Active high-frequency filter optimal configuration method and system

Abstract

The invention provides an active high-frequency filter optimal configuration method and system, and relates to the technical field of high-frequency filter configuration optimizing.The specific steps include: under different configuration parameter combinations, collecting power operation data, including power quality parameters and system performance parameters, of a power distribution network in real time, constructing a power operation data prediction model, and predicting the power operation data; performing training by taking configuration parameters as input and electric energy quality and system performance parameters as labels, randomly combining to generate an initial population and inputting the initial population into a prediction model to obtain related parameters, establishing a multi-term function relational expression, comprehensively evaluating high efficiency and stability of electric power operation, performing iterative optimization on the initial population, and extracting optimal configuration parameters. According to the method, the influence of different configuration combinations on the electric energy quality and the system performance can be systematically analyzed and evaluated, and a scientific basis and an evaluation standard are provided for optimization of configuration parameters, so that the operation efficiency and the stability of a power system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of high-frequency filter configuration optimization, and in particular to an active high-frequency filter optimization configuration method and system. Background Art

[0002] Active high-frequency filters are widely used in communications, electronic equipment, power systems and other fields. Their main function is to effectively suppress interference signals and improve signal quality. In practical applications, traditional filters often have problems such as insufficient gain, uneven frequency response, and phase delay, which directly affect the performance of the system and the integrity of the signal. Therefore, it is particularly important to optimize the configuration method of active high-frequency filters.

[0003] In the prior art, a method for optimizing the configuration of an active filter provided by publication number CN106786581A comprises the following steps: setting an objective function based on the influence of the active filter configuration; setting constraint conditions that satisfy the optimal configuration of the active filter of the smart distribution network; and optimizing the configuration of the active filter of the smart distribution network by using a configuration algorithm that combines modal analysis with a genetic algorithm. The method for optimizing the configuration of an active filter provided by the method introduces a modal analysis method to determine the candidate position nodes of the active filter, thereby reducing the amount of calculation when the genetic algorithm determines the configuration position of the active filter, and solving the problem that the genetic algorithm is prone to fall into a local optimum. Moreover, the method has a fast operation speed, which is conducive to the rapid convergence of the algorithm, and can find a better optimal solution, and can effectively calculate the optimal access position of the active filter, which plays a good role in improving the economy of system operation and improving the power quality.

[0004] However, there are still some deficiencies. As can be seen from the above statements, the existing technology lacks analysis of the functional relationship between power quality and system performance parameters, which makes it difficult to understand the interaction between the parameters during the optimization process, and then it is difficult to quantify the impact of different parameter combinations, which limits the accuracy of the comprehensive evaluation. It has not yet established a clear comprehensive evaluation standard, resulting in insufficient comparability of evaluation results in different optimization scenarios.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The object of the present invention is to provide an active high frequency filter optimization configuration method and system to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for optimizing the configuration of an active high-frequency filter, the specific steps comprising:

[0009] S1. Obtain the power operation data of the distribution network under different configuration parameter combinations of active high-frequency filters. The power operation data of the distribution network is the average of the power operation data of each node in the distribution network. The power operation data includes power quality parameters and system performance parameters. The power quality parameters include harmonic voltage content rate, harmonic current content rate, harmonic distortion rate and power factor. The system performance parameters include input signal amplitude, power loss and dynamic response time. The configuration parameters include gain, cutoff frequency, bandwidth and harmonic suppression rate.

[0010] S2. Construct a power operation data prediction model, take the configuration parameter combinations of different active high-frequency filters as input, and the power quality parameters and system performance parameters of the distribution network as label training models to train the power operation data prediction;

[0011] S3. Establishing constraints on configuration parameters, randomly combining configuration parameters under the constraints on configuration parameters, constructing individuals of the initial population of configuration parameters, inputting individuals of the initial population of configuration parameters into the power operation data prediction model, and obtaining power quality parameters and system performance parameters;

[0012] S4. Construct the functional relationship between harmonic voltage content rate, harmonic current content rate, harmonic distortion rate, power factor and power quality index, construct the functional relationship between input signal amplitude, power loss and dynamic response time and system performance index, construct the functional relationship between power quality index, system performance index and comprehensive evaluation index, and the comprehensive evaluation index is used to comprehensively evaluate the high efficiency and stability of power operation;

[0013] S5. Taking the maximization of the comprehensive evaluation index as the objective function, the individuals of the initial population of configuration parameters are iteratively optimized through a genetic algorithm under the constraints of the configuration parameters to obtain the optimal individuals, and based on the optimal individuals, the optimal values ​​of the configuration parameters are extracted.

[0014] Furthermore, the configuration parameters are randomly combined to construct individuals of the initial population of configuration parameters. The specific process is as follows:

[0015] The initial population is labeled as Q, and the initial population Q = uQ 1 ,Q 2 ,…,Q j ,…,Q m}, Q j is the jth individual in the initial population, j is the index of the individual in the initial population, and j∈[1,m], m is the number of individuals in the initial population, Q j = {G j ,f j ,BWj ,H j}, where G j ,f j ,BW j ,H j are the gain, cutoff frequency, bandwidth and harmonic suppression rate of the jth individual respectively.

