Resonance detection method and detection system for new energy power generation

Abstract

The application relates to the technical field of power generation resonance detection, in particular to a resonance detection method and system for new energy power generation, which comprises the following steps: based on the distance between each power generation node and each cluster center and the distance between all cluster centers, and the difference between the first resonance value between each power generation node and each cluster center and the degree of belonging to each cluster center, the optimal cluster center is obtained by iteratively optimizing a clustering algorithm; the change trend of the harmonic corresponding current signal of each power generation node and each optimal cluster center at the current moment is analyzed respectively, and the power generation node is subjected to resonance detection in combination with the degree of belonging of each power generation node to each optimal cluster center. The application solves the problem that resonance risk cannot be accurately and timely detected due to dynamic changes of resonance frequency and multi-node coupling interference in the new energy power generation process, and improves the accuracy and reliability of resonance risk detection for new energy power generation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power generation resonance detection, in particular to a resonance detection method and system for new energy power generation. BACKGROUND

[0002] In the process of new energy power generation, grid-connected inverters are used to convert the electrical energy generated by renewable energy into electrical energy that meets the requirements of the power grid, and then deliver it to the power grid. With the rapid development of new energy power generation, a large number of grid-connected inverters are connected to the power grid, forming a multi-inverter grid-connected system. The mutual coupling and interference between inverters through the power grid can cause a single grid-connected inverter that can work stably to become unstable after being connected in parallel with multiple inverters, and the grid-connected current of the grid-connected inverter can resonate. Resonance problems can seriously affect the reliability of the operation of the multi-inverter grid-connected system and the power quality of the power supply. Therefore, for the multi-inverter grid-connected system of new energy power generation, resonance detection is a key link to realize the safe and efficient operation of the multi-inverter grid-connected system.

[0003] The traditional method is to assume that the resonance frequency is basically unchanged, and then detect the resonance risk by injecting a current of a specific frequency. However, in practice, the resonance frequency of the multi-inverter grid-connected system changes with the operating conditions of each grid-connected inverter, and due to the instability of new energy power generation, the change of the resonance frequency is relatively frequent. Therefore, it is difficult to accurately detect the resonance state of different power generation nodes using the traditional method; at the same time, the interaction between multiple inverters through the power grid can exacerbate the occurrence of resonance phenomenon, and strong resonance can easily occur in a short period of time. Therefore, only considering the change of the resonance state of a single power generation node reduces the accuracy and reliability of the resonance risk detection of new energy power generation. SUMMARY

[0004] To solve the above technical problems, the purpose of the present application is to provide a resonance detection method and system for new energy power generation, and the technical solution adopted is as follows:

[0005] In a first aspect, the present application provides a resonance detection method for new energy power generation, which comprises the following steps:

[0006] Real-time acquisition of current data of each power generation node within a preset time length before each time, and composition of current signals of each power generation node at each time, acquisition of fundamental wave and all harmonics of the current signal;

[0007] The harmonic distortion rate of each power generation node at each time is determined based on the effective value of the fundamental wave and all harmonic amplitudes of the current signal of each power generation node at each time and the fluctuation degree of the current signal corresponding to each harmonic, so as to determine the first resonance value of each power generation node at each time, and the resonance feature vector and the resonance change vector of each power generation node at each time are constructed in combination with the difference of the effective value of all harmonic amplitudes in the current signal of each power generation node between each time and the previous time;

[0008] The all power generation nodes are clustered based on the difference of the resonance feature vectors between all power generation nodes at each time, the feature weight between each power generation node and each cluster center at each time is determined based on the distance between each power generation node and each cluster center and the distance between all cluster centers and the difference of the first resonance value between each power generation node and each cluster center, and the target function of the clustering algorithm is determined in combination with the difference of the resonance change vector and the difference of the first resonance value between each power generation node and each cluster center at each time and the degree of membership of each power generation node to each cluster center, so as to obtain the optimal cluster center through iterative optimization of the clustering algorithm.

[0009] The change trend of the harmonic corresponding current signal of each power generation node and each optimal cluster center at the current time is analyzed respectively, and the resonance speedup characteristic value of each harmonic of each power generation node at the current time is determined in combination with the degree of membership of each power generation node to each optimal cluster center, so as to determine the resonance risk characteristic value and perform resonance detection on the power generation node at the current time.

[0010] Preferably, the expression of the harmonic distortion rate of each power generation node at each time is: ; in the formula, , wherein , , respectively, represent the effective value of the fundamental wave amplitude and the effective value of the amplitude of the hth harmonic in the current signal of the jth power generation node at the ith time. , respectively, represent the effective value of the fundamental wave amplitude and the effective value of the amplitude of the hth harmonic in the current signal of the jth power generation node at the ith time. , respectively, represent the effective value of the fundamental wave amplitude and the effective value of the amplitude of the hth harmonic in the current signal of the jth power generation node at the ith time.

[0011] Preferably, the first resonance value of each power generation node at each time is the average of the harmonic distortion rates of each power generation node at all times within a preset time period before each time.

[0012] Preferably, the determination method of the resonance change vector of each power generation node at each time is:

[0013] The total difference between the effective values ​​of all subharmonic amplitudes of each power generation node at each time point and the previous time point is denoted as the second resonance value of each power generation node at each time point.

