optimized using particle swarm optimization algorithm. A fitness function with composite index as the core is constructed to screen the optimal decomposition parameters. Signal denoising and optimized detection are then performed in combination with a multi-chip collaborative architecture.
It effectively suppresses noise interference, improves the accuracy of harmonic analysis, and has a detection error of less than 2%, enabling power quality monitoring in high-noise environments.
Smart Images
Figure CN120950828A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system monitoring technology, specifically to a method, system, and application of power quality signal noise reduction and optimization detection. Background Technology
[0002] Power quality monitoring requires real-time analysis of parameters such as harmonics, flicker, and imbalance of voltage and current signals. However, in actual operating conditions, signals are often affected by high-frequency noise and electromagnetic interference. Traditional methods (such as Fourier transform and wavelet denoising) struggle to balance real-time performance and accuracy in complex noise environments. Therefore, an adaptive denoising method is urgently needed to improve signal quality and ensure detection accuracy. Summary of the Invention
[0003] The purpose of this invention is to address the shortcomings of existing technologies by providing a power quality signal noise reduction and optimization detection method, system, and application. By optimizing VMD decomposition parameters, it effectively suppresses noise interference, improves harmonic analysis accuracy, and is suitable for power quality monitoring in high-noise environments.
[0004] The technical solution of this invention to solve the above problems is: a method for power quality signal noise reduction and optimization detection, specifically including the following steps:
[0005] Step S1: Dynamically optimize the penalty factor δ and the number of modes β of variational mode decomposition (VMD) using the particle swarm optimization (PSO) algorithm, construct a fitness function with composite indices (permutation entropy and mutual information) as the core, and screen the optimal decomposition parameters.
[0006] Step S2: Perform VMD decomposition on the power quality signal based on the optimized decomposition parameters δ and β to obtain multiple intrinsic mode function components, i.e., IMF components;
[0007] Step S3: Based on composite indices, the optimal signal component is selected from the IMF components to achieve signal noise reduction.
[0008] Furthermore, the parameters of the particle swarm optimization algorithm in step S1 are set as follows:
[0009] The initial population size χ = 20, the number of iterations ψ = 50, the initial range of the penalty factor δ is [0, 6], and the initial range of the number of modes β is [0, 5000].
[0010] The particle position (δ,β) is constructed using the parameter penalty factor δ and the number of modes β.
[0011] Furthermore, step S2 specifically includes the following steps:
[0012] In each iteration of the particle swarm optimization algorithm:
[0013] Based on the positions (δ, β) of each particle, the signal is decomposed using VMD, and the corresponding fitness value f is calculated.
[0014] The fitness function is defined as follows:
[0015]
[0016] CI = MI / H′ p
[0017]
[0018] Where: CI is a composite index of permutation entropy Hp' and mutual information MI, mi(A,B)=H(B)—H(B|A), A and B represent two different events, H(B) is the entropy of B, H(A) is the entropy of A, and H(B|A) is the conditional entropy of B when A is known; A is the original input signal, and B is the signal after adding the IMF components after VMD decomposition;
[0019]
[0020] In the formula, D≥2 is the embedding dimension, and S D for The set of all possible permutations of the column positions of the elements in the array. Let τ be the D-dimensional delay embedding vector of the input signal time series {x(k), k = 1, 2, ..., N} at time i, where τ is the time delay, i = 1, 2, ..., N - (D-1)τ, Π j = (j1,j2,…,jD) is the set S D A permutation that simultaneously satisfies the following two conditions:
[0021] ①x(i+(j1-1)τ)≤x(i+(j2-1)τ)≤…≤x(i+(j D -1)τ);
[0022] ②If x(i+(j) s-1 -1)τ)=x(i+(j s –1)τ), then j s-1 ≤j s ;
[0023] P(Π j ) for Π j The probability of a sequence of symbols is expressed as:
[0024]
[0025] In the formula, for China has Π j The number of permutations, 1≤j≤N-(D-1)τ
[0026] The permutation entropy of the corresponding time series is defined as:
[0027]
[0028] Normalize the above expression:
[0029]
[0030] Furthermore, the criteria for selecting the optimal signal component in step S3 are: the IMF component with the largest composite index CI value, and the permutation entropy Hp' ≤ 0.3 and mutual information MI ≥ 0.8.
[0031] Based on the fitness value f, adjust the individual optimal value pbest(t) and the global optimal value p g (t)
[0032] The velocity and position update formulas of the particle swarm optimization algorithm are as follows:
[0033]
[0034] In the formula, t represents the number of iterations, vm(t) and xm(t) represent the velocity and position (δ,β) of particle m in the t-th iteration, respectively, w is the inertia weight factor, c1 and c2 are acceleration factors, and r1 and r2 are random numbers between [0,1].