[0016] Furthermore, the power operation data prediction model is composed of a deep learning network based on a multilayer perceptron, and the deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer, and the first hidden layer, the second hidden layer and the third hidden layer each have at least two neurons, and each uses ReLU as an activation function;

[0017] The process of training the power operation data prediction model is as follows:

[0018] The configuration parameter combinations of different active high-frequency filters are used as input, and the power quality parameters and system performance parameters of the distribution network are used as output labels for training. The mean square error is used as the loss function. When the mean square error is in the range of [0, 0.01], the training of the power operation data prediction model is completed.

[0019] Furthermore, a functional relationship between the harmonic voltage content rate, the harmonic current content rate, the harmonic distortion rate, the power factor and the power quality index is constructed as follows:

[0020]

[0021] Among them, PQI j is the power quality index of the jth individual, TV j is the harmonic voltage content rate of the jth individual, TI j is the harmonic current content of the jth individual, TH j is the harmonic distortion rate of the jth individual, PF j is the power factor of the jth individual, α 1 is the weight coefficient of harmonic voltage content, α 2 is the weight coefficient of harmonic current content, α 3 is the weight coefficient of harmonic distortion rate, α 4 is the weight coefficient of the power factor, α 3 +α 2 +α 1 +α 4 =1, and 0<α 3 <α 2 <α 1 <α 4 <1.

[0022] Furthermore, a functional relationship between input signal amplitude, power loss, dynamic response time and system performance index is constructed as follows:

[0023]

[0024] Among them, SZ j is the system performance index of the jth individual, A j is the input signal amplitude of the jth individual, P j is the power loss of the jth individual, T j is the dynamic response time of the jth individual, β 1 is the weight coefficient of the input signal amplitude, β 2 is the weight coefficient of power loss, β 3 is the weight coefficient of dynamic response time, β 1 +β 2 +β 3 =1, and 0<β 3 <β 2 <β 1 <1.

[0025] Furthermore, the functional relationship between the power quality index, system performance index and comprehensive evaluation index is constructed as follows:

[0026] ZGar j =γ 1 ·PQI j +γ 2 ·SZ j

[0027] Among them, ZPzs j is the comprehensive evaluation index, γ 1 and γ 2 are the weights in the calculation of power quality index and system performance index, and γ 1 and γ 2 The specific value of is determined by the hierarchical analysis method.

[0028] Furthermore, the specific process of step S5 is as follows:

[0029] Iterative optimization is performed on the individuals of the initial population of configuration parameters. During the iterative optimization process, the constraints of the configuration parameters are set, that is, the maximum and minimum values ​​of the gain, cutoff frequency, bandwidth and harmonic suppression rate are set respectively. Within the constraints of the gain, cutoff frequency, bandwidth and harmonic suppression rate, the configuration parameters of the active high-frequency filter are iteratively optimized. Specifically, the comprehensive evaluation index ZPzs is set to j Sort from large to small and select the comprehensive evaluation index ZPzs jThe individuals in the front row are taken as the parents. Through crossover and mutation operations, the genes of the parent individuals are exchanged, combined and mutated to generate new individuals. The power quality parameters and system performance parameters of the newly generated individuals are obtained by using the power operation data prediction model, and their comprehensive evaluation index is calculated. The new individuals and the parents are taken as the new population, and the selection, crossover and mutation operations are repeated until the predetermined number of iterations is reached. The individual corresponding to the maximum value of the comprehensive evaluation index is taken as the optimal combination of the configuration parameters of the active high-frequency filter, and the individual corresponding to the maximum value of the comprehensive evaluation index is calibrated as Q j1 = {G j1 ,f j1 ,BW j1 ,H j1}, then the configuration parameter combination of the optimal active high-frequency filter is gain G j1 , cut-off frequency f j1 , bandwidth BW j1 and harmonic suppression rate H j1 .

[0030] An active high-frequency filter optimization configuration system, the system is used to execute any one of the above-mentioned active high-frequency filter optimization configuration methods, comprising:

[0031] The data acquisition module is used to obtain the power operation data of the distribution network under different configuration parameter combinations of the active high-frequency filter. The power operation data of the distribution network is the average value of the power operation data of each node in the distribution network. The power operation data includes power quality parameters and system performance parameters. The power quality parameters include harmonic voltage content rate, harmonic current content rate, harmonic distortion rate and power factor. The system performance parameters include input signal amplitude, power loss and dynamic response time. The configuration parameters include gain, cutoff frequency, bandwidth and harmonic suppression rate.

[0032] The prediction model building module is used to build a power operation data prediction model, taking the configuration parameter combinations of different active high-frequency filters as input, and the power quality parameters and system performance parameters of the distribution network as label training models to train the power operation data prediction;

[0033] An initial population construction module is used to establish constraints on configuration parameters. Under the constraints on configuration parameters, the configuration parameters are randomly combined to construct individuals of the initial population of configuration parameters. The individuals of the initial population of configuration parameters are input into the power operation data prediction model to obtain power quality parameters and system performance parameters.

