Method and apparatus for improving performance of filter

Abstract

The present invention relates to the technical field of power system automation, and specifically relates to a method and apparatus for improving the performance of a filter. An adaptive filter based on an LMS algorithm is designed and improved, the input of the adaptive filter is set as an original signal, and the output of the filter is set as a signal after a harmonic component has been eliminated. In order to realize adaptive filtering, a model and an error function for the filter need to be established, and coefficients of the filter are updated by means of continuous iteration. After multiple iterations, the coefficients of the filter gradually tend to be stable, thereby eliminating a harmonic signal. In the design of the present invention, a more effective constraint function and step factor relationship model is first constructed, and the weight of a filter is adjusted by means of continuous iteration, thereby realizing adaptive filtering; and the filter has higher steady-state precision and a faster response speed, such that the real-time performance and accuracy of harmonic detection can be ensured, and the steady-state precision of the filter is higher under the condition that the response speeds are approximately the same.

Description

Method and device for improving filter performance Technical Field

[0001] The present invention relates to the technical field of power system automation, and in particular to a method and device for improving filter performance. Background Art

[0002] With the increasing application of electric energy, harmonics in power systems are becoming increasingly prominent. Harmonics are a common phenomenon in power systems, primarily due to the extensive use of power electronic devices and equipment with nonlinear electrical properties. Because harmonic signals in power systems have highly frequency-harmonic characteristics, they can affect the transmission and quality of power and, in severe cases, even cause power accidents. Therefore, research on harmonic control in power systems is particularly important. To effectively control system harmonics, appropriate filters must be designed. Research on filter design focuses primarily on response speed and steady-state accuracy.

[0003] Initially, practical applications typically employed "passive" filters consisting of reactors and capacitors to reduce harmonics. However, these filters are typically only suitable for static compensation. When the source impedance changes with system structure and operating conditions, their filtering performance is severely affected. Furthermore, series or parallel resonance between the passive filter and the source impedance is prone to occur. To overcome these issues, Yasuhiko Yasuno et al. in Japan pioneered an "active" harmonic compensator, achieving a compensation rate exceeding 80%. In actual distribution systems, active filters often experience unbalanced currents caused by unbalanced loads due to transient fluctuations. Later, Luo Shiguo et al. from Chongqing University proposed an adaptive detection method that largely overcomes the shortcomings of traditional detection methods, but suffers from slow dynamic response speed. To address the low steady-state accuracy and slow response speed of low-pass filters in traditional detection methods based on instantaneous reactive power, researchers introduced adaptive analysis theory and constructed adaptive filters to replace LPFs. This method significantly improves detection accuracy and response speed.

[0004] However, because the aforementioned methods all use a fixed-step least mean square algorithm when designing filters, the step size factor is fixed to a constant, resulting in limited adjustment flexibility. The algorithm requires a large number of steps, leading to slow system response. The power fluctuates significantly around the target value, significantly deviating from the expected value, and thus failing to meet the required steady-state accuracy. To address this issue, we propose a method and device to improve filter performance.

[0005] Summary of the Invention

[0006] The object of the present invention is to provide a method and device for improving filter performance to solve the problems raised in the above background technology.

[0007] To solve the above technical problems, one of the objectives of the present invention is to provide a method for improving filter performance, comprising the following steps:

[0008] S1. Detect three-phase harmonic current: through the instantaneous reactive power theory p -i q The harmonic detection method detects the harmonic current in the three-phase circuit;

[0009] S2. Calculate harmonic components: Calculate the voltage and current using moment theory to obtain the harmonic components in the current.

[0010] S3. Design and improve the adaptive filter based on the LMS algorithm, including:

[0011] S3.1. Set the input of the adaptive filter to the original signal;

[0012] S3.2. Set the output of the adaptive filter to the signal after eliminating harmonic components;

[0013] S3.3. Establish an adaptive filter model and error function, and update the coefficients of the adaptive filter through continuous iteration;

[0014] S3.4, calculate the instantaneous gradient;

[0015] S4. Construct constraint functions to ensure the stability of the algorithm;

[0016] S5. Filtering is performed according to the designed adaptive filter to obtain a filtering result.

