Transient voltage stability evaluation method based on combination of disturbance signal energy and decision tree

Abstract

The invention discloses a transient voltage stability evaluation method based on the combination of disturbance signal energy and a decision tree, and the method comprises the following steps: obtaining the voltage quantity of wide-area measurement information of each node in a power system, converting the voltage quantity into a corresponding disturbance signal energy value, building a model for evaluating the transient voltage stability based on an ID3 decision tree algorithm, and carrying out the evaluation of the transient voltage stability. And when short-term large disturbance occurs in the power system, measuring and obtaining a voltage dynamic time sequence by using each node PMU in the power system, calculating voltage disturbance signal energy obtained by converting each voltage time sequence, and comparing the voltage disturbance signal energy with a disturbance signal energy splitting threshold value of each node in the constructed decision tree to obtain an evaluation result. According to the method, the transient voltage stability of the power system is evaluated, the training time cost of the model is greatly reduced, the problem that voltage instability and voltage stability of different forms are easily mixed in an existing method is solved, and the evaluation accuracy of the model on the transient voltage stability of the power system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid security monitoring, and in particular to a transient voltage stability assessment method based on the combination of disturbance signal energy and a decision tree. Background Art

[0002] With the large-scale grid connection of new energy power generation equipment and the widespread application of power electronic equipment in power systems, the scale and complexity of power grids have increased, and the voltage regulation characteristics and reactive power distribution of power systems have changed significantly. This change makes the system more prone to transient voltage instability when disturbed, which seriously threatens the safe and stable operation of the power system. After a transient short-circuit fault occurs in the power system, the system voltage will drop briefly and rise rapidly after the fault is removed. Whether the system voltage can be quickly and accurately predicted to be able to restore stability after the fault is removed is an important task for evaluating the transient voltage stability of the system. In order to solve this type of problem, domestic and foreign scholars have carried out a lot of research. At present, the main methods can be divided into: one is an analysis method based on physical modeling, which relies on detailed model parameters of the system, such as time domain simulation method, power flow analysis method, energy function method, etc.; the other is a data-driven method based on wide-area measurement information, such as transient evaluation method based on voltage Shapelet time series trajectory feature learning. These methods have their own advantages and disadvantages, and the evaluation needs for specific scenarios still need further research and optimization.

[0003] However, as the scale of grid interconnection continues to expand, it becomes increasingly difficult to obtain detailed models and accurate parameters of the actual system, which limits the application of the first type of method. Although the second type can only use the data obtained from the wide area measurement system (WAMS) to learn the voltage time series trajectory features to evaluate transient stability, when many time series data sets have high dimensions and large sample sizes in actual applications, the model training time cost of shapelet-based time series trajectory feature learning is huge, which seriously affects the efficiency of time series feature learning and construction of transient voltage assessment models; and the time series trajectory feature learning based on voltage shapelet is based on the voltage time series curve shape for model training, and there are many forms of transient voltage instability, such as continuous low voltage, overvoltage, divergent oscillation, etc. The single optimal shapelet sequence obtained by the voltage shapelet-based time series trajectory feature learning method cannot distinguish multiple transient voltage instabilities from transient voltage stability, resulting in a decrease in assessment accuracy.

[0004] In recent years, the signal energy method has been widely used in the fields of machinery, power, and electricity, such as structural health monitoring, vibration analysis, fault diagnosis, and energy management systems. The theory of the signal energy method is to extract a certain signal from the time-domain simulation results of the system under study, and define the signal energy as the integral of the square of the transient component of the signal over the simulation time, while the corresponding disturbance signal energy is the integral of the square of the difference between the transient component and the steady-state component of the signal over the simulation time. In this theory, the disturbance signal energy is used to measure the energy accumulation of the signal after being disturbed over a period of time. It belongs to one-dimensional scalar data, and the time cost of its feature learning is very small. Combining it with binary classification machine learning methods, such as decision trees, provides a new idea for using data-driven methods based on wide-area measurement information to evaluate the transient voltage stability of power systems.

