Power system transient stability evaluation method and system considering data missing, and storage medium

Abstract

The invention discloses a power system transient stability evaluation method and system considering data missing, and a storage medium. The method comprises the following steps: step 1, data preprocessing; step 2, constructing a set of an optimal PMU combination; 3, training a transient stability evaluation integrated model; and 4, online adaptive transient stability evaluation is carried out. According to the method, the availability rate of a PMU combination is improved when data missing occurs, transient stability related information in time sequence features is fully explored by the model, and a historical evaluation result and an available sub-classifier evaluation result can be adaptively integrated and fused under PMU data missing; and the evaluation speed and the evaluation precision of the transient stability evaluation model of the power system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system automation, and in particular relates to a method, system and storage medium for evaluating transient stability of a power system taking into account data loss. Background Art

[0002] With the development of extensive grid interconnection, high penetration of renewable energy and liberalization of power market, the dynamic characteristics of power system are becoming more and more complex, so the risk of transient instability is also increasing. In order to prevent major losses caused by transient instability, transient stability assessment of power system is necessary.

[0003] With the installation of a large number of phasor measurement units (PMUs) in wide-area measurement systems and the development of artificial intelligence technology, data-driven transient stability assessment methods for power systems have received a lot of data support, making this type of method widely concerned by researchers. Data-driven transient stability assessment methods build a strong mapping relationship between certain features of the power system and the system stability state through model training, thereby achieving efficient and accurate transient stability assessment. However, in the data collection of actual power systems, PMU collection data may be missing due to communication delays, PMU failures, etc. In this case, the performance of the transient stability assessment model that lacks some features will be greatly reduced or even unusable. Therefore, a transient stability assessment method for power systems that can consider data missing due to PMU failures is proposed to enhance the tolerance of transient stability assessment to data missing situations, which has certain value. Summary of the invention

[0004] The present invention is made to solve the above-mentioned problem, and its purpose is to provide a method, system and storage medium for adaptive evaluation of transient stability of power system taking into account PMU data missing, which can perform adaptive transient stability evaluation when facing different degrees of data missing, thereby improving the accuracy of transient stability evaluation of power system after data missing occurs.

[0005] In order to achieve the above purpose, the present invention adopts the following scheme:

[0006] In a first aspect, the present invention provides a method for adaptively evaluating transient stability of a power system considering missing PMU data, the method comprising:

[0007] Step 1. Data preprocessing: Obtain all time series feature vectors collected within a preset period after the power system fault is removed under different situations, match them with the corresponding transient stable state, construct the original sample set, and then divide it into a training sample set and a verification sample set;

[0008] Step 2. Construct the set of optimal PMU combinations: Use the training sample set in step 1, calculate the importance of the time series features of the training sample set, construct the objective function of minimum risk optimization based on the importance of the time series features, and set the constraints that all network nodes are observable and there are still available PMU combinations under severe PMU failures to solve and obtain the set of optimal PMU combinations;

[0009] Step 3. Training of transient stability assessment integrated model: construct transient stability assessment sub-classifiers, obtain characteristic variable information and time information based on the transient stability assessment sub-classifiers and perform weighted fusion on them, use each optimal PMU combination in step 2 to correspond to a sub-classifier, use the time series features in the corresponding training sample set observed by the optimal PMU combination to train the corresponding sub-classifier, and then use the verification sample set to obtain the weight of each sub-classifier, and then construct a weighted integrated transient stability assessment integrated model;

[0010] Step 4. Online adaptive transient stability assessment: The PMU obtains the measurement data of the power system online, and obtains the final transient stability assessment result based on the transient stability assessment integrated model obtained in step 3.