[0014] The first resonance value of each power generation node at each time point and the second resonance value of the corresponding power generation node at all times within the preset time period before each time point are used to form the resonance feature vector of each power generation node at each time point.

[0015] The second resonance value of each power generation node at all times within the preset time period before each time is used to form the resonance change vector of each power generation node at each time.

[0016] Preferably, the expression for the feature weights between each power generation node and each cluster center at each time point is: ; This represents the feature weight between the m-th power generation node and the n-th cluster center at time t; This represents the distance between the m-th power generation node and the n-th cluster center at time t; This represents the maximum distance between all cluster centers at time t; The difference between the first resonance value of the m-th power generation node and the n-th cluster center at time t is represented by norm[ ], which represents the normalization function. This is a preset value.

[0017] Preferably, the method for determining the objective function of the clustering algorithm is as follows:

[0018] Objective function with respect to time t The expression is: In the formula, This indicates the degree to which the m-th power generation node belongs to the n-th cluster center at time t; The norm of the resonant change vector between the m-th power generation node and the n-th cluster center at time t is denoted by t. This represents the feature weight between the m-th power generation node and the n-th cluster center at time t; This represents the difference in the first resonance value between the m-th power generation node and the n-th cluster center at time t; This indicates the total number of all power generation nodes; This represents the number of all cluster centers; This indicates the preset fuzzy index.

[0019] Preferably, the iterative optimization of the clustering algorithm to obtain the optimal cluster centers includes:

[0020] Randomly select a preset number of power generation nodes as initial cluster centers in all power generation nodes, take all initial cluster centers as inputs of a fuzzy C-means clustering algorithm, wherein, take J(t) as an objective function of the fuzzy C-means clustering algorithm, take the difference between the resonance feature vectors of the power generation nodes as a metric distance of the fuzzy C-means clustering algorithm, and take the cluster center corresponding to the minimum value of the objective function as an optimal cluster center.

[0021] Preferably, the method for determining the resonance speed-up characteristic value of each harmonic of each power generation node at the current moment is as follows:

[0022] Fit all current data in the current current signal of each harmonic of each power generation node and each optimal cluster center respectively, and obtain the fitting straight line of each harmonic of each power generation node and each optimal cluster center respectively;

[0023] Calculate the product of the degree to which each power generation node belongs to each optimal cluster center and the slope of the fitting straight line of each harmonic of the corresponding optimal cluster center at the current moment, and record it as the comprehensive product between each power generation node and each optimal cluster center at each harmonic.

[0024] Calculate the accumulation result of the comprehensive product between each power generation node and all optimal cluster centers at each harmonic at the current moment, and take the sum of the slope of the fitting straight line of each harmonic of each power generation node at the current moment and the corresponding accumulation result as the resonance speed-up characteristic value of each harmonic of each power generation node at the current moment.

[0025] Preferably, the determination of the resonance risk characteristic value for the power generation node at the current moment includes:

[0026] If the resonance speed-up characteristic value of the zth harmonic of the qth power generation node at the current moment is less than or equal to 0, set the resonance risk characteristic value of the zth harmonic of the qth power generation node at the current moment to 0, otherwise, calculate the difference between the preset harmonic current allowable value of the zth harmonic of the qth power generation node and the effective value of the zth harmonic amplitude of the qth power generation node, and take the ratio of the resonance speed-up characteristic value of the zth harmonic of the qth power generation node and the difference as the resonance risk characteristic value of the zth harmonic of the qth power generation node at the current moment.

[0027] If the effective value of all harmonic amplitudes of the qth power generation node at the current moment is less than the corresponding preset harmonic current allowable value, and the resonance risk characteristic value of all harmonics of the qth power generation node does not exceed the preset resonance risk threshold, then the qth power generation node at the current moment does not have resonance risk, otherwise, the qth power generation node at the current moment has resonance risk, and the resonance detection is performed on all power generation nodes by traversing all power generation nodes.

[0028] In a second aspect, the embodiments of the present application also provide a resonance detection system for new energy power generation, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the resonance detection method for new energy power generation as described above when executing the computer program.

[0029] The present application has at least the following beneficial effects:

[0030] The present application combines the overall resonance risk of the node in the preset period, i.e., the first resonance value, with the instantaneous change intensity of the harmonic form, i.e., the second resonance value, by constructing a dynamic resonance feature vector and a change vector, thereby realizing the quantification and trend evaluation of the resonance risk state of the new energy power generation node, which helps to filter out the misjudgment caused by accidental fluctuations; further, the present application constructs an innovative objective function that simultaneously fuses the node behavior pattern similarity and the risk level closeness, thereby accurately dividing the power generation nodes with similar and close risks into the same group, which not only overcomes the defects of traditional clustering that ignores the absolute value difference of risks and effectively eliminates the interference of accidental fluctuations of a single node, but also accurately identifies the power generation node clustering cluster that "clusters" and risks synchronously and rapidly rise, thereby significantly improving the accuracy of resonance risk positioning and the reliability of early warning; further, the present application analyzes the harmonic change trend of the power generation node and the clustering center to which it belongs, calculates the accurate resonance speedup characteristic value, and combines the harmonic over-standard risk distance to finally calculate the resonance risk characteristic value that can quantify the resonance risk possibility, thereby realizing the rapid and accurate detection and early warning of the potential resonance risk of the multi-inverter grid-connected system, and improving the accuracy and reliability of resonance risk positioning. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0032] Figure 1 The step flowchart of the resonance detection method for new energy power generation provided by an embodiment of the present application;

[0033] Figure 2 The resonance risk detection process schematic diagram provided by an embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to further illustrate the technical means and effects taken by the present application to achieve the predetermined object, the specific implementation, structure, features and effects of the resonance detection method and system for new energy power generation according to the present application are described in detail below in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0036] The specific scheme of the resonance detection method and system for new energy power generation provided by the present application is described below in combination with the drawings.