[0005] In summary, there is an urgent need for a transient voltage stability assessment technology based on the combination of disturbance signal energy and decision tree, which should be applied to the research on safe and stable operation of power systems, so as to quickly and accurately realize the transient voltage stability assessment of the system after fault removal while greatly reducing the time cost of training the model. Summary of the invention

[0006] In order to overcome the defects and shortcomings of the prior art, the present invention provides a transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree. The present invention obtains the voltage of the wide-area measurement information of each node in the system, converts it into the corresponding disturbance signal energy value, and then uses the ID3 decision tree algorithm to construct a binary classification assessment model that can be used to assess the transient voltage stability of the system. It can predict whether the voltage curve can be restored to stability after a few seconds (generally 8 to 10 seconds in engineering) in a very short time after the fault is removed, thereby achieving the assessment of the transient voltage stability of the system and gaining more time for the operating personnel to take relevant control measures. The decision tree model training based on the disturbance signal energy greatly reduces the training time cost, and the transient voltage instability and stability of the system are assessed by the energy size, which can avoid the decrease in assessment accuracy caused by different voltage instability forms.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] The present invention provides a transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree, comprising the following steps:

[0009] For the selected power system, a simulation case is generated through numerical simulation, and the time-series voltage data of each node of the power system after the fault is removed is obtained as a data sample set, and each sample is normalized;

[0010] According to the voltage time series sample data, the difference between the transient voltage amplitude after fault removal and the voltage steady-state value of the corresponding node is calculated. Based on the signal energy method, the disturbance signal energy accumulated by each node in each sample within the set time period after fault removal is obtained. The time series data set is converted into a disturbance energy data set, and the disturbance energy data set is sorted and averaged to obtain a candidate set of disturbance signal energy splitting thresholds.

[0011] Taking the disturbance energy data set as the original data object and the disturbance signal energy splitting threshold candidate set as the condition, the classification learning is carried out based on the ID3 decision tree algorithm, and a decision tree model is established as the classification evaluation model of transient voltage stability. Each node in the tree corresponds to the disturbance signal energy splitting threshold of a certain node.

[0012] When it is determined that a short-term large disturbance occurs in the power system, the dynamic time series is obtained based on the PMU measurement of each node in the power system, and the disturbance signal energy obtained by converting each dynamic time series is calculated. It is compared with the disturbance signal energy splitting threshold of each node in the decision tree model to obtain the evaluation result.

[0013] As a preferred technical solution, the simulation case is generated by numerical simulation, the time series voltage data of each node of the power system after the fault is removed is obtained as a data sample set, and each sample is normalized. The specific steps include:

[0014] Generate n simulation cases through transient numerical simulation, one simulation case is one sample, one sample contains d nodes, and extract the voltage amplitude trajectory of each node;

[0015] Taking the fault removal time as the starting point, given the time window length T, the sampling time Δt, the time window is used to extract the per-unit time series of each node voltage, and the length of each sequence is m = T / Δt, and the transient voltage time series data set S of the d-dimensional power system after the fault removal is obtained. raw , S raw ={S raw1 ,S raw2 ,…,S rawd}, n is the number of samples, d is the number of nodes included in the power system, and m is the length of a single time series data;

[0016] The time series data set sample S raw Each data is divided by the reference voltage U of the corresponding node n , and obtain the normalized voltage time series data set S = {S1, S2, …, S d}={S raw1 / U n ,S raw2 / U n ,…,S rawd / Un};

[0017] The i-th dimension of the voltage time series data set S is a data set consisting of n time series with a length of m, that is:

[0018] S i = {U i,1 ,U i,2 ,...,U i,n},i=1,2,...d

[0019] Among them, U i,j = {u i,j (t1),u i,j (t2),…,u i,j (t m )},j=1,2,…,n,u i,j (t1) represents the value of the voltage of node i in the jth sample corresponding to the first measurement point.

[0020] As a preferred technical solution, the signal energy method is used to obtain the disturbance signal energy accumulated by each node in each sample within a set time period after the fault is removed, and the time series data set is converted into a disturbance energy data set. The specific steps include:

[0021] Based on the signal energy method, the transient voltage time series data set is converted into a voltage disturbance signal energy set, and each time series is converted into the disturbance signal energy generated in the T time period, which is expressed as:

[0022]

[0023] Where u(0) is the initial value of the voltage steady state, m is the number of measurement points in the time period considered in the timing trajectory, Δt is the time interval between two consecutive measurements, and u(t i ) is the voltage amplitude data corresponding to the i-th measurement point;

[0024] The time series data sets corresponding to all nodes of the voltage time series data set S are transformed to obtain the disturbance energy data set, which is expressed as:

[0025]

[0026] Among them, each element in the matrix of the perturbation energy data set is a scalar, and each element E i,j It represents the energy of the disturbance signal generated by the voltage of a node in the scene corresponding to a sample in a given time series.