[0034] The data processing and analysis module is used to construct the functional relationship between the harmonic voltage content rate, the harmonic current content rate, the harmonic distortion rate, the power factor and the power quality index, construct the functional relationship between the input signal amplitude, the power loss and the dynamic response time and the system performance index, and construct the functional relationship between the power quality index, the system performance index and the comprehensive evaluation index. The comprehensive evaluation index is used to comprehensively evaluate the high efficiency and stability of power operation;

[0035] The parameter optimization module is used to maximize the comprehensive evaluation index as the objective function, iteratively optimize the individuals of the initial population of configuration parameters through a genetic algorithm under the constraints of the configuration parameters, obtain the optimal individuals, and extract the optimal values ​​of the configuration parameters based on the optimal individuals.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention constructs a power operation data prediction model based on the acquisition of real-time data, takes the configuration parameter combinations of different active high-frequency filters as input, and takes the power quality parameters and system performance parameters as labels for training. Through the model, the influence of different configuration combinations on the power quality and system performance can be systematically analyzed and evaluated. On this basis, the functional relationship between the harmonic voltage content rate, the harmonic current content rate, the harmonic distortion rate, the power factor and the power quality index is established, and the functional relationship between the input signal amplitude, power loss, dynamic response time and the system performance index is constructed. These functional relationships make the relationship between the power quality index, the system performance index and the comprehensive evaluation index clearer, provide a scientific basis and evaluation standard for the optimization of the configuration parameters, thereby improving the operation efficiency and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0039] Figure 2 This is a block diagram of the module composition of the present invention. DETAILED DESCRIPTION

[0040] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0041] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0042] Embodiment 1:

[0043] See also Figure 1 , the present invention provides a technical solution:

[0044] An active high-frequency filter optimization configuration method, the specific steps include:

[0045] S1. Obtain the power operation data of the distribution network under different configuration parameter combinations of active high-frequency filters. The power operation data of the distribution network is the average of the power operation data of each node in the distribution network. The power operation data includes power quality parameters and system performance parameters. The power quality parameters include harmonic voltage content rate, harmonic current content rate, harmonic distortion rate and power factor. The system performance parameters include input signal amplitude, power loss and dynamic response time. The configuration parameters include gain, cutoff frequency, bandwidth and harmonic suppression rate.

[0046] S2. Construct a power operation data prediction model, take the configuration parameter combinations of different active high-frequency filters as input, and the power quality parameters and system performance parameters of the distribution network as label training models to train the power operation data prediction;

[0047] S3. Establishing constraints on configuration parameters, randomly combining configuration parameters under the constraints on configuration parameters, constructing individuals of the initial population of configuration parameters, inputting individuals of the initial population of configuration parameters into the power operation data prediction model, and obtaining power quality parameters and system performance parameters;

[0048] S4. Construct the functional relationship between harmonic voltage content rate, harmonic current content rate, harmonic distortion rate, power factor and power quality index, construct the functional relationship between input signal amplitude, power loss and dynamic response time and system performance index, construct the functional relationship between power quality index, system performance index and comprehensive evaluation index, and the comprehensive evaluation index is used to comprehensively evaluate the high efficiency and stability of power operation;

[0049] S5. Taking the maximization of the comprehensive evaluation index as the objective function, the individuals of the initial population of configuration parameters are iteratively optimized through a genetic algorithm under the constraints of the configuration parameters to obtain the optimal individuals, and based on the optimal individuals, the optimal values ​​of the configuration parameters are extracted.

[0050] On the basis of the above embodiment, the voltage signal is collected in real time by a harmonic analyzer, a discrete Fourier transform is performed on the voltage signal, the amplitude of each frequency component is obtained, and the ratio of the total amplitude of the harmonic voltage to the amplitude of the fundamental voltage is calculated to obtain the harmonic voltage content rate;

[0051] The current signal is collected in real time by a harmonic analyzer, and a discrete Fourier transform is performed on the current signal to obtain the amplitude of each frequency component, and the ratio of the total amplitude of the harmonic current to the amplitude of the fundamental current is calculated to obtain the harmonic current content rate;

[0052] The voltage or current waveform is collected in real time by a harmonic analyzer, and the voltage or current waveform is analyzed by Fourier transform to calculate the ratio of the harmonic component to the fundamental component to obtain the harmonic distortion rate;

[0053] The power analyzer measures the voltage, current and power values ​​simultaneously, and calculates the ratio of active power to reactive power to obtain the power factor;

[0054] The voltage or current amplitude of the input signal is directly measured by a voltage sensor, and the voltage or current amplitude of the input signal is used as the input signal amplitude;

[0055] The power loss is determined by measuring the current and voltage with a power analyzer and then calculating the product of active power and power factor;

[0056] The dynamic response time is evaluated by applying a step input signal and measuring the time delay of the system output response via a signal generator.

[0057] On the basis of the above embodiment, after collecting the harmonic voltage content rate, harmonic current content rate, harmonic distortion rate, power factor, input signal amplitude, power loss and dynamic response time, the above parameters are respectively subjected to maximum-minimum normalization processing, and then the normalized data is used for subsequent analysis and processing, so that in the subsequent analysis and processing process, various data can be analyzed and processed under the same dimension, avoiding the problem of some data being neglected due to different dimensions.

[0058] Among them, the harmonic analyzer is used to collect the harmonic voltage content rate, harmonic current content rate and harmonic distortion rate, the power analyzer is used to collect the power factor and power loss, the voltage sensor is used to collect the input signal amplitude, and the signal generator is used to collect the dynamic response time. The harmonic analyzer, power analyzer, voltage sensor and signal generator can all use models in existing equipment and are not restricted here.