[0037] Please refer to Figure 1 which shows the step flowchart of the resonance detection method for new energy power generation provided by one embodiment of the present application, which includes the following steps:

[0038] Step S1: Real-time acquisition of current data of each power generation node within a preset time length before each time, and composition of current signal at each time, acquisition of fundamental wave and all harmonics of current signal.

[0039] In the present embodiment, the nominal voltage of the grid of the multi-inverter grid-connected system is 0.38kV, the reference short-circuit capacity is 10MVA, the power frequency is 50Hz, and there are 500 power generation nodes. At the grid side, the current data of each power generation node is acquired in real time through a current sensor, and the current data of all times within a preset time length before each time is composed to form the current signal of each power generation node at each time, wherein the value of the preset time length is artificially set, and in the present embodiment, the value of the preset time length is 100ms, and the sampling frequency is 10kHz. In actual application, as other implementation manners, the implementer can also set it himself according to the specific circumstances, and the present embodiment does not make special limitation.

[0040] Further, the current signal of each power generation node at each time is taken as the input of the Fourier transform algorithm, and the fundamental wave and all harmonics of each power generation node current signal are output.

[0041] And the fundamental wave and all harmonics are inversely transformed by the Fourier transform algorithm to obtain the corresponding part of the fundamental wave and all harmonics in the current signal.

[0042] Wherein, the Fourier transform algorithm is a known technology, and the specific process of acquiring the fundamental wave and harmonics of the time domain signal is not repeated.

[0043] Step S2: By dynamically analyzing the current state, change trend and correlation of each node harmonic, nodes with similar behaviors are clustered, and the deterioration speed of each node harmonic is evaluated based on this to detect the resonance of the power generation node.

[0044] The new energy generation is strong at times and weak at times, which causes the size of the harmonic current to jump constantly. The simple instantaneous change at a single moment may have contingency and cannot represent the real risk state of the power generation node. For example, a very high harmonic peak may disappear soon, but it may be misjudged as a serious problem. Therefore, in order to avoid the misjudgment caused by the instantaneous accidental harmonic peak, the embodiment dynamically analyzes the current state, change trend and correlation of each node harmonic, clusters nodes with similar behaviors, and evaluates the deterioration speed of each node harmonic based on this to detect the resonance of the power generation node, so that the evaluation of the resonance state of the power generation node is more accurate. The resonance risk detection process provided by the embodiment is as shown in Figure 2 The specific process of detecting the resonance of the power generation node is as follows:

[0045] S2.1: Based on the effective value of the fundamental wave and all harmonic amplitudes of the current signal of each power generation node at each moment, and the fluctuation degree of the current signal corresponding to each harmonic, the harmonic distortion rate of each power generation node at each moment is determined to determine the first resonance value of each power generation node at each moment, and the difference between the effective value of all harmonic amplitudes of the current signal of each power generation node at each moment and the effective value of all harmonic amplitudes of the current signal of each power generation node at the previous moment is combined to construct the resonance feature vector and the resonance change vector of each power generation node at each moment.

[0046] Firstly, in the embodiment, based on the effective value of the fundamental wave and all harmonic amplitudes of the current signal of each power generation node at each moment, and the fluctuation degree of the current signal corresponding to each harmonic, the harmonic distortion rate of each power generation node at each moment is determined to determine the first resonance value of each power generation node at each moment, to filter the misjudgment of the harmonic risk caused by the accidental fluctuation of the harmonic data. Specifically,

[0047] As an implementation manner, in the embodiment, the expression of the harmonic distortion rate of the jth power generation node at the ith moment is as follows: ; in the formula, , respectively represent the effective value of the fundamental wave amplitude, the effective value of the amplitude of the hth harmonic in the current signal of the jth power generation node at the ith moment; represents the dispersion degree of all current data in the current signal corresponding to the hth harmonic of the jth power generation node at the ith moment; represents the number of all harmonics in the current signal of the power generation node at the ith moment.

[0048] ​It should be noted that there are many ways to measure the dispersion degree of a set of data. In the embodiment, the coefficient of variation of all current data in the hth harmonic current signal of the jth power generation node at time i is taken as the dispersion degree of all current data in the hth harmonic current signal of the jth power generation node at time i. In actual application, as an alternative, the implementer can also use other methods such as variance or standard deviation to measure the dispersion degree of data. The selection of the method for measuring the dispersion degree of data is not particularly limited in the embodiment.