[0027] As a preferred technical solution, the disturbance energy data set is sorted and averaged to obtain a candidate set of disturbance signal energy splitting thresholds, which specifically includes:

[0028] For each element E i,j As the characteristic attribute of the corresponding node i in the corresponding sample j, the perturbation signal energy set E of a single node i The energy values ​​of n samples in E are sorted from small to large to get i sort ={E' i,1 ,E' i,2 …,E' i,n}, calculate E' i,1 The average value between two adjacent elements in the node is used to obtain the candidate set of split thresholds for the energy of the node disturbance signal. The candidate set of split thresholds for the voltage disturbance signal energy of each node in the power system under a certain number of samples is obtained, which is expressed as:

[0029]

[0030] Among them, the split threshold candidate set E of the voltage disturbance signal energy is mean _ sort There are n-1 rows in total, representing n-1 mean data samples obtained by taking the mean of n adjacent samples. Each column represents the same system node under different samples. Each element is a scalar, which serves as the condition for the decision tree model based on conditional information entropy to divide the original data.

[0031] As a preferred technical solution, the disturbance energy data set is used as the original data object, the disturbance signal energy splitting threshold candidate set is used as the condition, and classification learning is performed based on the ID3 decision tree algorithm to establish a decision tree model as a classification evaluation model for transient voltage stability, which specifically includes:

[0032] According to the stable and unstable labels of the disturbance signal energy set E, the data samples in the disturbance signal energy set E are divided into stable and unstable categories, denoted as C1 and C2 respectively, and their proportions in the disturbance signal energy set E are p1 and p2 respectively. The original information entropy of the disturbance signal energy set E is calculated:

[0033]

[0034] Among them, Ent(E) represents the information entropy of the disturbance signal energy set E. For each node bus in the d-dimensional power system, Choose a candidate energy splitting threshold from When the disturbance signal energy set E is divided, The disturbance signal energy set E is divided into two subsets, denoted as E less , E greater , E less Every element in dimension i is less than or equal to And E greaterEvery element in is greater than

[0035] Count subset E separately less and subset E greater The number of stable samples and unstable samples in the classification is used to calculate the conditional information entropy after classification. The calculation expression is as follows:

[0036]

[0037] Among them, p less and p greater E less and E greater The proportion of the number of samples in E;

[0038] pass The information gain obtained by partitioning the data set is:

[0039]

[0040] Splitting threshold candidate set from perturbation signal energy set Select the one with the largest information gain from the n-1 candidate energy splitting thresholds as the optimal energy splitting threshold for node i Obtain the optimal energy splitting threshold of all nodes in the power system, and find out the optimal energy splitting threshold of all nodes in the power system by comparing the information gain corresponding to the optimal energy splitting threshold of multiple nodes;

[0041] The next decision tree node is selected by maximizing the information gain and its corresponding energy split threshold. After the original data samples are divided into two branches according to the energy split threshold of the decision tree node, the corresponding rows of the samples that are not in the branch in the original disturbance signal energy set E should be deleted according to the samples contained in the new branch, and the corresponding disturbance signal energy set E' should be updated again to serve as the original disturbance energy data set for the next decision tree node screening;

[0042] The decision tree classifies the disturbance signal energy splitting threshold candidate set E through multi-layer if-then rules. mean _sort After input, the data of each dimension are compared starting from the root node, and the optimal attribute is selected as the node according to the information gain. The data is recursively divided step by step until each subset contains only samples of the same type or reaches the preset threshold. The leaf nodes store the labels of the same type of samples or the labels of the majority of samples, and a decision tree model is established as a classification evaluation model for transient voltage stability.

[0043] As a preferred technical solution, the energy of the disturbance signal obtained by converting each dynamic time series is calculated and compared with the disturbance signal energy splitting threshold of each node in the decision tree model to obtain an evaluation result, which specifically includes:

[0044] Based on the PMU measurement of each node in the power system, the voltage dynamic time series is obtained, the voltage disturbance signal energy obtained by the transformation of each voltage time series is calculated, and compared with the disturbance signal energy splitting threshold of each node in the decision tree, and the path search and matching are performed from top to bottom. When the path reaches the terminal (leaf) node, the evaluation result is given by the class label in the terminal node. If the power system is judged to be unstable, an alarm signal is issued; if the power system is judged to be stable, monitoring continues.