[0059] Harmonic analyzers, power analyzers, voltage sensors, and signal generators are all in multiple groups (such as 3 groups). Harmonic voltage content, harmonic current content, harmonic distortion rate, power factor, input signal amplitude, power loss, and dynamic response time are detected at different nodes. The same data detected from different nodes are averaged, and the final average is used as the corresponding data in the power quality parameters and system performance parameters to avoid accidental errors in individual points.

[0060] On the basis of the above embodiment, the configuration parameters are randomly combined to construct individuals of the initial population of configuration parameters. The specific process is as follows:

[0061] The initial population is labeled as Q, and the initial population Q = {Q 1 ,Q 2 ,…,Q j ,…,Q m}, Q j is the jth individual in the initial population, j is the index of the individual in the initial population, and j∈[1,m], m is the number of individuals in the initial population, Q j = {G j ,f j ,BW j ,H j}, where G j ,f j ,BW j ,H j are the gain, cutoff frequency, bandwidth and harmonic suppression rate of the jth individual respectively.

[0062] On the basis of the above embodiment, the power operation data prediction model is composed of a deep learning network based on a multilayer perceptron, and the deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer, and the first hidden layer, the second hidden layer and the third hidden layer each have at least two neurons, and each uses ReLU as an activation function;

[0063] In this embodiment, the input features of the deep learning network of the multilayer perceptron include four features: gain, cutoff frequency, bandwidth, and harmonic suppression rate.

[0064] The structure of the deep learning network of multi-layer perceptron is:

[0065] Input layer: receives input of 4 features;

[0066] The first hidden layer has 64 neurons and uses ReLU as the activation function.

[0067] The second hidden layer has 32 neurons and also uses the ReLU activation function.

[0068] The third hidden layer has 16 neurons and uses the ReLU activation function.

[0069] Output layer: has 2 neurons, power quality parameters and system performance parameters.

[0070] The process of training the power operation data prediction model is as follows:

[0071] The configuration parameter combinations of different active high-frequency filters are used as input, and the power quality parameters and system performance parameters of the distribution network are used as output labels for training. The mean square error is used as the loss function. When the mean square error is in the range of [0, 0.01], the training of the power operation data prediction model is completed.

[0072] Based on the above embodiment, the correlation between the harmonic voltage content rate, the harmonic current content rate, the harmonic distortion rate, the power factor and the power quality index is as follows:

[0073] The harmonic voltage content rate is negatively correlated with the power quality index. The increase in the harmonic voltage content rate generates a high harmonic current content rate, which also means that the harmonic pollution in the power grid is aggravated. The high harmonic voltage content rate indicates that there are more harmonic components in the voltage waveform, which leads to voltage waveform distortion and a decrease in power quality, thereby reducing the power quality index.

[0074] The harmonic current content rate is negatively correlated with the power quality index. As the harmonic current content rate increases, a high harmonic current content rate is generated. A high harmonic current content rate indicates that the current waveform contains more harmonic components, and the current waveform flowing in the system becomes more complex, which may cause overheating and failure of equipment, and then interfere with the normal operation of the power system, reduce the reliability and stability of power supply, and reduce the power quality index accordingly.

[0075] The harmonic distortion rate is negatively correlated with the power quality index. This is because an increase in the harmonic distortion rate will cause distortion of the voltage and current waveforms in the power system, affecting equipment performance and system stability, thereby reducing power quality and the power quality index will decrease accordingly.

[0076] The power factor and the power quality index are positively correlated. As the power factor increases, a high power factor is generated. A high power factor means that the current waveform is closer to an ideal sine wave, reducing the harmonic components, thereby improving the quality of the electric energy, and the power quality index increases accordingly.

[0077] According to the correlation between harmonic voltage content rate, harmonic current content rate, harmonic distortion rate, power factor and power quality index, the harmonic voltage content rate, harmonic current content rate, harmonic distortion rate and power factor are processed and correlated to generate a power quality index for evaluating power quality. The formula is as follows:

[0078]

[0079] Among them, PQI j is the power quality index of the jth individual. The power quality index is used to comprehensively evaluate the influence of power quality parameters on power quality from four aspects: harmonic voltage content, harmonic current content, harmonic distortion rate and power factor. The larger the power quality index, the better the power quality.

[0080] TV j is the harmonic voltage content rate of the jth individual, TI j is the harmonic current content of the jth individual, TH j is the harmonic distortion rate of the jth individual, PF j is the power factor of the jth individual, α 1 is the weight coefficient of harmonic voltage content, α 2 is the weight coefficient of harmonic current content, α 3 is the weight coefficient of harmonic distortion rate, α 4 is the weight coefficient of the power factor.

[0081] The reasons for setting the above function form to express the functional relationship between the power quality index and the harmonic voltage content rate, harmonic current content rate, harmonic distortion rate, and power factor are as follows:

[0082] First, the larger the power factor, the more effective the use of electricity, making the power quality index QPI j Therefore, the power factor, as the numerator in the formula, positively promotes the improvement of the power quality index.