[0017] As a further improvement of the present technical solution, in step S2, calculating the harmonic component specifically includes the following steps:

[0018] S2.1. Calculate active current; active current i p with i q The calculation formula is:

[0019] Where, i a 、i b 、i c is the three-phase current directly obtained through detection;

[0020] S2.2, get the three-phase fundamental current: calculate the active current i p with i q After filtering with a low-pass filter, its DC component is obtained and Then, the three-phase fundamental current i can be obtained by calculating the inverse transformation matrix.af 、i bf 、i cf :

[0021] Where C 23 C 32 The transposed matrix of C -1 is the inverse matrix of C;

[0022] S2.3, get three-phase harmonic current: use three-phase current i a 、i b 、i c Subtracting the corresponding fundamental current can get the three-phase harmonic current i ah 、i bh 、i ch :

[0023] Where i ah 、i bh 、i ch is the three-phase harmonic current.

[0024] As a further improvement of the present technical solution, in step S3.1, the input of the adaptive filter is set to the original signal, and the signal vector X(t) is expressed as: X(t) = [x1(t), x2(t), ..., x n-1 (t),x n (t)] T

[0025] Where x(t) represents the input signal of the adaptive filter;

[0026] The corresponding weight coefficient vector W(t) is expressed as: W(t)=[w1(t),w2(t),...,w n-1 (t),w n (t)] T

[0027] Among them, w(t) represents the weight signal.

[0028] As a further improvement of the present technical solution, in step S3.2, the output of the adaptive filter is set to be the signal after the harmonic components are eliminated, and the value of the output signal y(t) is:

[0029] Where i represents the i-th signal.

[0030] As a further improvement of the present technical solution, in the step S3.3, when establishing the model and error function of the adaptive filter, the error is first generated by calculating the difference between the reference signal and the actual signal; at the same time, for the selection of the adaptive filter structure, a transverse FIR (Finite Impulse Response) structure is adopted; then, an improved epicycloid function is used to adjust and improve the variable step size factor of the variable step size LMS algorithm, so that it can effectively overcome the shortcomings of the traditional LMS algorithm; the step size factor is dynamically adjusted by the change in the size of the filter output error; and the iterative relationship of the new LMS adaptive algorithm is obtained through the obtained filter parameters and algorithm relationship; after multiple iterations, the filter coefficients will gradually stabilize, thereby achieving the elimination of harmonic signals.

[0031] As a further improvement of the present technical solution, the step size factor is adjusted using an improved epicycloid curve function, thereby designing an LMS adaptive filter, which uses an epicycloid curve function to combine the step size factor and the error function, and the expression is:

[0032] Where μ(t) represents the epicycloidal function, and the parameter β is used to control the value range of the function, that is, the upper and lower limits of the output of the function; the parameters α and δ are used to control the shape change of the function, that is, the curvature of the function and enhance the robustness of the filter algorithm.

[0033] As a further improvement of the present technical solution, in step S3.3, the error function calculation expression is as follows: e(t) = d(t) - y(t) = d(t) - W T (txt)

[0034] Where d(t) represents the expected signal value of the system; e(t) represents the error signal, whose value is the difference between the expected signal and the output signal;

[0035] The mean square error is further obtained as: J=E[e 2 (t)]=E[d 2 (t)]-2E[d(n)W T (n)X(n)]+E[W T (n)X(n)X T (n)W(n)]

[0036] Where J represents the mean square error of the error function; E[·] represents the mean square error.

[0037] As a further improvement of the present technical solution, in step S3.4, the instantaneous gradient is calculated by using the square of the error obtained by one sampling to obtain the instantaneous gradient value under the LMS algorithm, including: using the steepest descent algorithm to obtain the weighted iterative expression as follows:

[0038] in, Represents the instantaneous gradient value under the LMS algorithm;

[0039] Use e 2 (t) replace [e 2 (t)] calculates the instantaneous gradient value; the instantaneous gradient value under the LMS algorithm for:

[0040] Furthermore, the iterative relationship of the LMS adaptive algorithm is as follows:

[0041] Where μ represents the step size factor, which is a positive real number.