[0049] The calculation method of the harmonic amplitude effective value and the calculation method of the coefficient of variation are both known technologies, and the specific calculation processes are not described in detail.

[0050] It should be noted that only the fundamental wave and all harmonics before the 20th harmonic are calculated and analyzed in the embodiment, and the 20th harmonic is not included.

[0051] According to the harmonic distortion rate of each power generation node at each time, it can be understood that the harmonic distortion rate is used to represent the severity of power quality and harmonic pollution of the grounding. If the fundamental wave amplitude effective value of the jth power generation node is smaller, it means that the power generation power of the jth power generation node is smaller, and the power generation node is in a weak working state. At this time, even a small harmonic current will be amplified, and the probability of the current power generation node harmonic being disturbed is greater, so the corresponding harmonic distortion rate is greater. At the same time, if the ratio of the effective value of the amplitude of the hth harmonic of the jth power generation node to the dispersion degree of the corresponding current data is greater, it means that there is a greater possibility of stable and serious harmonic pollution problem in the jth power generation node, so the corresponding harmonic distortion rate is greater.

[0052] On the contrary, if the fundamental wave amplitude effective value of the jth power generation node is greater, it means that the power generation power of the jth power generation node is greater, and the power generation node is in a strong working state. At this time, the inhibition ability of the power generation node to the harmonic current mutation is stronger, and even if there is a certain amount of harmonic current, its relative influence will be weakened, and the probability of the current power generation node harmonic being disturbed is smaller, so the corresponding harmonic distortion rate is smaller. At the same time, if the ratio of the effective value of the amplitude of the hth harmonic of the jth power generation node to the dispersion degree of the corresponding current data is smaller, it means that the harmonic energy in the jth power generation node is lower, or the harmonic component is very unstable and has strong randomness, and the possibility of stable and serious harmonic pollution problem is smaller, so the corresponding harmonic distortion rate is smaller.

[0053] Further, the embodiment takes the average of the harmonic distortion rates of all time points of each power generation node within a preset time period before each time point as the first resonance value of each power generation node at each time point, which is used to represent the overall resonance state and severity of the power generation node within the preset time period. The greater the first resonance value of the current power generation node, the higher the resonance risk of the current power generation node.

[0054] It should be noted that the value of the preset time period length is artificially set, and in the embodiment, the value of the preset time period length is 3s. In actual application, as other implementation manners, the implementer can also set it by himself according to the specific situation, which is not specially limited in the embodiment.

[0055] Further, the embodiment constructs the resonance feature vector and the resonance change vector of each power generation node at each time point based on the first resonance value of each power generation node at each time point and the difference between the effective values of all harmonic amplitudes of each power generation node between each time point and the previous time point, so as to quantify the harmonic state and dynamic change trend of each node, and provide a multi-dimensional data basis for subsequent accurate identification of the power generation node group with similar behaviors and accurate assessment of the development speed. Specifically,

[0056] In the embodiment, the overall difference between the effective values of all harmonic amplitudes of each power generation node between each time point and the previous time point is recorded as the second resonance value of each power generation node at each time point, which is used to represent the similarity and change intensity of the harmonic energy distribution of a node at different time points. The greater the second resonance value of the current power generation node, the more intense the change of the harmonic form of the current power generation node, which indicates that the resonance risk may be rapidly evolving. On the contrary, the smaller the second resonance value of the current power generation node, the more gentle the change of the harmonic form of the current power generation node, which indicates that the operation state of the power generation node is relatively stable, and the resonance risk is low or develops slowly.

[0057] The first resonance value of each power generation node at each time point and the second resonance value of the corresponding power generation node at all time points within the preset time period before each time point are combined to form the resonance feature vector of each power generation node at each time point.

[0058] The second resonance value of each power generation node at all time points within the preset time period before each time point is combined to form the resonance change vector of each power generation node at each time point.

[0059] So far, the embodiment combines the overall resonance risk of the node within the preset time period, i.e. the first resonance value, with the instantaneous change intensity of the harmonic form, i.e. the second resonance value, by constructing the dynamic resonance feature vector and change vector, so as to realize the quantification and trend assessment of the resonance risk state of the new energy power generation node, which helps to filter out the false judgments caused by accidental fluctuations.

[0060] S2.2: based on the difference between the resonance feature vectors of all power generation nodes at each time, clustering all power generation nodes, determining the feature weight between each power generation node and each cluster center at each time based on the distance between each power generation node and each cluster center and the distance between all cluster centers, and the difference between the first resonance values between each power generation node and each cluster center, and combining the difference between the resonance change vectors between each power generation node and each cluster center at each time and the difference between the first resonance values, and the degree to which each power generation node belongs to each cluster center, determining the objective function of the clustering algorithm to obtain the optimal cluster center through iterative optimization of the clustering algorithm.

[0061] In order to more accurately evaluate the resonance risk of the power generation node, the embodiment is used to exclude the problem that the resonance is misjudged due to the contingency of a single power generation node, and the contingency problem is solved through clustering analysis of the power generation node. The traditional clustering method usually only divides the power generation nodes into groups based on the difference between the feature vectors of the power generation nodes. However, in a huge power grid, a power generation node with a very high harmonic distortion rate and a power generation node with a very low harmonic distortion rate have completely the same harmonic change trend, but their risk levels are very different.