[0045] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0046] (1) The present invention combines the data-driven method based on the wide-area measurement system with the disturbance signal energy, converting the original two-dimensional trajectory-type time series data set into a one-dimensional flat energy data set, which greatly reduces the training time cost of constructing a data-driven transient voltage stability assessment model and improves the efficiency of constructing the assessment model.

[0047] (2) In view of the fact that the energy of the voltage disturbance signal can fully reflect the difference in the size of the accumulated energy of the node voltage during transient voltage instability and stability, the data-driven method based on a large amount of sample data can perform feature learning on different transient voltage instability and stability situations, thereby evaluating whether the subsequent node voltage can restore stability within a short time (within 1 second) after the fault is removed based on the learned stability characteristics; and combining the disturbance signal energy with the decision tree algorithm, it not only has clear physical interpretability, but also improves the problem that the original shapelet sequence method based on the voltage time series trajectory is easy to confuse different forms of voltage instability and voltage stability, thereby improving the accuracy of the model in evaluating the transient voltage stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 Schematic diagram of the process of transient voltage stability assessment method based on disturbance signal energy and decision tree in this embodiment;

[0049] Figure 2 Schematic diagram of the architecture of the decision tree model of this embodiment;

[0050] Figure 3 This is a topology diagram of the IEEE standard 39-node system in this embodiment;

[0051] Figure 4 A classification assessment model for transient voltage stability constructed for this embodiment;

[0052] FIG5( a ) is a schematic diagram of the voltage amplitude trajectory and the voltage disturbance signal energy accumulation under the transient voltage stability condition of this embodiment;

[0053] FIG5( b ) is a schematic diagram of the voltage amplitude trajectory and the voltage disturbance signal energy accumulation in the case of transient voltage instability in this embodiment. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0055] Example

[0056] like Figure 1 As shown, the present invention provides a transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree, comprising the following steps:

[0057] S1. For a given power system, considering its various possible operating conditions and accident types, a large number of simulation cases are generated through numerical simulation, and the time-series voltage data of each node of the power system after the fault is removed is obtained as a data sample set, and each sample is normalized;

[0058] (1) For the power system under study, various possible operating conditions and accident types are considered. Through transient numerical simulation, n simulation cases are generated. One simulation case is a sample, and one sample contains d nodes.

[0059] Based on the standard operation scenario of the power system, under the premise of system flow convergence, the system load is randomly set between 85% and 125% of the standard scenario; the randomness of wind power output is simulated by the Monte Carlo method; the synchronous machine is randomly set to 90% to 100% of its rated capacity. The fault setting scheme is: three-phase short-circuit faults are randomly set at 10%, 30%, 60%, 90% and other locations on all transmission lines from the head end, the fault occurrence time is set to 0s, and the fault removal time is fixed to 0.1s.

[0060] (2) Under the above conditions, data samples are generated through transient numerical simulation and the voltage amplitude trajectory of each node is extracted. Taking the fault removal time as the starting point, the time window length is given as T, the sampling time is Δt, and the time window is used to extract the voltage U time series of each node from each sample. The length of each sequence is m = T / Δt, thus obtaining the transient voltage time series data set S of the d-dimensional (i.e., containing d nodes) power system after the fault removal. raw , S raw ={S raw1 ,S raw2 ,…,Srawd}, n is the number of samples, d is the number of nodes included in the power system, and m is the length of a single time series data;

[0061] Specifically, the time window length is given as T, and the measurement interval Δt is the same as the electromechanical transient simulation step length of 0.01s. The time window is used to extract the per-unit time series of the voltage of each node. The length of each sequence is m = T / Δt, T is 0.8s, and m = 80. 800 data samples are randomly generated through transient numerical simulation and used as the data sample set S raw , The time series data set sample S raw Each data is divided by the reference voltage U of the corresponding node n , and obtain the normalized voltage time series data set S = {S1, S2, …, S 39}.