[0042] As a further improvement of the present technical solution, in step S4, the specific expression of the constraint function is:

[0043] Among them, μ represents the step size factor, which is a positive real number. max is chosen to be a positive number not greater than 1, μ min Satisfy the convergence speed and ensure the stability of the algorithm; in the above formula, the upper limit of the step size factor μ is β.

[0044] A second object of the present invention is to provide a device for improving filter performance, which is used to implement the steps of the above-mentioned method for improving filter performance, including a harmonic current detection module, a harmonic component calculation module, an adaptive filtering module, an algorithm constraint module, and a result verification module that are communicatively connected in sequence; wherein:

[0045] The harmonic current detection module is used to detect the instantaneous reactive power through the theory of i p -i q The harmonic detection method detects the harmonic current in the three-phase circuit;

[0046] The harmonic component calculation module is used to obtain the harmonic components in the current by performing moment theory calculation on the voltage and current;

[0047] The adaptive filtering module is used to adjust the step size factor using an improved epicycloid function to design an LMS adaptive filter, set the input of the adaptive filter to be the original signal and the output to be the signal after eliminating harmonic components, establish a model and error function of the adaptive filter, and update the coefficients of the adaptive filter through continuous iteration;

[0048] The algorithm constraint module is used to construct a constraint function to ensure the stability of the algorithm;

[0049] The result verification module is used to perform filtering according to the designed adaptive filter to obtain filtering results, and compare them with traditional adaptive low-pass filters and variable step-size adaptive low-pass filters of Lorentzian functions to verify the performance of the aforementioned LMS adaptive filter.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. In this method and device for improving filter performance, a more effective constraint function and step-size factor relationship model is first constructed. Adaptive filtering is achieved by continuously iteratively adjusting the filter weights. An improved epicycloid function is introduced to achieve continuous adjustment of the step-size factor and error signal, effectively improving the steady-state accuracy and response speed of the filtering process. The designed adaptive filter is then compared with a traditional adaptive low-pass filter and a variable-step-size adaptive low-pass filter using a Lorentzian function. The filter has higher steady-state accuracy and response speed, ensuring real-time and accurate harmonic detection. Under conditions of roughly the same response speed, the filter achieves higher steady-state accuracy. Simulations verify the effectiveness of this approach.

[0052] 2. In the method and device for improving filter performance, the improved epicycloid function variable step size adaptive filter proposed in this scheme filters in i p -i q The harmonic detection algorithm has good practicality and application value, and can provide a new solution for harmonic detection in power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] FIG1 is a flow chart of an exemplary overall method of the present invention;

[0054] FIG2 is an exemplary instantaneous reactive power theory based on the present invention. p -i q Schematic diagram of the algorithm principle;

[0055] FIG3 is a schematic diagram of an exemplary adaptive filter based on the LMS algorithm in the present invention;

[0056] FIG4 is a schematic diagram of an LMS adaptive filter in an exemplary harmonic detection method of the present invention;

[0057] FIG5 is a schematic diagram of an exemplary LMS filter step size factor adjustment based on a Lorentzian function in the present invention;

[0058] FIG6 is a schematic diagram of an exemplary LMS filter step size factor adjustment based on an improved epicycloidal curve function in the present invention;

[0059] FIG. 7 is an exemplary diagram of i under different cutoff frequencies in the present invention. pSchematic diagram of the DC component;

[0060] FIG8 is an exemplary diagram of i under different filters in the present invention. p -i q Phase A fundamental current diagram detected by the harmonic detection method. DETAILED DESCRIPTION

[0061] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0062] Example 1

[0063] As shown in FIG1 to FIG8 , this embodiment provides a method for improving filter performance, including the following steps:

[0064] S1. Detect three-phase harmonic current: through the instantaneous reactive power theory p -i q The harmonic detection method can detect the harmonic current in the three-phase circuit;

[0065] This step specifically includes: first, using a phase locked loop (PLL) to extract the sine and cosine signals with the same phase and frequency as the phase A voltage, and then obtaining the active current i according to the definition of instantaneous reactive power theory. p with i q , as shown in Figure 2.

[0066] S2. Calculate harmonic components: Calculate the voltage and current using moment theory to obtain the harmonic components in the current.