[0062] Therefore, in order to more accurately cluster the power generation nodes and thus evaluate the resonance risk of the power generation nodes, the embodiment is used to cluster all power generation nodes based on the difference between the resonance feature vectors of all power generation nodes at each time, determine the feature weight between each power generation node and each cluster center at each time based on the distance between each power generation node and each cluster center and the distance between all cluster centers, and the difference between the first resonance values between each power generation node and each cluster center, and combine the difference between the resonance change vectors between each power generation node and each cluster center at each time and the difference between the first resonance values, and the degree to which each power generation node belongs to each cluster center, determine the objective function of the clustering algorithm to obtain the optimal cluster center through iterative optimization of the clustering algorithm, and the specific process is as follows.

[0063] Firstly, the embodiment is used to cluster all power generation nodes based on the difference between the resonance feature vectors of all power generation nodes at each time, and determine the feature weight between each power generation node and each cluster center at each time based on the distance between each power generation node and each cluster center and the distance between all cluster centers, and the difference between the first resonance values between each power generation node and each cluster center, and the specific process is as follows.

[0064] As a specific implementation, in the embodiment, the expression of the feature weight between the mth power generation node and the nth cluster center at time t is as follows. ; represents the distance between the mth power generation node and the nth cluster center at time t; ​This represents the maximum distance between all cluster centers at time t; The difference between the first resonance value of the m-th power generation node and the n-th cluster center at time t is represented by norm[ ], which represents the normalization function. This is a preset value.

[0065] It should be noted that there are many methods to measure the differences between data. In this embodiment, the absolute value of the difference between the first resonance value between the m-th power generation node and the n-th cluster center at time t is taken as the difference between the first resonance value between the m-th power generation node and the n-th cluster center at time t. In practical applications, as other implementation methods, implementers may also use the square of the difference or other methods to measure the differences between data in combination with the specific situation. This embodiment does not impose any special restrictions on the selection of methods to measure the differences between data.

[0066] It should be understood that, The value is a preset value, which is set manually. In this embodiment... The value of is 5. The reason for choosing a value of 5 is to effectively amplify the significant differences in the level of risk that are of concern, while smoothing out those negligible minor fluctuations. In practical applications, implementers can also set it according to specific circumstances. This embodiment does not impose any special restrictions.

[0067] It should be noted that, unless otherwise specified, in this embodiment, all methods involving measuring differences between data use the method of taking the absolute value of the difference.

[0068] Based on the feature weights between each power generation node and each cluster center at each time point, it can be understood that the feature weights reflect the proportion of the power generation node's tendency to assign to each cluster center during the clustering process. If the distance between the m-th power generation node and the n-th cluster center at time t accounts for a larger proportion of the maximum distance between all cluster centers at time t, then... The larger the value, the lower the similarity between the m-th power generation node and the n-th cluster center in terms of static resonance characteristics. Therefore, the corresponding feature weight is larger. At the same time, if the difference between the first resonance value of the m-th power generation node and the n-th cluster center is larger at time t, it means that the difference between the m-th power generation node and the n-th cluster center in the harmonic variation trend at time t is larger. That is, the more inconsistent the trends are, the less likely the m-th power generation node and the n-th cluster center should be classified into the same category, and the corresponding feature weight is larger.

[0069] Conversely, if the proportion of the distance between the m-th power generation node and the n-th cluster center at time t to the maximum value of the distances between all cluster centers at time t is smaller, that is... The smaller, the higher the similarity of the mth power generation node to the nth cluster center on the static resonance characteristic, and thus the smaller the corresponding feature weight. At the same time, if the difference between the first resonance value between the mth power generation node and the nth cluster center at time t is smaller, the difference between the mth power generation node and the nth cluster center in the harmonic change trend at time t is smaller, that is, the trend is more similar, which means that the mth power generation node and the nth cluster center should be classified into the same class, and the corresponding feature weight is smaller.

[0070] Further, the feature weight between each power generation node and each cluster center at each time in the embodiment, in combination with the difference between the resonance change vectors and the difference between the first resonance values between each power generation node and each cluster center at each time, and the degree to which each power generation node belongs to each cluster center, determines the objective function of the clustering algorithm to obtain the optimal cluster center by iterative optimization of the clustering algorithm, specifically:

[0071] In the embodiment, the objective function with respect to time t is expressed as: ; in the formula, degree to which the mth power generation node belongs to the nth cluster center at time t; norm of the resonance change vector between the mth power generation node and the nth cluster center at time t; feature weight between the mth power generation node and the nth cluster center at time t; difference between the first resonance value between the mth power generation node and the nth cluster center at time t; number of all power generation nodes; number of all cluster centers; preset fuzzy index.

[0072] It should be noted that the value of the preset fuzzy index in the embodiment is artificially set, and the value of the preset fuzzy index in the embodiment is 1.5. In actual application, the implementer can also set it by himself according to the specific situation, and the embodiment does not make special limitation.

[0073] It should be noted that the degree to which the power generation node belongs to the cluster center, that is, the membership degree, can be obtained by the fuzzy C-means clustering algorithm, and the specific calculation process is not described again.