[0011] Furthermore, the step 1 includes the following sub-steps:

[0012] Step 1.1: Based on the historical operation data of the power system and the time domain simulation records after the fault is cleared, max The active power and reactive power of the generator, the voltage and phase angle of the bus, the active power and reactive power of the line transmission, and the active power and reactive power of the load in the sampling period are used to construct the feature vector x containing the time series;

[0013] Step 1.2: The transient stability state y of the power system is set to 0 or 1, which represents the transient stability and transient instability of the system respectively. The transient stability state y under each operating state corresponds to the feature vector x, and the original sample set {x, y} is constructed. The original sample set {x, y} is randomly divided into the training sample set {x 1 ,y 1} and the validation sample set {x 2 ,y 2}.

[0014] Furthermore, the step 2 includes the following sub-steps:

[0015] Step 2.1, calculation of time series feature importance: using the training sample set {x 1 ,y 1}Iterate the importance calculation process of time series features, execute the loop m=1 until m=m max , where m is the number of iterative sampling, m maxis the maximum number of iterative sampling;

[0016] Determine whether the iterative cycle of the time series feature importance calculation is completed. If so, end the iterative process and obtain the weight of each time series feature calculated at the end as the importance measure of the feature, using h α Indicates the importance of the αth time series feature;

[0017] Step 2.2, minimum risk PMU combination optimization:

[0018] Based on the importance of PMU, the risk of unavailability of PMU combination is measured, and the objective function of minimum risk optimization is constructed to find the optimal PMU combination, where the objective function of minimum risk optimization is expressed as:

[0019]

[0020] Where K represents the number of PMU combinations, N represents the number of PMUs installed in the power grid, and I is an indicator function. represents the jth combination of i PMUs, x k represents the combination mode of the kth PMU combination, H k represents the importance of the kth PMU combination, P is the probability of PMU failure, and P i It represents the probability of i PMUs failing at the same time; represents the number of all combinations of any i out of N PMUs;

[0021] The indicator function I is specifically expressed as:

[0022]

[0023] The calculation formula for the importance of the k-th PMU combination is:

[0024]

[0025] Where Ω k represents all observable features of the kth PMU combination, h α represents the importance of the αth time series feature, which is calculated by sub-step 2.1;

[0026] The constraints are expressed as:

[0027]

[0028] Where f(·) represents the PMU observation range calculation function, f(x K-N+1 ) to f(x K ) is the observation range of each PMU in the N PMUs of the power grid obtained by the observation algorithm, f(x 1) to f(x K-N ) is the observation range of other combinations containing 2 or more PMUs, V k represents the observable node obtained by the kth PMU combination according to the PMU observation algorithm, q represents the number of nodes in the whole network, when the node is observable, V is 1, when the node is not observable, V is 0;

[0029] The optimal solution of the objective function of minimum risk optimization is obtained. When the risk value is minimized, a set of K PMU combinations is obtained. These combinations are the optimal PMU combinations.

[0030] Furthermore, the step 3 includes the following sub-steps:

[0031] Step 3.1, training a transient stability assessment sub-classifier: using a convolutional neural network for time series feature classification as a sub-classifier for transient stability assessment, wherein the one-dimensional convolutional neural network and the two-dimensional convolutional neural network in the sub-classifier are represented by the following functions:

[0032]

[0033] Where X represents the input feature, x 1 is a one-dimensional convolutional neural network C 1 (·) Get timestamp-related information, x D is a two-dimensional convolutional neural network, C 2 (·) The information obtained related to the observed characteristic variables; θ 1 and θ 2 are the parameters of one-dimensional and two-dimensional convolutional neural networks respectively;

[0034] Introduce the information weighted fusion mechanism and connect the weighted module. First, the timestamp-related information x 1 Information related to the observed characteristic variables x D The information after splicing is represented as x D+1 =[x D |x 1 ], then x D+1 The global time information is compressed into the input channel through the global average pooling layer, and then the information passes through the first fully connected layer with ReLU activation function and the second fully connected layer with Sigmoid activation function to obtain the information weight s D+1 ;

[0035] Using the information weights obtained D+1 Calculate the output of the attention module It is expressed as:

[0036]

[0037] at last The final output result of the entire network is obtained through the global average pooling layer and the softmax operator;