[0035] After each iteration, VMD decomposition is performed on the updated particle position (δ,β), and the corresponding fitness value f is calculated. After the maximum number of iterations, the optimal particle position (δ,β) is output, and the optimal VMD denoising parameters δ and β are obtained. Finally, the denoising processing of the input signal is completed using these optimized parameters.
[0036] Furthermore, it also includes: step S4, combining a multi-chip collaborative architecture to perform real-time harmonic analysis on the noise-reduced signal and output power quality parameters.
[0037] Furthermore, the multi-chip collaborative architecture is FPGA+ARM+DSP.
[0038] Furthermore, it also includes: step S5, performing transient disturbance detection on the denoised signal, extracting transient features through short-time Fourier transform (STFT), with a response time ≤10ms.
[0039] A power quality signal denoising and optimization detection system, employing the aforementioned power quality signal denoising and optimization detection method, includes:
[0040] The signal acquisition module is used to acquire power quality signals in real time.
[0041] The parameter optimization module is used to optimize VMD parameters;
[0042] The noise reduction module is used to perform VMD decomposition and component selection on the signal based on optimized parameters.
[0043] The harmonic analysis module is used to perform real-time harmonic detection on the noise-reduced signal and output power quality parameters.
[0044] The transient detection module is used to identify voltage surges, drops, interruptions, and transient oscillations.
[0045] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method.
[0046] The above-mentioned power quality signal noise reduction and optimization detection methods are applied in online power quality monitoring and fault diagnosis of microgrids, rail transit power supply systems, new energy power plants, and industrial power distribution systems.
[0047] The present invention has the following beneficial effects:
[0048] This invention provides a power quality signal noise reduction and optimization detection method, which can optimize the parameters of VMD decomposition for power quality signals, thereby overcoming noise interference in the signal, solving the problem of low signal-to-noise ratio, and ultimately effectively improving the detection accuracy of power quality, achieving a measurement error of less than 2%. Attached Figure Description
[0049] Figure 1 This is a flowchart of VMD noise reduction for laser ultrasound signals in this invention;
[0050] Figure 2 The figure shows the optimization results of VMD noise reduction parameters in this test case;
[0051] Figure 3 This is a diagram showing the VMD decomposition results of the acquired signals in this test case. Detailed Implementation
[0052] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.
[0053] As shown in the figure, a method for power quality signal noise reduction and optimization detection specifically includes the following steps:
[0054] Step S1: Dynamically optimize the penalty factor δ and the number of modes β of variational mode decomposition (VMD) using the particle swarm optimization (PSO) algorithm, construct a fitness function with composite indices (permutation entropy and mutual information) as the core, and screen the optimal decomposition parameters.
[0055] The parameters of the particle swarm optimization algorithm in step S1 are set as follows:
[0056] The initial population size χ = 20, the number of iterations ψ = 50, the initial range of the penalty factor δ is [0, 6], and the initial range of the number of modes β is [0, 5000].
[0057] The particle position (δ,β) is constructed using the parameter penalty factor δ and the number of modes β.
[0058] Step S2 involves performing VMD decomposition on the power quality signal based on the optimized decomposition parameters δ and β to obtain multiple intrinsic mode function components, i.e., IMF components. This specifically includes the following steps:
[0059] In each iteration of the particle swarm optimization algorithm:
[0060] Based on the positions (δ, β) of each particle, the signal is decomposed using VMD, and the corresponding fitness value f is calculated.
[0061] The fitness function is defined as follows:
[0062]
[0063] CI = MI / H′ p
[0064]
[0065] Where: CI is a composite index of permutation entropy Hp' and mutual information MI, mi(A,B)=H(B)—H(B|A), A and B represent two different events, H(B) is the entropy of B, H(A) is the entropy of A, and H(B|A) is the conditional entropy of B when A is known; A is the original input signal, and B is the signal after adding the IMF components after VMD decomposition.