[0083] Second, the higher the harmonic indexes such as harmonic voltage content, harmonic current content, and harmonic distortion rate, the lower the power quality, which may lead to overheating or unstable operation of the equipment. Therefore, these parameters, as denominators, negatively affect the power quality index, which is in line with the normal logic of power quality.

[0084] Third, by introducing the weight coefficient (α1 , α 2 , α 3 , α 4 ), the importance of each factor in power quality assessment can be flexibly adjusted according to different application scenarios or specific needs, making the assessment process more accurate and adaptable to different power system characteristics.

[0085] Fourth, as a power quality index, PQI can be conveniently used to compare power quality at different times and locations. This standardized evaluation method enables power operators to quickly identify which areas or time periods have poor power quality, so as to make targeted improvements.

[0086] Power factor (PF) is one of the key indicators for evaluating power quality. It directly reflects the relationship between effective power and apparent power in the power system. A high power factor means that power is used more efficiently, thereby improving power quality. Therefore, the weight coefficient of the power factor (α 4 ) should be set relatively high.

[0087] Harmonic voltage content (TV) is also an important factor affecting power quality, but its impact is relatively small compared to power factor. High harmonic voltage content may cause equipment overheating and failure, but under normal circumstances, the impact of power factor is more significant. Therefore, the weight coefficient of harmonic voltage content (α 1 ) should be greater than the harmonic current content rate (α 2 ) and harmonic distortion rate (α 3 ).

[0088] The influence of harmonic current content (TI) and harmonic distortion (TH) is a minor factor in power quality assessment. In most cases, the influence of harmonic current content is often smaller than that of harmonic voltage content, while harmonic distortion (TH) is usually considered to be the factor with the least influence on power quality. Therefore, setting α 3 <α 2 .

[0089] In summary, the weight coefficient of power factor is greater than the weight coefficient of harmonic voltage content, the weight coefficient of harmonic voltage content is greater than the weight coefficient of harmonic current content, and the weight coefficient of harmonic current content is greater than the weight coefficient of harmonic distortion rate, that is, in α 3 +α 2 +α 1 +α 4 =1, set 0<α 3 <α 2 <α 1 <α 4 <1.

[0090] As an embodiment, α1 The value range is an open interval of 0.2-0.3, α 2 The value range is 0.15-0.2, α 3 The value range is an open interval of 0.1-0.15, α 4 The value range is an open interval of 0.3-0.4. The specific value is set by the technicians according to the actual situation and is not limited here.

[0091] Based on the above embodiment, the correlation among input signal amplitude, power loss, dynamic response time and system performance index is as follows:

[0092] A larger input signal amplitude can usually improve the filter's response and processing capabilities, thereby improving the system performance index. When the filter receives a sufficiently large signal, it can more effectively suppress harmonics and improve power quality. Therefore, there is usually a positive correlation between the input signal amplitude and the system performance index.

[0093] An increase in power loss means a decrease in system efficiency, which may lead to an increase in heat and energy waste, which will affect the stability and reliability of the system, thereby reducing the system performance index. Therefore, there is usually a negative correlation between power loss and system performance index.

[0094] Filters with shorter dynamic response times can adapt to changes in input signals more quickly, which usually improves system performance. In a dynamic power environment, the ability to respond quickly is crucial; conversely, longer dynamic response times may cause the system to be unable to adjust in time, thereby reducing performance. Therefore, there is usually a negative correlation between dynamic response time and system performance index.

[0095] According to the correlation between input signal amplitude, power loss, dynamic response time and system performance index, data processing and correlation analysis are performed on input signal amplitude, power loss and dynamic response time to generate a system performance index for evaluating system performance. The formula is as follows:

[0096]

[0097] Among them, SZ j is the system performance index of the jth individual. The system performance index is used to comprehensively evaluate the influence of system performance parameters on system performance from three levels: input signal amplitude, power loss, and dynamic response time. The larger the system performance index, the better the system performance.

[0098] A j is the input signal amplitude of the jth individual, P j is the power loss of the jth individual, T j is the dynamic response time of the jth individual, β1 is the weight coefficient of the input signal amplitude, β 2 is the weight coefficient of power loss, β 3 is the weight coefficient of dynamic response time.

[0099] The reasons for setting the above function form to express the functional relationship between the system performance index and the input signal amplitude, power loss, and dynamic response time are as follows:

[0100] First, the input signal amplitude A j The larger the system performance index SZ is, the j The higher the input signal, the better the filter response and system performance. j As a numerator, it can express the input signal amplitude A j and system performance index SZ j The positive correlation between them.

[0101] Second, power loss P j and dynamic response time T j The larger the system performance index SZ is, the j The lower the power loss P j and dynamic response time T j As the denominator, it can express the power loss P j , Dynamic response time T j and system performance index SZ j The negative correlation between them is consistent with the normal logic of system performance.

[0102] Third, by introducing the weight coefficient (β 1 , β 2 , β 3 ), which can flexibly adjust the influence of various factors on the system performance index. This design allows the contribution value of each parameter to be dynamically optimized according to actual application requirements or system characteristics, making the evaluation more accurate.

[0103] Fourth, this formula can convert the input signal amplitude A j , power loss P j and dynamic response time T j These three key factors are evaluated together to form a unified system performance index SZ j , used to compare the performance of different systems.