[0038] The optimizer learning rate of each neural network of the sub-classifier is set to η, and the process of updating each network parameter is:

[0039]

[0040] Where L 1 , L 2 , L W1 and L W2 They are the loss functions of the one-dimensional convolutional neural network, the two-dimensional convolutional neural network, the first fully connected layer in the weighted module, and the first fully connected layer, and all use the cross entropy loss function;

[0041] According to the optimal K PMU combinations obtained in step 2, each PMU combination corresponds to a sub-classifier, and the training sample set {x 1 ,y 1} features as the input features of the sub-classifier;

[0042] Execute loop i t =1 until i t =i t_max , iteratively train each sub-classifier separately;

[0043] Determine whether the iterative cycle of sub-classifier training is completed, if so, end the training iteration process to obtain K trained transient stability assessment sub-classifiers;

[0044] Step 3.2: Construct the transient stability assessment integrated model: Use the validation sample set {x 2 ,y 2} Obtain the weight of each sub-classifier in sub-step 3.1, and search for the corresponding weight of each sub-classifier when the prediction result of the verification sample set is closest to the true result. The specific process is expressed as the optimization solution process:

[0045]

[0046]

[0047] Where M represents the number of samples in the validation set, y i represents the true category of the i-th sample in the validation sample set, Represents the predicted category of the i-th sample in the validation sample set, ω k represents the weight of the k-th subclassifier, P(x ik ) represents the probabilistic output result of the i-th sample in the validation sample set in the k-th sub-classifier;

[0048] After the optimization solution is completed, the corresponding weight ω of each sub-classifier is obtained, and then the integrated model for transient stability assessment is constructed. The weights of each sub-classifier are used for weighted integration. In the case of no missing data, the final transient stability prediction result y given by the integrated model at time t is t ′ can be expressed as:

[0049]

[0050] In the formula represents the transient stability assessment output of the kth sub-classifier at the tth time.

[0051] Furthermore, the step 4 includes the following sub-steps:

[0052] Step 4.1, calculation of the output result of the integrated model at the current moment: Use the phasor measurement unit PMU in the wide-area measurement system to obtain the real-time measurement data of the power system during operation, and input it into the transient stability assessment integrated model obtained in step 3 for online assessment. When data missing occurs, it is assumed that the power system fault removal time is t c , which is the time when the model starts transient stability assessment, and the time when the PMU fails is t m , the longest transient stability assessment time is t max , the sub-classifier corresponding to the PMU combination where the faulty PMU causing data loss is located is classified as an unavailable sub-classifier, and the rest are classified as available sub-classifiers. The evaluation result obtained by the transient stability evaluation integrated model at the current moment is the weighted fusion of the evaluation result of the unavailable sub-classifier at the moment before the PMU failure and the evaluation result of the available sub-classifier at the current moment. The specific expression of this process is y t for:

[0053]

[0054] Where S represents the remaining available transient stability assessment sub-classifier after the PMU fails, represents the evaluation result of the i-th sub-classifier at the t-th time, Indicates that the jth subclassifier is m-1 The evaluation result at time t is that if the PMU fails at time t m ≤t c ,So The intermediate value of 0 and 1 is 0.5, which means that the sub-classifier cannot distinguish whether the input sample belongs to transient stability or transient instability because of the failure of PMU before the transient stability assessment begins;

[0055] Step 4.2, online evaluation final result determination: Set the observation window length of the transient stability evaluation integrated model to T, make predictions by inputting time series features with a time length of T, and define the transient instability judgment threshold τ∈(0.5,1). If the prediction result y of the model at the current moment is t ∈[τ,1], then the final evaluation result output at the current time t is transient instability. If y t ∈[0,1-τ], then the final evaluation result output at the current time t is transient stability. These two situations are called satisfying the threshold conditions. However, if y t ∈(1-τ,τ), that is, the threshold condition is not met, then it is determined that the current time t cannot give an accurate prediction result temporarily, and it is necessary to enter the next time t+1 to repeat the evaluation of step 4 until the threshold condition is met or the maximum evaluation time t is reached. max , the specific expression of transient stability Y predicted by online evaluation is:

[0056]

[0057] The final transient stability Y is the online transient stability assessment result under the condition of data loss caused by any number of PMU failures at any time.