[0066]
[0067] In the formula, D≥2 is the embedding dimension, and S D for The set of all possible permutations of the column positions of the elements in the array. Let τ be the D-dimensional delay embedding vector of the input signal time series {x(k), k = 1, 2, ..., N} at time i, where τ is the time delay, i = 1, 2, ..., N - (D-1)τ, ∏ j= (j1,j2,…,jD) is the set S D A permutation that simultaneously satisfies the following two conditions:
[0068] ①x(i+(j1-1)T)≤x(i+(j2-1)τ)≤…≤x(i+(j D -1)τ);
[0069] ②If x(i+(j) s-1 –1)τ)=x(i+(j s -1)τ), then js-1≤j s ;
[0070] P(Π j ) for Π j The probability of a sequence of symbols is expressed as:
[0071]
[0072] In the formula, for China has Π j The number of permutations, 1≤j≤N-(D-1)τ
[0073] The permutation entropy of the corresponding time series is defined as:
[0074]
[0075] For the above equation, it is clear that 0 ≤ Hp ≤ ln(D!), because for a monotonic sequence, Hp = 0; while for a completely random sequence, since the probability of each symbol sequence appearing is 1 / ln(D!), its Hp = ln(D!). Therefore, for the convenience of statistical analysis, the above equation is normalized:
[0076]
[0077] As can be seen from the above definition, the permutation entropy Hp' can reflect the local characteristics of a time series (signal). The smaller Hp' is, the more regular the time series (signal) is.
[0078] Step S3: Based on composite indices, the optimal signal component is selected from the IMF components to achieve signal noise reduction.
[0079] The criteria for selecting the optimal signal component in step S3 are: the IMF component with the largest composite index CI value, and the permutation entropy Hp' ≤ 0.3 and mutual information MI ≥ 0.8.
[0080] Based on the fitness value f, adjust the individual optimal value pbest(t) and the global optimal value p g (t)
[0081] The velocity and position update formulas of the particle swarm optimization algorithm are as follows:
[0082]
[0083] In the formula, t represents the iteration number, vm(t) and xm(t) represent the velocity and position (δ,β) of particle m in the t-th iteration, respectively, w is the inertia weight factor, c1 and c2 are acceleration factors, and r1 and r2 are random numbers between [0,1].
[0084] After each iteration, VMD decomposition is performed on the updated particle position (δ,β), and the corresponding fitness value f is calculated. After the maximum number of iterations, the optimal particle position (δ,β) is output, and the optimal VMD denoising parameters δ and β are obtained. Finally, the denoising processing of the input signal is completed using these optimized parameters.
[0085] Step S4: Combine the multi-chip collaborative architecture to perform real-time harmonic analysis on the noise-reduced signal and output power quality parameters.
[0086] The multi-chip collaborative architecture consists of FPGA + ARM + DSP. The FPGA is used for high-speed signal acquisition and preprocessing, including signal filtering, amplification, and analog-to-digital conversion. The ARM is used to execute the PSO algorithm to optimize VMD parameters and schedule the data flow between the FPGA and DSP. The DSP is used to calculate harmonic parameters, including total harmonic distortion (THD), amplitude, and phase of each harmonic. The chips communicate and work together via PCIe or Gigabit Ethernet bus, with a data transmission latency of ≤1ms.
[0087] Step S5: Perform transient disturbance detection on the denoised signal and extract transient features through short-time Fourier transform (STFT), with a response time ≤10ms.
[0088] Transient disturbances include voltage spikes, drops, interruptions, and transient oscillations, with a detection accuracy error ≤2%.
[0089] The VMD denoising method optimized by the particle swarm optimization algorithm described above is used to denoise the acquired signal. The optimization results for VMD denoising are as follows: Figure 2 As shown in the figure, the left horizontal axis represents the number of iterations, the vertical axis represents the optimal position value in the PSO algorithm, the right horizontal axis represents the value of β, and the vertical axis represents the value of δ. The final optimized particle position is (4.78036, 2468), which is approximately taken as δ = 5 and β = 2468.
[0090] The signal obtained by decomposing based on the above VMD noise reduction method is as follows: Figure 3 As shown in the figure, the input signal is decomposed into 5 IMF components, among which IMF2 is the optimal signal component. It can be clearly seen that the noise is reduced and the RSR wave signal can be clearly seen.
[0091] A power quality signal noise reduction and optimization detection system, comprising:
[0092] The signal acquisition module is used to acquire power quality signals in real time.
[0093] The parameter optimization module is used to optimize VMD parameters;
[0094] The noise reduction module is used to perform VMD decomposition and component selection on the signal based on optimized parameters.
[0095] The harmonic analysis module is used to perform real-time harmonic detection on the noise-reduced signal and output power quality parameters.
[0096] The transient detection module is used to identify voltage surges, drops, interruptions, and transient oscillations.