[0104] Input signal amplitude A j Directly determines the input strength of the system. The larger the amplitude, the stronger the system's response ability is when processing signals. Therefore, a larger β 1It can better reflect the positive impact of the input signal amplitude on system performance. This direct positive impact makes the input signal amplitude A j In many applications, it becomes a key factor to improve system performance.

[0105] Power loss P j It is an important indicator to measure the efficiency of the system. Higher power loss means that the system is not efficient in energy use, resulting in heat loss and equipment aging. j The performance of the system is significantly affected, but its impact is usually manifested in indirect ways such as reduced efficiency and increased risk of failure. This indirectness makes the impact of power loss less important than the input signal amplitude A in some cases. j Intuitive. Therefore, a lower weight β 2 .

[0106] Dynamic response time T j It reflects the system's ability to adapt to changes in the input signal. Although a long dynamic response time will reduce the performance of the system, in many cases, its impact is not as significant as the input signal amplitude and power loss. 3 Smaller weight.

[0107] In summary, the weight coefficient of the input signal amplitude is greater than the weight coefficient of the power loss, and the weight coefficient of the power loss is greater than the weight coefficient of the dynamic response time, that is, in β 1 +β 2 +β 3 =1, set 0<β 3 <β 2 <β 1 <1.

[0108] As an embodiment, β 1 The value range is an open interval of 0.4-0.5, β 2 The value range is an open interval of 0.3-0.4, β 3 The value range is an open interval of 0.2-0.3. The specific value is set by the technicians according to the actual situation and is not limited here.

[0109] Based on the above embodiment, a functional relationship between the power quality index, the system performance index and the comprehensive evaluation index is constructed, and the formula is as follows:

[0110] ZGar j =γ 1 ·PQI j +γ 2 ·SZ j

[0111] Among them, ZPzs jIt is a comprehensive evaluation index, which is used to combine the power quality index and the system performance index to comprehensively evaluate the efficiency and stability of power operation. The larger the comprehensive evaluation index, the higher the efficiency and stability of power operation.

[0112] It should be noted that, as can be seen from the above description, the power quality index PQI j The larger the value, the better the power quality and the system performance index SZ. j The larger the value, the better the system performance. Therefore, the comprehensive evaluation index ZPzs j Power Quality Index PQI j , System Performance Index SZ j All are positively correlated, so the above weighted summation form of comprehensive evaluation index calculation formula is set;

[0113] In the formula, γ 1 and γ 2 are the weights in the calculation of power quality index and system performance index, and γ 1 and γ 2 The specific value of is determined by the hierarchical analysis method, and the specific logic is as follows:

[0114] The two indicators, power quality index and system performance index, are marked, and the relative importance between them is determined by the nine-scale method to construct a judgment matrix, where the index of the power quality index is marked as 1 and the index of the system performance index is marked as 2. The constructed judgment matrix [q uv ] 2×2 for:

[0115]

[0116] Among them, u and v both represent the index of the coefficient, and u∈[1,2], v∈[1,2], indicating the importance of the index with index u to the comprehensive evaluation index relative to the index with index v, q uv The specific value of q is determined by relevant experts using a 1-9 scoring method. uv =9 means that the index with index u is more important to the comprehensive evaluation index than the index with index v. uv =1 means that the index with index u is extremely unimportant to the comprehensive evaluation index compared with the index with index v;

[0117] Each element value in the judgment matrix is ​​divided by the sum of its columns to obtain a normalized judgment matrix. The mean of the element values ​​in each row of the normalized judgment matrix is ​​calculated, and the mean of the element values ​​in the first row is used as the proportional coefficient of the power quality index, and the mean of the element values ​​in the second row is used as the proportional coefficient of the system performance index. With the constraint that the sum of the scaled values ​​is equal to 1, the two proportional coefficients are scaled in equal proportion, and the scaled values ​​are used as the weights of the corresponding indexes.

[0118] Based on the above embodiment, the specific process of step S5 is as follows:

[0119] Iterative optimization is performed on the individuals of the initial population of configuration parameters. During the iterative optimization process, the constraints of the configuration parameters are set, that is, the maximum and minimum values ​​of the gain, cutoff frequency, bandwidth and harmonic suppression rate are set respectively. Within the constraints of the gain, cutoff frequency, bandwidth and harmonic suppression rate, the configuration parameters of the active high-frequency filter are iteratively optimized. Specifically, the comprehensive evaluation index ZPzs is set to j Sort from large to small and select the comprehensive evaluation index ZPzs j The individuals in the front rank are regarded as the parents, and the front rank refers to the individuals in the comprehensive evaluation index ZPzs j For the first 50% of individuals, the genes of the parent individuals are exchanged, combined and mutated through crossover and mutation operations to generate new individuals. The power quality parameters and system performance parameters of the newly generated individuals are obtained using the power operation data prediction model, and their comprehensive evaluation index is calculated. The new individuals and the parent generation are used as a new population, and the selection, crossover and mutation operations are repeated until the predetermined number of iterations is reached. The individual corresponding to the maximum value of the comprehensive evaluation index is used as the optimal combination of the configuration parameters of the active high-frequency filter, and the individual corresponding to the maximum value of the comprehensive evaluation index is calibrated as Q j1 = {G j1 ,f j1 ,BW j1 ,H j1}, then the configuration parameter combination of the optimal active high-frequency filter is gain G j1 , cut-off frequency fj j1 , bandwidth BWj j1 Harmonic suppression rate H j1 .