[0074] It should be noted that the value of the number of cluster clusters in the embodiment is the same as the value of the preset number, which is 20, that is, the number of cluster centers is always 20 in the entire clustering process. In actual application, as other implementation manners, the implementer can also set it by himself according to the specific situation, and the embodiment does not make special limitation.

[0075] It should be further explained that the norm of the resonance change vector between the mth power generation node and the nth cluster center at time t in the embodiment is an L2 norm. In actual application, an L1 norm or other method for measuring the difference between vectors can also be used according to specific conditions, and the embodiment is not specially limited.

[0076] The calculation method of the L2 norm is a known technology, and the specific calculation process is not described again.

[0077] According to the target function J(t), the target function is set in this way because the resonance risk is often not a problem of a single power generation node, but a result of interaction of a group of power generation nodes with similar behavior patterns and similar resonance risk levels. Therefore, in the design of the target function, the norm of the resonance change vector between the mth power generation node and the nth cluster center at time t is based on the degree to which the mth power generation node belongs to the nth cluster center at time t to ensure that the behavior patterns of the power generation nodes in the group are similar. The smaller the norm, the more similar the harmonic change between the mth power generation node and the nth cluster center, and the more the mth power generation node should be divided into a group. At the same time, to ensure that the risk levels of all power generation nodes in the same cluster are on the same echelon. Through double-constraint clustering, power generation nodes that are in a group and have rapidly rising group risk are accurately identified.

[0078] Further, the embodiment randomly selects a preset number of power generation nodes from all power generation nodes as initial cluster centers, takes all initial cluster centers as the input of the fuzzy C-means clustering algorithm, takes J(t) as the target function of the fuzzy C-means clustering algorithm, takes the difference between the resonance feature vectors of the power generation nodes as the measurement distance of the fuzzy C-means clustering algorithm, and takes the cluster center corresponding to the minimum value of the target function as the optimal cluster center.

[0079] It should be noted that the difference between the resonance feature vectors of the power generation nodes is taken as the measurement distance of the fuzzy C-means clustering algorithm in the embodiment, which actually takes the Euclidean distance between the resonance feature vectors of the power generation nodes as the measurement distance of the fuzzy C-means clustering algorithm.

[0080] The calculation method of the Euclidean distance and the fuzzy C-means clustering algorithm are known technologies, and the calculation process of the Euclidean distance and the specific process of clustering the power generation nodes by using the fuzzy C-means clustering are not described again.

[0081] So far, the embodiment accurately divides the power generation nodes with similar and similar risks into the same group by constructing an innovative objective function that simultaneously combines the node behavior pattern similarity and the risk level closeness. The method not only overcomes the defects of traditional clustering ignoring the absolute value difference of risk, effectively eliminates the interference of accidental fluctuations of a single node, but also accurately identifies the power generation node clustering cluster that "clusters" and risks synchronously and rapidly rises, thereby significantly improving the accuracy of resonance risk positioning and the reliability of early warning.

[0082] S2.3: Analyze the change trend of the harmonic corresponding current signal of each power generation node and each optimal clustering center at the current time respectively, and determine the resonance speedup characteristic value of each harmonic of each power generation node at the current time in combination with the degree to which each power generation node belongs to each optimal clustering center, to determine the resonance risk characteristic value for resonance detection of the power generation node at the current time.

[0083] Based on the optimal clustering center obtained in S2.2, the embodiment further analyzes the change trend of the harmonic corresponding current signal of each power generation node and each optimal clustering center at the current time, and determines the resonance speedup characteristic value of each harmonic of each power generation node at the current time in combination with the degree to which each power generation node belongs to each optimal clustering center, to determine the resonance risk characteristic value for resonance detection of the power generation node at the current time, the specific process is as follows:

[0084] In the embodiment, all current data in the harmonic corresponding current signal of each power generation node and each optimal clustering center at the current time are fitted respectively to obtain the fitting straight line of each harmonic of each power generation node and each optimal clustering center;

[0085] It should be noted that there are many commonly used fitting methods, and in the embodiment, the least squares method is used to fit all current data in the harmonic corresponding current signal. In actual application, as other implementation manners, the implementer can also use other fitting methods such as polynomial fitting algorithm according to the specific situation, and the selection of the fitting algorithm is not specially limited in the embodiment.

[0086] Among them, the least squares method is a known technology, and the specific process of using it to fit the current data will not be repeated.

[0087] Further, the embodiment calculates the product of the degree to which each power generation node belongs to each optimal clustering center at the current time and the slope of the fitting straight line of each harmonic of the corresponding optimal clustering center, which is recorded as the comprehensive product between each power generation node and each optimal clustering center at each harmonic;

[0088] Further, the sum of the fitting straight line slope of each generating node of each harmonic at the current time and the corresponding accumulated result is taken as the resonance speed-up characteristic value of each generating node of each harmonic at the current time.

[0089] According to the resonance speed-up characteristic value of each generating node of each harmonic at the current time, the resonance speed-up characteristic value quantifies the speed and trend of the harmonic pollution and deterioration of the generating node, and is used to represent the dynamic intensity of the resonance risk development. If the resonance speed-up characteristic value of the current generating node is larger, it indicates that the harmonic problem of the current generating node is deteriorating at a faster speed. Conversely, if the resonance speed-up characteristic value of the current generating node is smaller, it indicates that the harmonic problem of the current generating node is deteriorating at a slower speed, or even tends to be stable or is being alleviated, indicating that the resonance risk is in a controllable or weakened state.