[0058] On the other hand, the present invention provides a power system transient stability assessment system considering data loss, the system comprising:

[0059] The power system data preprocessing module is used to obtain all time series feature vectors collected after the power system fault is removed under different situations, match them with the corresponding transient stable state, construct the original sample set, and then divide it into a training sample set and a verification sample set;

[0060] The module for constructing the optimal PMU combination of the power system is used to obtain the set of PMU combinations with the lowest fault risk: the training sample set features in the data preprocessing module are used to calculate the importance of the time series features of the training sample set, and then based on the feature importance, a set of optimal PMU combinations that can be observed by all network nodes and still have available PMU combinations under severe PMU failures is constructed to minimize the fault risk;

[0061] The power system transient stability assessment integrated model training module is used to train sub-classifiers to build an integrated transient stability assessment model: a convolutional neural network for time series feature classification is used to construct a transient stability assessment sub-classifier, which can identify the observed characteristic variable information and the time information of the input data that is very important for prediction, and perform weighted fusion on the information. Each PMU combination obtained by the optimal PMU combination construction module corresponds to a sub-classifier, and the corresponding sub-classifier is trained using the time series features in the corresponding training sample set observed by the PMU combination. Then, the validation sample set is used to obtain the weight of each sub-classifier, and then a weighted integrated transient stability assessment integrated model is constructed;

[0062] The power system online adaptive transient stability assessment module is used to online assess the transient stability state of the power system when PMU data is missing: the PMU obtains the measurement data of the power system online, and the transient stability assessment integrated model obtained by the transient stability assessment integrated model training module performs online adaptive fusion of the output results of the unaffected sub-classifiers and the historical output results of the affected sub-classifiers according to the data missing caused by the PMU fault, thereby obtaining the final transient stability assessment result;

[0063] The power system transient stability assessment system considering data missing is used to execute the steps in the power system transient stability assessment method considering data missing.

[0064] Furthermore, it also includes:

[0065] An input display module, which is used to display the input, output and intermediate processing data of the corresponding part in the form of text, list, or two-dimensional / three-dimensional static / dynamic graph according to the operation instructions input by the operator;

[0066] The control module is used to communicate with the power system data preprocessing module, the power system optimal PMU combination construction module, the power system transient stability assessment integrated model training module, the power system online adaptive transient stability assessment module, and the input display module to control their operation.

[0067] On the other hand, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for transient stability assessment of a power system considering data missing.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] The present invention relates to a method, system and storage medium for adaptively evaluating transient stability of a power system considering missing PMU data. The optimal PMU combination constructed aims to minimize the risk of failure in various situations where different numbers of PMUs fail, thereby improving the availability of the PMU combination when data is missing, and ensuring that there are still available sub-classifiers for transient stability evaluation in severe data missing situations; secondly, the adopted sub-classifiers can extract information related to observed feature variables and timestamps in parallel, and enhance information mapping useful for transient stability classification through an information weighted fusion mechanism, more fully explore the transient stability related information of the power system in time series features, and at the same time improve the evaluation accuracy of the classifier; further, in the online transient stability evaluation stage, the historical evaluation results and the available sub-classifier evaluation results can be adaptively integrated and fused in the data missing scenario, and compared with other commonly used methods for processing missing data, the evaluation speed and evaluation accuracy of the transient stability evaluation model of the power system are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0071] Figure 1 A flowchart of a method for adaptively evaluating transient stability of a power system considering missing PMU data according to an embodiment of the present invention;

[0072] Figure 2 This is a time series feature importance ranking diagram involved in an embodiment of the present invention.

[0073] Figure 3 This is a risk value variation diagram of the optimal PMU combination optimization process with the minimum failure risk involved in an embodiment of the present invention.