[0120] See also Figure 2 , the present invention also provides a technical solution:

[0121] An active high-frequency filter optimization configuration system, the system is used to execute any one of the above-mentioned active high-frequency filter optimization configuration methods, comprising:

[0122] The data acquisition module is used to obtain the power operation data of the distribution network under different configuration parameter combinations of the active high-frequency filter. The power operation data of the distribution network is the average value of the power operation data of each node in the distribution network. The power operation data includes power quality parameters and system performance parameters. The power quality parameters include harmonic voltage content rate, harmonic current content rate, harmonic distortion rate and power factor. The system performance parameters include input signal amplitude, power loss and dynamic response time. The configuration parameters include gain, cutoff frequency, bandwidth and harmonic suppression rate.

[0123] The prediction model building module is used to build a power operation data prediction model, taking the configuration parameter combinations of different active high-frequency filters as input, and the power quality parameters and system performance parameters of the distribution network as label training models to train the power operation data prediction;

[0124] An initial population construction module is used to establish constraints on configuration parameters. Under the constraints on configuration parameters, the configuration parameters are randomly combined to construct individuals of the initial population of configuration parameters. The individuals of the initial population of configuration parameters are input into the power operation data prediction model to obtain power quality parameters and system performance parameters.

[0125] The data processing and analysis module is used to construct the functional relationship between the harmonic voltage content rate, the harmonic current content rate, the harmonic distortion rate, the power factor and the power quality index, construct the functional relationship between the input signal amplitude, the power loss and the dynamic response time and the system performance index, and construct the functional relationship between the power quality index, the system performance index and the comprehensive evaluation index. The comprehensive evaluation index is used to comprehensively evaluate the high efficiency and stability of power operation;

[0126] The parameter optimization module is used to maximize the comprehensive evaluation index as the objective function, iteratively optimize the individuals of the initial population of configuration parameters through a genetic algorithm under the constraints of the configuration parameters, obtain the optimal individuals, and extract the optimal values ​​of the configuration parameters based on the optimal individuals.

[0127] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0128] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by computer software, electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0129] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0130] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for optimizing the configuration of an active high-frequency filter, characterized in that: The specific steps include: S1. Obtain the power operation data of the distribution network under different configuration parameter combinations of active high-frequency filters. The power operation data of the distribution network is the average of the power operation data of each node in the distribution network. The power operation data includes power quality parameters and system performance parameters. The power quality parameters include harmonic voltage content rate, harmonic current content rate, harmonic distortion rate and power factor. The system performance parameters include input signal amplitude, power loss and dynamic response time. The configuration parameters include gain, cutoff frequency, bandwidth and harmonic suppression rate. S2. Construct a power operation data prediction model, take the configuration parameter combinations of different active high-frequency filters as input, and the power quality parameters and system performance parameters of the distribution network as label training models to train the power operation data prediction model; S3. Establishing constraints on configuration parameters, randomly combining configuration parameters under the constraints on configuration parameters, constructing individuals of the initial population of configuration parameters, inputting individuals of the initial population of configuration parameters into the power operation data prediction model, and obtaining power quality parameters and system performance parameters; S4. Construct the functional relationship between harmonic voltage content rate, harmonic current content rate, harmonic distortion rate, power factor and power quality index, construct the functional relationship between input signal amplitude, power loss and dynamic response time and system performance index, construct the functional relationship between power quality index, system performance index and comprehensive evaluation index, and the comprehensive evaluation index is used to comprehensively evaluate the high efficiency and stability of power operation; S5. Taking the maximization of the comprehensive evaluation index as the objective function, the individuals of the initial population of configuration parameters are iteratively optimized through a genetic algorithm under the constraints of the configuration parameters to obtain the optimal individuals, and based on the optimal individuals, the optimal values ​​of the configuration parameters are extracted.

2. The active high frequency filter optimization configuration method according to claim 1, characterized in that: The configuration parameters are randomly combined to construct individuals of the initial population of configuration parameters. The specific process is as follows: The initial population is labeled as Q, and the initial population Q={Q1,Q2,…,Q j ,…,Q m }, Q j is the jth individual in the initial population, j is the index of the individual in the initial population, and j∈[1,m], m is the number of individuals in the initial population, Q j = {G j ,f j ,BW j ,H j }, where G j ,f j ,BW j ,H j are the gain, cutoff frequency, bandwidth and harmonic suppression rate of the jth individual respectively.

3. The active high frequency filter optimization configuration method according to claim 1, characterized in that: The power operation data prediction model is composed of a deep learning network based on a multilayer perceptron, and the deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer, and the first hidden layer, the second hidden layer and the third hidden layer each have at least two neurons, and each uses ReLU as an activation function; The process of training the power operation data prediction model is as follows: The configuration parameter combinations of different active high-frequency filters are used as input, and the power quality parameters and system performance parameters of the distribution network are used as output labels for training. The mean square error is used as the loss function. When the mean square error is in the range of [0, 0.01], the training of the power operation data prediction model is completed.