[0067] In this step, calculating the harmonic components specifically includes the following steps:

[0068] S2.1. Calculate active current; active current i p with i q The calculation formula is:

[0069] Where, i a 、i b 、i c is the three-phase current directly obtained through detection;

[0070] S2.2, get the three-phase fundamental current: calculate the active current i p with i qAfter filtering with a low-pass filter, its DC component is obtained and Then, the three-phase fundamental current i can be obtained by calculating the inverse transformation matrix. af 、i bf 、i cf :

[0071] Where C 23 C 32 The transposed matrix of C -1 is the inverse matrix of C;

[0072] S2.3, get three-phase harmonic current: use three-phase current i a 、i b 、i c Subtracting the corresponding fundamental current can get the three-phase harmonic current i ah 、i bh 、i ch :

[0073] Where i ah 、i bh 、i ch is the three-phase harmonic current.

[0074] Specifically, it can be seen from the above harmonic detection process that in order to obtain the three-phase fundamental current, a low-pass filter is needed to filter out high-frequency noise and high-frequency components in the harmonic signal. The filtering performance of the low-pass filter will have a significant impact on the steady-state accuracy and response speed of the entire harmonic current.

[0075] S3. Design and improve the adaptive filter based on the LMS algorithm: As shown in Figures 3-5, an improved adaptive filter based on the LMS algorithm is designed. The input of the adaptive filter is set as the original signal, and the output of the filter is the signal after the harmonic components are eliminated. To achieve adaptive filtering, it is necessary to establish a filter model and error function, and update the filter coefficients through continuous iteration. After multiple iterations, the filter coefficients will gradually stabilize, thereby achieving the elimination of harmonic signals. Specifically, it includes:

[0076] S3.1. Set the input of the adaptive filter to the original signal; the signal vector X(t) is expressed as: X(t) = [x1(t), x2(t), ..., x n-1 (t),x n (t)] T

[0077] Where x(t) represents the input signal of the adaptive filter;

[0078] The corresponding weight coefficient vector W(t) is expressed as: W(t)=[w1(t),w2(t),...,w n-1 (t),w n (t)] T

[0079] Among them, w(t) represents the weight signal.

[0080] S3.2. Set the output of the adaptive filter to the signal after eliminating the harmonic components; the value of the output signal y(t) is obtained from the above formula:

[0081] Where i represents the i-th signal.

[0082] S3.3. Establish an adaptive filter model and error function, and update the coefficients of the adaptive filter through continuous iteration;

[0083] In this step, when establishing the model and error function of the adaptive filter, the error is first calculated by the difference between the reference signal and the actual signal; at the same time, for the selection of the adaptive filter structure, the transverse FIR (Finite Impulse Response) structure is used for research, which has good filtering performance and real-time performance; then the variable step size factor of the variable step size LMS algorithm is adjusted and improved using the improved epicycloid curve function, so that it can effectively overcome the shortcomings of the traditional LMS algorithm; the step size factor is dynamically adjusted by the change in the size of the filter output error; and the iterative relationship of the new LMS adaptive algorithm is obtained through the obtained filter parameters and algorithm relationship; after multiple iterations, the coefficients of the filter will gradually stabilize, thereby achieving the elimination of harmonic signals.

[0084] Among them, the improved epicycloid function is used to adjust the step size factor, so as to design the LMS adaptive filter. As shown in Figure 6, compared with the Lorentz function, a larger step size factor can be obtained when the error is large. However, when the error is close to zero, the step size factor changes faster, resulting in poor stability of the algorithm and easy oscillation. The epicycloid function used in this scheme combines the step size factor and the error function, and the expression is:

[0085] Where μ(t) represents the epicycloidal function, and the parameter β is used to control the value range of the function, that is, the upper and lower limits of the output of the function; the parameters α and δ are used to control the shape change of the function, that is, the curvature of the function and enhance the robustness of the filter algorithm.