[0097] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments for application in other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A method for power quality signal noise reduction and optimization detection, characterized in that: Specifically, the following steps are included: Step S1: Dynamically optimize the penalty factor δ and the number of modes β of variational mode decomposition using the particle swarm optimization algorithm, construct a fitness function with composite index as the core, and screen the optimal decomposition parameters; Step S2: Perform VMD decomposition on the power quality signal based on the optimized decomposition parameters δ and β to obtain multiple intrinsic mode function components, i.e., IMF components; Step S3: Based on composite indices, the optimal signal component is selected from the IMF components to achieve signal noise reduction.
2. The power quality signal noise reduction and optimization detection method as described in claim 1, characterized in that: The parameters of the particle swarm optimization algorithm in step S1 are set as follows: The initial population size χ = 20, the number of iterations ψ = 50, the initial range of the penalty factor δ is [0, 6], and the initial range of the number of modes β is [0, 5000]. The particle position (δ,β) is constructed using the parameter penalty factor δ and the number of modes β.
3. The power quality signal noise reduction and optimization detection method as described in claim 1, characterized in that: Step S2 specifically includes the following steps: In each iteration of the particle swarm optimization algorithm: Based on the positions (δ, β) of each particle, the signal is decomposed using VMD, and the corresponding fitness value f is calculated. The fitness function is defined as follows: CI=MI / H′ p Where: CI is a composite index of permutation entropy Hp' and mutual information MI, mi(A,B)=H(B)—H(B|A), A and B represent two different events, H(B) is the entropy of B, H(A) is the entropy of A, and H(B|A) is the conditional entropy of B when A is known; A is the original input signal, and B is the signal after adding the IMF components after VMD decomposition; In the formula, D≥2 is the embedding dimension, and S D for The set of all possible permutations of the column positions of the elements in the array. Let τ be the D-dimensional delay embedding vector of the input signal time series {x(k), k = 1, 2, ..., N} at time i, where τ is the time delay, i = 1, 2, ..., N - (D-1)τ, and Πj = (j1, j2, ..., jD) be the set S. D A permutation that simultaneously satisfies the following two conditions: ①x(i+(j1-1)τ)≤x(i+(j2-1)τ)≤…≤x(i+(j D -1)t); ②If x(i+(j) s-1 -1)τ)=x(i+(j s -1)τ), then j s-1 ≤j s ; P(Π j ) for Π j The probability of a symbol sequence is expressed as: In the formula, for China has Π j The number of permutations, 1≤j≤N-(D-1)τ The permutation entropy of the corresponding time series is defined as: Normalize the above expression:
4. The power quality signal noise reduction and optimization detection method as described in claim 3, characterized in that: Based on the fitness value f, adjust the individual optimal value pbest(t) and the global optimal value p g (t) The velocity and position update formulas of the particle swarm optimization algorithm are as follows: In the formula, t represents the iteration number, and vm(t) and xm(t) represent the velocity and position (δ, β) of particle m in the t-th iteration, respectively. ) w is the inertia weighting factor, c1 and c2 are acceleration factors, and r1 and r2 are random numbers between [0,1]. After each iteration, VMD decomposition is performed on the updated particle position (δ,β), and the corresponding fitness value f is calculated. After the maximum number of iterations, the optimal particle position (δ,β) is output, and the optimal VMD denoising parameters δ and β are obtained. Finally, the denoising processing of the input signal is completed using these optimized parameters.
5. The power quality signal noise reduction and optimization detection method as described in claim 1, characterized in that: Also includes: Step S4: Combine the multi-chip collaborative architecture to perform real-time harmonic analysis on the noise-reduced signal and output power quality parameters.
6. The power quality signal noise reduction and optimization detection method as described in claim 5, characterized in that: The multi-chip collaborative architecture is FPGA+ARM+DSP.
7. A power quality signal noise reduction and optimization detection method as described in claim 1 or 5, characterized in that: Also includes: Step S5: Perform transient disturbance detection on the denoised signal and extract transient features through short-time Fourier transform (STFT), with a response time ≤10ms.
8. A power quality signal noise reduction and optimization detection system, characterized in that: The power quality signal noise reduction and optimization detection method as described in any one of claims 1-7 includes: The signal acquisition module is used to acquire power quality signals in real time. The parameter optimization module is used to optimize VMD parameters; The noise reduction module is used to perform VMD decomposition and component selection on the signal based on optimized parameters. The harmonic analysis module is used to perform real-time harmonic detection on the noise-reduced signal and output power quality parameters. The transient detection module is used to identify voltage surges, drops, interruptions, and transient oscillations.
9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.
10. The power quality signal noise reduction and optimization detection method as described in any one of claims 1-7 is applied to online power quality monitoring and fault diagnosis in microgrids, rail transit power supply systems, new energy power plants, and industrial power distribution systems.