[0062] The time series data set sample S raw Each data is divided by the reference voltage U of the corresponding node n , and obtain the normalized voltage time series data set S = {S1, S2, …, S d}={S raw1 / U n ,S raw2 / U n ,…,S rawd / U n The i-th dimension (1≤i≤d) of the voltage time series data set S, a single-dimensional data set (i.e., a transient voltage time series data set of a specific node) is a data set consisting of n time series of length m, that is:

[0063] S i = {U i,1 ,U i,2 ,...,U i,n},i=1,2,...d

[0064] Among them, U i,j = {u i,j (t1),u i,j (t2),…,u i,j (t m )},j=1,2,…,n,u i,j (t1) represents the value of the voltage of node i in the jth sample corresponding to the first measurement point.

[0065] S2. According to the voltage time series sample data obtained in step S1, the difference between the transient voltage amplitude after the fault is removed and the voltage steady-state value of the corresponding node is calculated; and the disturbance signal energy accumulated at each node in each sample within a set time period after the fault is removed is obtained by using the signal energy method, thereby converting the time series data set into a disturbance energy data set, further sorting and averaging the disturbance energy data set, and obtaining a candidate set of disturbance signal energy splitting thresholds, which specifically includes:

[0066] (1) According to the signal energy method, the transient voltage time series data set S obtained in step S1 is converted into a voltage disturbance signal energy set. Each time series can be converted into the disturbance signal energy generated in the time period T according to the following formula:

[0067]

[0068] Where u(0) is the initial value of the steady-state voltage, m is the number of measurement points in the time period considered in the timing trajectory, Δt is the time interval between two consecutive measurements, i.e., the step length, and u(t i ) is the voltage amplitude data corresponding to the i-th measurement point;

[0069] After the transformation of n (preferably 800) time series is completed, the disturbance signal energy set E of a single node is obtained. i ={E i,1 ,E i,2 ,…,E i,800 The same processing is performed on the time series data sets corresponding to all nodes in S to generate the disturbance signal energy set E, thereby completing the transformation of the original trajectory-type time series data set to the flat energy data set. The disturbance signal energy set E is expressed as:

[0070]

[0071] Each element in the above matrix is ​​a scalar, and each element E i,j It represents the energy of disturbance signal generated by the voltage of a node in a scene corresponding to a sample in a given time series;

[0072] (2) To disturb each element E in the signal energy set E i,j As the characteristic attribute of the corresponding node i in the corresponding sample j, the perturbation signal energy set E of a single node i The energy values ​​of the 800 samples are sorted from small to large to get E i sort ={E' i,1 ,E' i,2 …,E' i,800}, calculate E' i,1 The average value between two adjacent elements in the node is used to obtain the candidate set of split thresholds for the energy of the node disturbance signal. Similarly, after the same processing is performed on all nodes in the power system, the split threshold candidate set E of the voltage disturbance signal energy of each node under a certain number of samples is obtained. mean_sort , as follows:

[0073]

[0074] The candidate set E of split threshold of voltage disturbance signal energy mean_sort There are 799 rows in total, representing 799 mean data samples obtained by taking the mean of 800 adjacent samples. Each column represents the same system node under different samples, and each element is a scalar. These values ​​will be used as conditions for the subsequent decision tree algorithm based on conditional information entropy to divide the original data.

[0075] S3, taking the disturbance energy data set obtained in step S2 as the original data object, taking the disturbance signal energy splitting threshold candidate set as the condition, using the ID3 decision tree algorithm to perform classification learning, and establishing a decision tree model. The established decision tree model is a classification evaluation model for transient voltage stability, and each node in the tree corresponds to the disturbance signal energy splitting threshold of a node. The specific steps include:

[0076] (1) The disturbance signal energy set E obtained in step S2 is taken as the original data object. Since the evaluation of transient voltage stability in the present invention belongs to a binary classification problem, the data samples in the disturbance signal energy set E are divided into stable and unstable categories according to the stable and unstable labels of the disturbance signal energy set E, which are respectively denoted as C1 and C2. Their proportions in the disturbance signal energy set E are p1 and p2, respectively. The original information entropy of the disturbance signal energy set E is calculated:

[0077]

[0078] Among them, Ent(E) is called the information entropy of E. For each busbar in the d-dimensional power system, Choose a candidate energy splitting threshold (denoted as ) when E is divided, Divide E into two subsets, denoted by E less , E greater (E less Every element in dimension i is less than or equal to And E greater Every element in is greater than );

[0079] Specifically, the disturbance signal energy set splitting threshold candidate set E is obtained in step S2: mean_sort is the condition for dividing the original data. Figure 3As shown, for each bus in the given 39-bus power system, Choose a candidate energy splitting threshold (denoted as ) when E is divided, Divide E into two subsets, denoted by E less , E greater (E less Every element in dimension i is less than or equal to And E greater Every element in is greater than ).