[0086] Furthermore, the error function calculation expression is as follows: e(t) = d(t) - y(t) = d(t) - W T (txt)

[0087] Where d(t) represents the expected signal value of the system; e(t) represents the error signal, whose value is the difference between the expected signal and the output signal;

[0088] The mean square error is further obtained as: J=E[e 2 (t)]=E[d 2 (t)]-2E[d(n)W T (n)X(n)]+E[W T (n)X(n)X T (n)W(n)]

[0089] Where J represents the mean square error of the error function; E[·] represents the mean square error.

[0090] S3.4. Calculate the instantaneous gradient. Specifically, use the square of the error obtained from a single sampling to obtain the instantaneous gradient value under the LMS algorithm, including: using the steepest descent algorithm to obtain the weighted iterative expression as:

[0091] in, Represents the instantaneous gradient value under the LMS algorithm;

[0092] Use e 2 (t) replace [e 2 (t)] calculates the instantaneous gradient value; thus, the instantaneous gradient value under the LMS algorithm is for:

[0093] Furthermore, the iterative relationship of the LMS adaptive algorithm is as follows:

[0094] Where μ represents the step size factor, which is a positive real number.

[0095] Furthermore, by comparing the step-size factor adjustment range with that of a variable-step-size adaptive filter based on the Lorentzian function, we found that as the error decreases, that is, in areas closer to the optimal value, the step-size factor corresponding to the Lorentzian function and the pre-improved epicycloidal function changes more rapidly, while the step-size factor of the improved epicycloidal function changes more slowly. Furthermore, in areas closer to the optimal value, the step-size factor corresponding to the improved epicycloidal function is also relatively small. This indicates that the improved epicycloidal function exhibits superior smoothness and continuity. Based on the above theoretical analysis, the improved epicycloidal function, which combines the step-size factor and the error signal, performs better than the Lorentzian function and the pre-improved standard epicycloidal function. Therefore, this solution adopts the improved epicycloidal function variable-step-size LMS adaptive filter for filtering.

[0096] S4. Construct a constraint function to ensure the stability of the algorithm. By comparing the step size factor adjustment range with that of the variable step size adaptive filter based on the Lorentzian function, as the error decreases, that is, in the area closer to the optimal value, the step size factor corresponding to the Lorentzian function and the epicycloid function before improvement changes faster, resulting in poor stability of the algorithm and prone to oscillation. Therefore, this embodiment ensures the stability of the algorithm by constructing a constraint function. The specific expression of the constraint function is:

[0097] Among them, μ represents the step size factor, which is a positive real number. max is chosen to be a positive number not greater than 1, μ min Satisfy the convergence speed and ensure the stability of the algorithm; in the above formula, the upper limit of the step size factor μ is β.

[0098] S5. Filtering is performed according to the designed adaptive filter to obtain a filtering result.

[0099] In order to verify the performance of the above LMS adaptive filter, we also provide a set of verification examples, as shown in Figures 7 and 8, which specifically include:

[0100] First, set the experimental environment: set the three-phase voltage in the power grid to 220V / 50HZ, and connect the load side to a three-phase uncontrolled rectifier bridge with an inductive load, where the load resistance R = 10Ω and L = 0.001H. In addition, take μ max =0.1, μ min =0.001. The harmonic current generated on the load side is the detection object.

[0101] Secondly, select an appropriate cutoff frequency. When using a low-pass filter for filtering, the filter's cutoff frequency affects the accuracy and response speed of harmonic detection. A smaller cutoff frequency ensures higher steady-state accuracy, but the response speed decreases. A larger cutoff frequency increases the response speed but reduces the steady-state accuracy. Therefore, it is necessary to conduct comparative studies using filters with different cutoff frequencies to determine the optimal cutoff frequency. As shown in Figure 8, a cutoff frequency of 20 Hz results in lower ripple, meaning the steady-state offset is minimized, but the response speed is slowest. A cutoff frequency of 40 Hz results in a faster response, but higher ripple, meaning the steady-state offset is more severe. A cutoff frequency of 30 Hz achieves both high response speed and steady-state accuracy.