[0074] Figure 4 It is a network structure diagram of a transient stability assessment classifier involved in an embodiment of the present invention.

[0075] Figure 5 The present invention is a block diagram of an online adaptive transient stability assessment of a power system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] It should be noted that, in the description of the embodiments of the present invention, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "include one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0078] The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0079] <Example 1>

[0080] like Figure 1 As shown, the power system transient stability adaptive assessment method considering PMU data loss provided in this embodiment includes the following steps:

[0081] Step 1: Data preprocessing. Obtain all time series feature vectors collected after the power system fault is removed under different situations, match them with the corresponding transient stable state, construct the original sample set, and then divide it into a training sample set and a validation sample set.

[0082] Step 1.1, in this embodiment, two wind farms with equal capacity are used to replace the synchronous generators of bus No. 34 and No. 36, and a photovoltaic solar field with equal capacity is used to replace the synchronous generator of bus No. 37, so that the penetration rate of wind and solar power generation reaches 30.2% of the IEEE-39 node system as the test power system, and the T after the fault is cleared is recorded. max= The active power and reactive power of the generator, the voltage and phase angle of the bus, the active power and reactive power of the line transmission, and the active power and reactive power of the load in 20 sampling periods are used to construct the feature vector x containing the time series.

[0083] Step 1.2: The transient stability state y of the power system is set to 0 or 1, which represents the transient stability and transient instability of the system respectively. The transient stability state y under each operating state corresponds to the feature vector x, and the original sample set {x, y} is constructed. The original sample set {x, y} is randomly divided into the training sample set {x 1 ,y 1} and the validation sample set {x 2 ,y 2}.

[0084] Step 2: Construct a set of optimal PMU combinations: Use the training sample set in step 1 to calculate the importance of the time series features of the training sample set, and then construct the objective function of minimum risk optimization based on the importance of the time series features. Set the constraints that all network nodes are observable and that there are still available PMU combinations in the event of a serious PMU failure, and solve them to obtain the set of optimal PMU combinations.

[0085] Step 2.1, calculation of time series feature importance.

[0086] In this embodiment, the importance of different features in the power system to transient stability assessment varies. The Relief algorithm can calculate feature weights to measure the importance of features. A variant algorithm of the Relief algorithm, Relief-FT, is used to calculate the importance of time series features.

[0087] Execute the loop from m = 1 to m = 10, using the training sample set {x 1 ,y 1}Iterative process of calculating the importance of time series features;

[0088] The iterative formula for calculating the importance of time series features is as follows:

[0089]

[0090] Where A t represents the time series feature, m represents the number of iterative sampling, k represents the number of nearest training samples, M j (C) indicates class The jth nearest training sample in the t ,R 1 ,R 2 ) represents different training samples R 1 , R 2 The same features of At The differences between them are specifically expressed as:

[0091]

[0092] Where ||R 1 [·]-R 2 [·]|| represents the bi-norm of the difference between two training samples;

[0093] Determine whether the iterative cycle of the time series feature importance calculation is completed. If so, end the iterative process and obtain the weight of each time series feature calculated at the end as the importance measure of the feature. α It represents the importance of the αth time series feature. The results of sorting the feature importance from large to small are as follows Figure 2 As shown;

[0094] Step 2.2, search for the optimal combination of PMU with minimum risk.

[0095] In this embodiment, in order to reduce the impact of data loss caused by PMU failure on transient stability assessment and resist the risk of unavailability of the assessment model, the PMU is divided into multiple combinations, and the acquisition features of each combination correspond to a sub-classifier. When data loss occurs, the sub-classifier that is not affected by the missing data can still be used to integrate the historical assessment results to complete the transient stability assessment;

[0096] In order to find the optimal PMU combination with the appropriate number, the risk of measuring the unavailability of the PMU combination is defined based on the importance of PMU. The objective function of the minimum risk optimization is expressed as:

[0097]

[0098] Where K represents the number of PMU combinations, N represents the number of PMUs installed in the power grid, and I is an indicator function. represents the jth combination of i PMUs, x k It represents the combination mode of the kth PMU combination, including H k represents the importance of the kth PMU combination, P is the probability of PMU failure, and P i It represents the probability of i PMUs failing at the same time, It represents the number of all possible combinations of any i PMUs from N PMUs.