4. The active high frequency filter optimization configuration method according to claim 2, characterized in that: Construct the functional relationship between harmonic voltage content rate, harmonic current content rate, harmonic distortion rate, power factor and power quality index. The formula is as follows: Among them, PQI j is the power quality index of the jth individual, TV j is the harmonic voltage content rate of the jth individual, TI j is the harmonic current content of the jth individual, TH j is the harmonic distortion rate of the jth individual, PF j is the power factor of the jth individual, α1 is the weight coefficient of the harmonic voltage content, α2 is the weight coefficient of the harmonic current content, α3 is the weight coefficient of the harmonic distortion rate, α4 is the weight coefficient of the power factor, α3+α2+α1+α4=1, and 0<α3<α2<α1<α4<1.

5. The active high frequency filter optimization configuration method according to claim 4, characterized in that: Construct a functional relationship between input signal amplitude, power loss and dynamic response time and system performance index, the formula is as follows: Among them, SZ j is the system performance index of the jth individual, A j is the input signal amplitude of the jth individual, P j is the power loss of the jth individual, T j is the dynamic response time of the jth individual, β1 is the weight coefficient of the input signal amplitude, β2 is the weight coefficient of the power loss, β3 is the weight coefficient of the dynamic response time, β1+β2+β3=1, and 0<β3<β2<β1<1.

6. The active high frequency filter optimization configuration method according to claim 5, characterized in that: Construct the functional relationship between power quality index, system performance index and comprehensive evaluation index, the formula is as follows: ZPzs j =γ1·PQI j +γ2·SZ j Among them, ZPzs j is a comprehensive evaluation index, γ1 and γ2 are the weights in the calculation of power quality index and system performance index respectively, and the specific values ​​of γ1 and γ2 are determined by the hierarchical analysis method.

7. The active high frequency filter optimization configuration method according to claim 6, characterized in that: The specific process of step S5 is as follows: Iterative optimization is performed on the individuals of the initial population of configuration parameters. During the iterative optimization process, the constraints of the configuration parameters are set, that is, the maximum and minimum values ​​of the gain, cutoff frequency, bandwidth and harmonic suppression rate are set respectively. Within the constraints of the gain, cutoff frequency, bandwidth and harmonic suppression rate, the configuration parameters of the active high-frequency filter are iteratively optimized. Specifically, the comprehensive evaluation index ZPzs is set to j Sort from large to small and select the comprehensive evaluation index ZPzs j The individuals in the front row are taken as the parents. Through crossover and mutation operations, the genes of the parent individuals are exchanged, combined and mutated to generate new individuals. The power quality parameters and system performance parameters of the newly generated individuals are obtained by using the power operation data prediction model, and their comprehensive evaluation index is calculated. The new individuals and the parents are taken as the new population, and the selection, crossover and mutation operations are repeated until the predetermined number of iterations is reached. The individual corresponding to the maximum value of the comprehensive evaluation index is taken as the optimal combination of the configuration parameters of the active high-frequency filter, and the individual corresponding to the maximum value of the comprehensive evaluation index is calibrated as Q j1 = {G j1 ,f j1 ,BW j1 ,H j1 }, then the configuration parameter combination of the optimal active high-frequency filter is gain G j1 , cut-off frequency f j1 , bandwidth BW j1 Harmonic suppression rate H j1 .

8. An active high-frequency filter optimization configuration system, the system is used to execute an active high-frequency filter optimization configuration method according to any one of claims 1 to 7, characterized in that: include: The data acquisition module is used to obtain the power operation data of the distribution network under different configuration parameter combinations of the active high-frequency filter. The power operation data of the distribution network is the average value of the power operation data of each node in the distribution network. The power operation data includes power quality parameters and system performance parameters. The power quality parameters include harmonic voltage content rate, harmonic current content rate, harmonic distortion rate and power factor. The system performance parameters include input signal amplitude, power loss and dynamic response time. The configuration parameters include gain, cutoff frequency, bandwidth and harmonic suppression rate. The prediction model building module is used to build a power operation data prediction model, taking the configuration parameter combinations of different active high-frequency filters as input, and the power quality parameters and system performance parameters of the distribution network as label training models to train the power operation data prediction; An initial population construction module is used to establish constraints on configuration parameters. Under the constraints on configuration parameters, the configuration parameters are randomly combined to construct individuals of the initial population of configuration parameters. The individuals of the initial population of configuration parameters are input into the power operation data prediction model to obtain power quality parameters and system performance parameters. The data processing and analysis module is used to construct the functional relationship between the harmonic voltage content rate, the harmonic current content rate, the harmonic distortion rate, the power factor and the power quality index, construct the functional relationship between the input signal amplitude, the power loss and the dynamic response time and the system performance index, and construct the functional relationship between the power quality index, the system performance index and the comprehensive evaluation index. The comprehensive evaluation index is used to comprehensively evaluate the high efficiency and stability of power operation; The parameter optimization module is used to maximize the comprehensive evaluation index as the objective function, iteratively optimize the individuals of the initial population of configuration parameters through a genetic algorithm under the constraints of the configuration parameters, obtain the optimal individuals, and extract the optimal values ​​of the configuration parameters based on the optimal individuals.

Images