[0102] Then, the fundamental current of phase A detected by the other filters is compared: the adaptive filter using the improved epicycloid function can obtain a smoother ideal fundamental current curve. Next, the harmonic content of the fundamental current of phase A detected by these four different filters is compared. When the improved epicycloid function variable step size adaptive filter is used for filtering, the harmonic content of the fundamental current is the lowest. This shows that the filtering effect of the adaptive filter designed in this scheme is better than that of the other three filters. It can be seen that the adaptive filter designed in this scheme makes the fundamental current curve smoother, closer to the standard sine wave, and has less harmonic content.

[0103] This embodiment further provides a device for improving filter performance, which is used to implement the steps of the above-mentioned method for improving filter performance, including a harmonic current detection module, a harmonic component calculation module, an adaptive filtering module, an algorithm constraint module, and a result verification module that are communicatively connected in sequence; wherein:

[0104] The harmonic current detection module is used to detect the instantaneous reactive power through the i p -i q The harmonic detection method detects the harmonic current in the three-phase circuit;

[0105] The harmonic component calculation module is used to obtain the harmonic components in the current by performing moment theory calculation on the voltage and current;

[0106] The adaptive filtering module is used to adjust the step size factor using the improved epicycloid function to design an LMS adaptive filter. The input of the adaptive filter is set to the original signal and the output is the signal after the harmonic components are eliminated. The model and error function of the adaptive filter are established, and the coefficients of the adaptive filter are updated through continuous iteration.

[0107] The algorithm constraint module is used to construct constraint functions to ensure the stability of the algorithm;

[0108] The result verification module is used to perform filtering according to the designed adaptive filter to obtain the filtering results, and compare them with the traditional adaptive low-pass filter and the variable step-size adaptive low-pass filter of the Lorentzian function to verify the performance of the aforementioned LMS adaptive filter.

[0109] Those skilled in the art will appreciate that the process of implementing all or part of the steps of the above embodiments may be accomplished by hardware, or by instructing related hardware through a program.

[0110] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for improving the performance of a filter, characterized in that, It includes the following steps: S1. Detect three-phase harmonic current: Detect the harmonic current in the three-phase circuit by using the i p -i q harmonic detection method of instantaneous reactive power theory; S2. Calculate the harmonic components: By performing moment theory calculations on voltage and current, the harmonic components in the current are obtained; S3. Design and improve an adaptive filter based on the LMS algorithm, specifically including: S3.

1. Set the input of the adaptive filter as the original signal; S3.

2. Set the output of the adaptive filter as the signal after eliminating the harmonic components; S3.

3. Establish the model and error function of the adaptive filter, and continuously update the coefficients of the adaptive filter through iteration; S3.

4. Calculate the instantaneous gradient; S4. Construct a constraint function to ensure the stability of the algorithm; S5. Filter according to the designed adaptive filter to obtain the filtering result.

2. The method and device for improving filter performance according to claim 1, wherein: In step S2, calculating the harmonic components specifically includes the following steps: S2.

1. Calculate the active current; the active current i p and i q The calculation formula is as follows: Wherein, i a 、i b 、i c are three-phase currents directly obtained through detection; S2.

2. Obtain the three-phase fundamental current: Filter the calculated active current \(i\) p and \(i\) q through a low-pass filter to obtain their DC components With The three-phase fundamental currents i af 、i bf 、i cf can be obtained through the calculation of the inverse transformation matrix: where C 23 is the transpose matrix of C 32 , that is C -1 is the inverse matrix of C; S2.

3. Obtain the three-phase harmonic currents: Use the three-phase currents i a 、i b 、i c Subtract their corresponding fundamental currents to obtain the three-phase harmonic currents i ah 、i bh 、i ch : where i ah , i bh , i ch are three-phase harmonic currents.

3. The method for improving the performance of a filter according to claim 1, characterized in that: In the step S3.1, the input of the adaptive filter is set as the original signal, and the signal vector X(t) is expressed as: X(t) = [x1(t), x2(t),..., x n-1 (t), x n (t)] T Among them, x(t) represents the input signal of the adaptive filter; The corresponding weight coefficient vector W(t) is expressed as: W(t) = [w1(t), w2(t),..., w n-1 (t), w n (t)] T Among them, w(t) represents the weight signal.