[0090] Further, based on the resonance speed-up characteristic value, the resonance risk characteristic value is determined for the resonance detection of the generating node at the current time, specifically:

[0091] If the resonance speed-up characteristic value of the zth harmonic of the qth generating node at the current time is less than or equal to 0, the resonance risk characteristic value of the zth harmonic of the qth generating node at the current time is set to 0. Otherwise, the difference between the preset harmonic current allowable value of the zth harmonic of the qth generating node and the effective value of the zth harmonic amplitude of the qth generating node is calculated, and the ratio of the resonance speed-up characteristic value of the zth harmonic of the qth generating node to the difference is taken as the resonance risk characteristic value of the zth harmonic of the qth generating node at the current time.

[0092] It should be noted that the resonance risk characteristic value quantifies the possibility of the resonance risk of the generating node. If the difference between the preset harmonic current allowable value of the zth harmonic of the qth generating node and the effective value of the corresponding harmonic amplitude is smaller, and the resonance speed-up characteristic value is larger, it indicates that the possibility of the resonance risk of the qth generating node is larger, and the corresponding resonance risk characteristic value is larger. Conversely, if the difference between the preset harmonic current allowable value of the zth harmonic of the qth generating node and the effective value of the corresponding harmonic amplitude is larger, and the resonance speed-up characteristic value is smaller, it indicates that the possibility of the resonance risk of the qth generating node is smaller, and the corresponding resonance risk characteristic value is smaller.

[0093] If the effective values of all the harmonic amplitudes of the qth power generation node at the current time are less than the respective preset harmonic current allowable values, and the resonance risk characteristic values of all the harmonics of the qth power generation node do not exceed the preset resonance risk threshold value, the qth power generation node at the current time does not have a resonance risk, otherwise, the qth power generation node at the current time has a resonance risk, and all the power generation nodes are traversed to perform resonance detection.

[0094] It should be noted that the preset harmonic current allowable value can be valued with reference to the harmonic current allowable values of each harmonic in the GB / T 14549-93 standard, which will not be listed one by one in the embodiment.

[0095] It should be noted that the value of the preset resonance risk threshold value in the embodiment is artificially set, and the value of the preset resonance risk threshold value in the embodiment is 10. In actual application, as other implementation manners, the implementer can also set it by himself according to the specific situation, and the embodiment does not have special restrictions.

[0096] So far, by analyzing the harmonic change trend of the power generation node and the cluster center to which it belongs, the embodiment calculates the accurate resonance speedup characteristic value, and combines the harmonic over-standard risk distance to finally calculate the resonance risk characteristic value that can quantify the resonance risk possibility, so as to realize the rapid and accurate detection and early warning of the potential resonance risk of the multi-inverter grid-connected system, and improve the accuracy and reliability of the resonance risk positioning.

[0097] Based on the same inventive concept as the above method, the embodiment of the present application also provides a resonance detection system for new energy power generation, which comprises a memory, a processor and a computer program stored in the memory and running on the processor. The processor implements the steps of any one of the methods in the above resonance detection method for new energy power generation when executing the computer program.

[0098] It should be noted that the above sequence of the embodiments is only for description, and does not represent the advantages and disadvantages of the embodiments. The above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0099] Each embodiment in the present application is described in a progressive manner, and the same or similar parts of each embodiment can be referred to each other. Each embodiment mainly describes the differences from other embodiments.

[0100] The above description is only the preferred embodiment of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. made within the principle of the present application should be included in the protection scope of the present application.