[0080] (2) is the calculation subset E less and E greater The information entropy of the two subsets is calculated by counting the number of stable samples and unstable samples in each subset. less and E greater The respective information entropy of , and then calculate the conditional information entropy after classification, the calculation expression is as follows:

[0081]

[0082] Among them, p less and p greater E less and E greater The proportion of the number of samples in E;

[0083] Therefore, through After dividing the data set to obtain the conditional information entropy, the corresponding information gain is calculated as:

[0084]

[0085] The three formulas in step S3 are The 799 candidate energy splitting thresholds are selected as the optimal energy splitting threshold for node i. Similarly, the optimal energy splitting threshold of all nodes in the power system can be obtained. By comparing the information gain corresponding to the optimal energy splitting threshold of multiple nodes, the optimal energy splitting threshold of all nodes in the power system can be found. The next decision tree node is selected by maximizing the information gain and its corresponding energy splitting threshold. less and E greater The corresponding It will be further used to update the original information entropy and corresponding information gain of the disturbance signal energy set E;

[0086] (3) Figure 2 As shown, the decision tree is classified through multiple layers of if-then rules, and the disturbance signal energy splitting threshold candidate set Emean_sort After input, the model compares the data of each dimension starting from the root node, selects the optimal attribute (voltage node) as the node according to the information gain, and recursively divides the data step by step until each subset contains only samples of the same type or reaches the preset threshold. Finally, the leaf node stores the label of the same type of sample or the label of the majority of samples. Figure 4 As shown in the figure, the decision tree model constructed in this example has three internal nodes, which correspond to the disturbance energy splitting thresholds of nodes 5, 11, and 7 of the given power system. The category labels on the leaf nodes represent the stability judgment results, and -1 and 1 represent instability and stability respectively. The constructed classification model is tested for accuracy using k-fold cross validation, with k set to 10, and the cross validation accuracy is 99.8%, which shows that the model has ideal classification performance.

[0087] S4. When a short-term large disturbance occurs in the power system, the voltage dynamic time series is obtained by measuring the PMU of each node in the power system, and the voltage disturbance signal energy obtained by the transformation of each voltage time series is calculated. It is compared with the disturbance signal energy splitting threshold of each node in the decision tree constructed in step S3, and the path search and matching are performed from top to bottom. When the path reaches the terminal (leaf) node, the evaluation result is given by the class label in the terminal node. If the power system is judged to be unstable, an alarm signal is immediately issued; if the power system stability judgment result is stable, continuous monitoring is carried out.

[0088] (1) The same random fault scenario setting as in step S1 is adopted, and one line is randomly set for maintenance in the system line to simulate N-1 faults, and 400 transient simulation cases are randomly generated again as test samples.

[0089] (2) The voltage time series of nodes 5, 11, and 7 within 0.8s of fault removal are collected through the wide-area measurement system, and then the energy of the disturbance signal generated by this sequence and the steady-state value is calculated, and compared with the energy splitting threshold in the model from top to bottom until a leaf node is encountered, and the stability judgment result can be obtained. The overall accuracy of the test evaluation obtained is 99.25%, of which the false positive rate (the probability of evaluating stability as instability) is 0.5% and the missed positive rate (the probability of evaluating instability as stability) is 0.25%, which proves that the classification model can be used reliably and effectively for online monitoring. As shown in Figures 5(a) and 5(b), the typical system node voltage response curves of the two types of samples evaluated as transient voltage stability and transient voltage instability in this example are respectively shown, which intuitively presents the voltage response characteristics under different conditions. The local enlarged figure is a physical schematic diagram of the accumulated disturbance signal energy, which more clearly shows the difference between transient voltage stability and instability.