[0099] The indicator function I is specifically expressed as:

[0100]

[0101] The calculation formula for the importance of the k-th PMU combination is:

[0102]

[0103] Where Ω k represents all observable features of the kth PMU combination, h α represents the importance of the αth time series feature, which is calculated by sub-step 2.1;

[0104] The constraints of the objective function must first ensure that the voltage of all nodes in the network is observable. Secondly, considering the different severity of PMU failures, the PMU combination set should include each individual PMU as a combination form. This ensures that except for the extreme case where all PMUs in the network fail, there are available PMU combination sets that can collect information. Therefore, the constraints are expressed as:

[0105]

[0106] Where f(·) represents the calculation function of the PMU observation range, f(x K-N+1 ) to f(x K ) is the observation range of each of the N PMUs in the power grid obtained by the observation algorithm, V k represents the observable node obtained by the kth PMU combination according to the PMU observation algorithm, q represents the number of nodes in the whole network, when the node is observable, V is 1, when the node is not observable, V is 0, where the calculation function f(·) of the PMU observation range is described as:

[0107] ① Determine the node location where the PMU is configured. The node voltage, connected branch current, and connected node voltage of the node where the PMU is configured should be observable.

[0108] ② If the voltage across the branch is considerable, the current in the branch is considerable;

[0109] ③If the branch current and the voltage at one end are considerable, then the voltage at the other end of the branch is considerable;

[0110] ④ If the zero injection node is not equipped with a PMU, and there is only one branch current connected to the zero injection node that is unknown, then the branch current is observable;

[0111] ⑤ If the zero injection node is not configured with a PMU and the voltages of all adjacent nodes of the zero injection node are known, the voltage of the node is observable.

[0112] Find the optimal solution of the objective function of minimum risk optimization. The risk value changes in the solution process are as follows: Figure 3 As shown, when the risk value reaches the minimum of 0.875, a set of K=15 PMU combinations is obtained, and these combinations are the optimal PMU combinations.

[0113] Step 3: Training of transient stability assessment integrated model. This embodiment uses a convolutional neural network for time series feature classification to construct a transient stability assessment sub-classifier, which can identify the observed characteristic variable information and the time information of the input data that is very important for prediction, and weightedly fuse the information. Each PMU combination in step 2 corresponds to a sub-classifier, and the corresponding sub-classifier is trained using the time series features in the corresponding training sample set observed by the PMU combination. Then, the validation sample set is used to obtain the weight of each sub-classifier, and then a weighted integrated transient stability assessment integrated model is constructed.

[0114] Step 3.1: Train the transient stability assessment sub-classifier.

[0115] In this embodiment, a convolutional neural network for time series feature classification is used as a sub-classifier for transient stability assessment. The network can identify the observed feature variables and the timestamps of the input data that are important for prediction, and directly extract the information related to the observed feature variables and the timestamps from the input data in parallel. At the same time, an information weighted fusion mechanism is introduced to enhance useful information mapping and suppress less useful information mapping by automatically learning the importance of each information, thereby realizing automatic weighted fusion of information related to the observed feature variables and the timestamps, thereby improving the performance of the sub-classifier and performing accurate classification tasks. The structure of the sub-classifier is as follows: Figure 4 As shown, T is set to represent the time series length of the feature, D represents the number of features, and F=64 represents the number of filters;

[0116] According to the optimal K = 15 PMU combinations obtained in step 2, each PMU combination corresponds to a sub-classifier, and the training sample set {x 1 ,y 1} features as the input features of the sub-classifier;

[0117] Execute loop i t =1 until i t =200, iteratively train each sub-classifier;