4. The method for improving the performance of a filter according to claim 3, characterized in that: In the step S3.2, set the output of the adaptive filter to the signal after eliminating the harmonic components, and the value of the output signal y(t) is: Among them, i represents the i-th signal.

5. The method for improving the performance of a filter according to claim 4, characterized in that: In step S3.3, when establishing the model and error function of the adaptive filter, first calculate the error by the difference between the reference signal and the actual signal; at the same time, for the selection of the adaptive filter structure, a transverse FIR structure is adopted; then use the improved outer pendulum curve function to adjust and improve the variable step size factor of the variable step size LMS algorithm; dynamically adjust the step size factor according to the change in the magnitude of the filter output error; and obtain the iterative relationship of the new LMS adaptive algorithm through the obtained filter parameters and algorithm relationship; After multiple iterations, the coefficients of the filter will gradually tend to be stable, thereby achieving the elimination of harmonic signals.

6. The method for improving the performance of a filter according to claim 5, characterized in that: The step size factor is adjusted using an improved outside-swing curve function to design an LMS adaptive filter. The outside-swing curve function used combines the step size factor and the error function, and the expression is as follows: In the formula, μ(t) represents the outer pendulum curve function, and the parameter β is used to control the value range of the function, that is, the output upper limit and output lower limit of the function; the parameters α and δ are used to control the shape change of the function, that is, the curvature of the function and enhance the robustness of the filter algorithm.

7. The method for improving the performance of a filter according to claim 6, wherein: In the step S3.3, the calculation expression of the error function is as follows: e(t) = d(t) - y(t) = d(t) - W T (t)X(t) Among them, d(t) represents the expected signal value of the system; e(t) represents the error signal, and its value is the difference between the expected signal and the output signal; Further, its mean square error value is obtained as: J = E[e 2 (t)] = E[d 2 (t)] - 2E[d(n)W T (n)X(n)] + E[W T (n)X(n)X T (n)W(n)] Among them, J represents the mean square error value of the error function; E[·] represents the mean square error value.

8. The method for improving the performance of a filter according to claim 7, wherein: In the step S3.4, calculating the instantaneous gradient specifically includes: obtaining the instantaneous gradient value under the LMS algorithm by using the error square obtained from one sampling, including: using the steepest descent algorithm to obtain the weight iteration expression as: Among them, Represents the instantaneous gradient value under the LMS algorithm; Use e 2 (t) to replace [e 2 (t)] to calculate the instantaneous gradient value; the instantaneous gradient value under the LMS algorithm is: Furthermore, the iterative relationship of the LMS adaptive algorithm is as follows: In the formula, μ represents the step size factor, and its value is a positive real number.

9. The method for improving the performance of a filter according to claim 8, characterized in that: In the step S4, the specific expression of the constraint function is as follows: Among them, μ represents the step size factor, and its value is a positive real number, μ max is selected as a positive number less than or equal to 1, μ min satisfies the convergence rate and ensures the stability requirements of its algorithm; in the above formula, the upper limit of the value of the step size factor μ is β.

10. A device for improving the performance of a filter, which is used to implement the steps of the method for improving the performance of a filter according to any one of claims 1-9, characterized in that: It includes a harmonic current detection module, a harmonic component calculation module, an adaptive filtering module, an algorithm constraint module, and a result verification module that are sequentially communicatively connected; among them: The harmonic current detection module is used to detect the harmonic current in the three-phase circuit by the i p -i q harmonic detection method based on the instantaneous reactive power theory; The harmonic component calculation module is used to obtain the harmonic components in the current by performing moment theory calculations on voltage and current; The adaptive filtering module is used to adjust the step size factor using the improved outer pendulum curve function Thereby designing an LMS adaptive filter, setting the input of the adaptive filter as the original signal, the output as the signal after eliminating the harmonic components, establishing the model and error function of the adaptive filter, and continuously updating the coefficients of the adaptive filter through iteration; The algorithm constraint module is used to construct a constraint function to ensure the stability of the algorithm; The result verification module is used to filter according to the designed adaptive filter, obtain the filtering result, and compare it with the traditional adaptive low-pass filter and the variable-step-size adaptive low-pass filter of the Lorentzian function to verify the performance of the aforementioned LMS adaptive filter.

Images