[0090] (3) In order to demonstrate the superiority of the present invention in terms of model building efficiency and evaluation accuracy, the method of the present invention is compared with the transient voltage stability evaluation method based on voltage shapelet time series feature learning using consistent parameters and environment. The performance comparison results are shown in Table 1:

[0091] Table 1 Comparison of training time and accuracy of the two methods

[0092]

[0093] In terms of model training time, compared with the shapelet-based method, the model training time of the method proposed in the present invention is reduced by two orders of magnitude, greatly shortened to less than 300 seconds, and greatly reduced the model training time cost. In terms of accuracy, compared with the shapelet-based method, the method proposed in the present invention has reduced the missed detection rate and the false detection rate, and improved the overall accuracy.

[0094] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A transient voltage stability assessment method based on disturbance signal energy and decision tree, characterized in that: The steps include: For the selected power system, a simulation case is generated through numerical simulation, and the time-series voltage data of each node of the power system after the fault is removed is obtained as a data sample set, and each sample is normalized; According to the voltage time series sample data, the difference between the transient voltage amplitude after fault removal and the voltage steady-state value of the corresponding node is calculated. Based on the signal energy method, the disturbance signal energy accumulated by each node in each sample within the set time period after fault removal is obtained. The time series data set is converted into a disturbance energy data set, and the disturbance energy data set is sorted and averaged to obtain a candidate set of disturbance signal energy splitting thresholds. Taking the disturbance energy data set as the original data object and the disturbance signal energy splitting threshold candidate set as the condition, the classification learning is carried out based on the ID3 decision tree algorithm, and a decision tree model is established as the classification evaluation model of transient voltage stability. Each node in the tree corresponds to the disturbance signal energy splitting threshold of a certain node. When it is determined that a short-term large disturbance occurs in the power system, the dynamic time series is obtained based on the PMU measurement of each node in the power system, and the disturbance signal energy obtained by converting each dynamic time series is calculated. It is compared with the disturbance signal energy splitting threshold of each node in the decision tree model to obtain the evaluation result.

2. The transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree according to claim 1 is characterized in that: The simulation case is generated by numerical simulation, the time series voltage data of each node of the power system after the fault is removed is obtained as a data sample set, and each sample is normalized. The specific steps include: Generate n simulation cases through transient numerical simulation, one simulation case is one sample, one sample contains d nodes, and extract the voltage amplitude trajectory of each node; Taking the fault removal time as the starting point, given the time window length T, the sampling time Δt, the time window is used to extract the per-unit time series of each node voltage, and the length of each sequence is m = T / Δt, and the transient voltage time series data set S of the d-dimensional power system after the fault removal is obtained. raw , S raw ={S raw1 ,S raw2 ,…,S rawd }, n is the number of samples, d is the number of nodes included in the power system, and m is the length of a single time series data; The time series data set sample S raw Each data is divided by the reference voltage U of the corresponding node n , and obtain the normalized voltage time series data set S = {S1, S2, …, S d }={S raw1 / U n ,S raw2 / U n ,…,S rawd / U n }; The i-th dimension of the voltage time series data set S is a data set consisting of n time series with a length of m, that is: S i ={U i,1 ,U i,2 ,...,U i,n },i=1,2,...d Among them, U i,j = {u i,j (t1),u i,j (t2),…,u i,j (t m )},j=1,2,…,n,u i,j (t1) represents the value of the voltage of node i in the jth sample corresponding to the first measurement point.

3. The transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree according to claim 1 is characterized in that: The signal energy method is used to obtain the disturbance signal energy accumulated by each node in each sample within a set time period after the fault is removed, and the time series data set is converted into a disturbance energy data set. The specific steps include: Based on the signal energy method, the transient voltage time series data set is converted into a voltage disturbance signal energy set, and each time series is converted into the disturbance signal energy generated in the T time period, which is expressed as: Where u(0) is the initial value of the voltage steady state, m is the number of measurement points in the time period considered in the timing trajectory, Δt is the time interval between two consecutive measurements, and u(t i ) is the voltage amplitude data corresponding to the i-th measurement point; The time series data sets corresponding to all nodes of the voltage time series data set S are transformed to obtain the disturbance energy data set, which is expressed as: Among them, each element in the matrix of the perturbation energy data set is a scalar, and each element E i,j It represents the energy of the disturbance signal generated by the voltage of a node in the scene corresponding to a sample in a given time series.