Claims

1. A resonance detection method for new energy power generation, characterized in that, The method includes the following steps: The current data of each power generation node within a preset time period before each moment is acquired in real time, and the current signal of each power generation node at each moment is composed, and the fundamental frequency and all subharmonics of the current signal are acquired. Based on the effective values ​​of the fundamental wave and all subharmonic amplitudes of the current signal at each power generation node at each time point, and the fluctuation degree of the current signal corresponding to each subharmonic, the harmonic distortion rate of each power generation node at each time point is determined to determine the first resonance value of each power generation node at each time point. In combination with the difference of the effective values ​​of all subharmonic amplitudes in the current signal of each power generation node between each time point and the previous time point, the resonance characteristic vector and resonance change vector of each power generation node at each time point are constructed. Based on the differences in the resonant feature vectors among all power generation nodes at each time point, all power generation nodes are clustered. Based on the distance between each power generation node and each cluster center, the distance between all cluster centers, and the difference in the first resonance value between each power generation node and each cluster center, the feature weight between each power generation node and each cluster center at each time point is determined. Combining the differences in the resonant change vectors and the first resonance values ​​between each power generation node and each cluster center at each time point, as well as the degree to which each power generation node belongs to each cluster center, the objective function of the clustering algorithm is determined, so as to iteratively optimize the clustering algorithm to obtain the optimal cluster center. The changing trends of the harmonic current signals of each power generation node and each optimal cluster center at the current moment are analyzed separately. Combined with the degree to which each power generation node belongs to each optimal cluster center, the characteristic value of the resonant growth rate of each harmonic of each power generation node at the current moment is determined, so as to determine the resonant risk characteristic value and perform resonance detection on the power generation node at the current moment. The method for determining the characteristic value of the resonant growth rate of each harmonic at each power generation node at the current moment is as follows: Fit all current data in the current signals corresponding to each harmonic of each power generation node and each optimal cluster center at the current time, and obtain the fitting straight lines for each harmonic of each power generation node and each harmonic of each optimal cluster center. Calculate the product of the degree to which each power generation node belongs to each optimal cluster center at the current time and the slope of the fitting line of each harmonic of the corresponding optimal cluster center. This product is recorded as the comprehensive product of each power generation node and each optimal cluster center on each harmonic. Calculate the cumulative result of the comprehensive product of each power generation node and all optimal cluster centers on each harmonic at the current time. Take the slope of the fitted straight line of each harmonic of each power generation node at the current time and the sum of the corresponding cumulative results as the characteristic value of the resonant growth rate of each harmonic of each power generation node at the current time. 2.The new energy power generation oriented resonance detection method according to claim 1, characterized in that, The expression of the harmonic distortion rate of each power generation node at each time is: ; in the formula, represents the harmonic distortion rate of the jth power generation node at time i; , respectively represent the effective value of the fundamental amplitude and the effective value of the amplitude of the hth harmonic in the current signal of the jth power generation node at time i; represents the discrete degree of all current data in the hth harmonic current signal of the jth power generation node at time i; represents the number of all harmonics in the current signal of the power generation node at time i. 3.The method of claim 1, wherein, The first resonance value of each power generation node at each time point is the average harmonic distortion rate of each power generation node at all times within a preset time period prior to each time point. 4.The method of claim 1, wherein, The method for determining the resonance change vector of each power generation node at each time point is as follows: The total difference between the effective values ​​of all subharmonic amplitudes of each power generation node at each time point and the previous time point is denoted as the second resonance value of each power generation node at each time point. The first resonance value of each power generation node at each time point and the second resonance value of the corresponding power generation node at all times within the preset time period before each time point are used to form the resonance feature vector of each power generation node at each time point. The second resonance value of each power generation node at all times within the preset time period before each time is used to form the resonance change vector of each power generation node at each time. 5.The method for new energy power generation oriented resonance detection according to claim 1, characterized in that, The expression of the feature weight between each power generation node and each cluster center at each time point is: ; The feature weight between the mth power generation node and the nth cluster center at time t is represented as The distance between the mth power generation node and the nth cluster center at time t is represented as The maximum distance between all cluster centers at time t is represented as The difference between the first resonance value between the mth power generation node and the nth cluster center at time t is represented as norm[ ] represents a normalization function. is a preset value. 6.The new energy power generation oriented resonance detection method according to claim 1, characterized in that, The method for determining the objective function of the clustering algorithm is as follows: Objective function about time t The expression is: ; In the formula, indicates the degree to which the mth power generation node belongs to the nth cluster center at time t; indicates the norm of the resonance change vector between the mth power generation node and the nth cluster center at time t; indicates the feature weight between the mth power generation node and the nth cluster center at time t; indicates the difference between the first resonance value between the mth power generation node and the nth cluster center at time t; indicates the number of all power generation nodes; indicates the number of all cluster centers; indicates a preset fuzzy index. 7.The method of claim 6, wherein, The iterative optimization of the clustering algorithm to obtain the optimal cluster centers includes: A predetermined number of power generation nodes are randomly selected from all power generation nodes as initial cluster centers. All initial cluster centers are used as inputs to the fuzzy C-means clustering algorithm. J(t) is used as the objective function of the fuzzy C-means clustering algorithm, the difference in the resonant feature vectors between power generation nodes is used as the distance metric of the fuzzy C-means clustering algorithm, and the cluster center corresponding to the minimum value of the objective function is used as the optimal cluster center. 8.The method of claim 1, wherein, The determination of resonance risk characteristic values ​​for resonance detection of the power generation node at the current moment includes: If the resonant growth rate characteristic value of the z-th harmonic of the q-th power generation node at the current time is less than or equal to 0, then the resonant risk characteristic value of the z-th harmonic of the q-th power generation node at the current time is set to 0. Otherwise, the difference between the preset allowable value of the harmonic current of the z-th harmonic of the q-th power generation node and the effective value of the amplitude of the z-th harmonic of the q-th power generation node is calculated, and the ratio of the resonant growth rate characteristic value of the z-th harmonic of the q-th power generation node to the difference is taken as the resonant risk characteristic value of the z-th harmonic of the q-th power generation node at the current time. If, at the current moment, the effective values ​​of the amplitudes of all harmonics of the q-th power generation node are less than their respective preset harmonic current allowable values, and the resonance risk characteristic values ​​of all harmonics of the q-th power generation node do not exceed the preset resonance risk threshold, then the q-th power generation node does not have resonance risk at the current moment; otherwise, the q-th power generation node has resonance risk at the current moment. Traverse all power generation nodes and perform resonance detection on all power generation nodes. 9.A resonance detection system for new energy power generation, comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the resonance detection method for new energy power generation as described in any one of claims 1-8.

Images