4. The transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree according to claim 3 is characterized in that: The disturbance energy data set is sorted and averaged to obtain a candidate set of disturbance signal energy splitting thresholds, which specifically includes: For each element E i,j As the characteristic attribute of the corresponding node i in the corresponding sample j, the perturbation signal energy set E of a single node i The energy values ​​of n samples in E are sorted from small to large to get i sort ={E' i,1 ,E' i,2 …,E' i,n }, calculate E' i,1 The average value between two adjacent elements in the node is used to obtain the candidate set of split thresholds for the energy of the node disturbance signal. The candidate set of split thresholds for the voltage disturbance signal energy of each node in the power system under a certain number of samples is obtained, which is expressed as: Among them, the split threshold candidate set E of the voltage disturbance signal energy is mean_sort There are n-1 rows in total, representing n-1 mean data samples obtained by taking the mean of n adjacent samples. Each column represents the same system node under different samples. Each element is a scalar, which serves as the condition for the decision tree model based on conditional information entropy to divide the original data.

5. The transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree according to claim 1 is characterized in that: Taking the disturbance energy data set as the original data object and the disturbance signal energy splitting threshold candidate set as the condition, classification learning is performed based on the ID3 decision tree algorithm, and a decision tree model is established as a classification evaluation model for transient voltage stability, which specifically includes: According to the stable and unstable labels of the disturbance signal energy set E, the data samples in the disturbance signal energy set E are divided into stable and unstable categories, denoted as C1 and C2 respectively, and their proportions in the disturbance signal energy set E are p1 and p2 respectively. The original information entropy of the disturbance signal energy set E is calculated: Among them, Ent(E) represents the information entropy of the disturbance signal energy set E. For each node bus in the d-dimensional power system, Choose a candidate energy splitting threshold from When the disturbance signal energy set E is divided, The disturbance signal energy set E is divided into two subsets, denoted as E less , E greater , E less Every element in dimension i is less than or equal to And E greater Every element in is greater than Count subset E separately less and subset E greater The number of stable samples and unstable samples in the classification is used to calculate the conditional information entropy after classification. The calculation expression is as follows: Among them, p less and p greater E less and E greater The proportion of the number of samples in E; pass The information gain obtained by partitioning the data set is: Splitting threshold candidate set from perturbation signal energy set Select the one with the largest information gain from the n-1 candidate energy splitting thresholds as the optimal energy splitting threshold for node i Obtain the optimal energy splitting threshold of all nodes in the power system, and find out the optimal energy splitting threshold of all nodes in the power system by comparing the information gain corresponding to the optimal energy splitting threshold of multiple nodes; The next decision tree node is selected by maximizing the information gain and its corresponding energy split threshold. After the original data samples are divided into two branches according to the energy split threshold of the decision tree node, the corresponding rows of the samples that are not in the branch in the original disturbance signal energy set E should be deleted according to the samples contained in the new branch, and the corresponding disturbance signal energy set E' should be updated again to serve as the original disturbance energy data set for the next decision tree node screening; The decision tree classifies the disturbance signal energy splitting threshold candidate set E through multi-layer if-then rules. mean_sort After input, the data of each dimension are compared starting from the root node, and the optimal attribute is selected as the node according to the information gain. The data is recursively divided step by step until each subset contains only samples of the same type or reaches the preset threshold. The leaf nodes store the labels of the same type of samples or the labels of the majority of samples, and a decision tree model is established as a classification evaluation model for transient voltage stability.

6. The transient voltage stability assessment method based on the combination of disturbance signal energy and decision tree according to claim 1 is characterized in that: The energy of the disturbance signal obtained by converting each dynamic time series is calculated and compared with the splitting threshold of the disturbance signal energy of each node in the decision tree model to obtain an evaluation result, which specifically includes: Based on the PMU measurement of each node in the power system, the voltage dynamic time series is obtained, the voltage disturbance signal energy obtained by the transformation of each voltage time series is calculated, and compared with the disturbance signal energy splitting threshold of each node in the decision tree, and the path search and matching are performed from top to bottom. When the path reaches the terminal (leaf) node, the evaluation result is given by the class label in the terminal node. If the power system is judged to be unstable, an alarm signal is issued; if the power system is judged to be stable, monitoring continues.

Images