[0118] The one-dimensional convolutional neural network in the blue part and the two-dimensional convolutional neural network in the green part of the subclassifier are represented by the following functions:

[0119]

[0120] Where X represents the input feature, x 1 is a one-dimensional convolutional neural network C 1 (·) Get timestamp-related information, x D It is a two-dimensional convolutional neural network C 2 (·) The information obtained related to the observed characteristic variables; θ 1and θ 2 are the parameters of one-dimensional and two-dimensional convolutional neural networks respectively;

[0121] The information weighted fusion mechanism is introduced and the weighted module is connected. First, the global time information is compressed into the input channel through the global average pooling layer, and then the information passes through the first fully connected layer with the ReLU activation function and the second fully connected layer with the Sigmoid activation function to obtain the information weight s D+1 , the process is expressed as:

[0122] s D+1 =σ(W 2 δ(W 1 x D+1 ))

[0123] Where δ and σ are ReLU and sigmoid activation functions respectively, W 1 , and W 2 are the parameters of the two fully connected layers, x D+1 =[x D+1 |x 1 ] represents the concatenation of two information vectors;

[0124] Output of the attention module is obtained by adding the weighting coefficient s D+1 Applied to the feature map, it is calculated as:

[0125]

[0126] Then the global average pooling layer is used to improve the generalization ability of the network. The final output of the last layer of the entire network is obtained through the softmax operator, as shown below:

[0127]

[0128] Where h j is the output of the jth neuron in the output layer, z is the input of the last layer of the network, and the vector h(z) has all components between 0 and 1, and the sum of its components is 1;

[0129] The optimizer learning rate of each neural network of the sub-classifier is set to η = 0.005, and the process of updating each network parameter is:

[0130]

[0131] Where L 1 , L 2 , L W1 and L W2They are the loss functions of the one-dimensional convolutional neural network, the two-dimensional convolutional neural network, the first fully connected layer in the weighted module, and the first fully connected layer, and all use the cross entropy loss function;

[0132] Determine whether the iteration cycle of sub-classifier training is completed, if so, end the training iteration process, and obtain K=15 trained transient stability assessment sub-classifiers;

[0133] Step 3.2: Construct an integrated model for transient stability assessment.

[0134] In this embodiment, the verification sample set {x 2 ,y 2} Obtain the weight of each sub-classifier in sub-step 3.1, and search for the corresponding weight of each sub-classifier when the prediction result of the verification sample set is closest to the true result. The specific process is expressed as the optimization solution process:

[0135]

[0136] Where M represents the number of samples in the validation set, y i represents the true category of the i-th sample in the validation sample set, Represents the predicted category of the i-th sample in the validation sample set, ω k represents the weight of the k-th subclassifier, P(x ik ) represents the probabilistic output result of the i-th sample in the validation sample set in the k-th sub-classifier;

[0137] After the optimization solution is completed, the corresponding weight ω of each sub-classifier is obtained, and then the integrated model for transient stability assessment is constructed. The weights of each sub-classifier are used for weighted integration. In the case of no missing data, the final transient stability prediction result given by the integrated model at time t can be expressed as:

[0138]

[0139] In the formula represents the transient stability assessment output of the kth sub-classifier at the tth time;

[0140] Step 4: Online adaptive transient stability assessment. The PMU obtains the measurement data of the power system online. The transient stability assessment integrated model obtained in step 3 performs online adaptive fusion of the output results of the unaffected sub-classifiers and the historical output results of the affected sub-classifiers according to the data missing caused by the PMU failure, thereby obtaining the final transient stability assessment result. The online assessment process is as follows: Figure 5 shown.

[0141] Step 4.1, calculate the output result of the integrated model at the current moment.

[0142] In this embodiment, the phasor measurement unit PMU in the wide area measurement system is used to obtain real-time measurement data during the operation of the power system, and input it into the transient stability assessment integrated model obtained in step 3 for online assessment. When data loss occurs, it is assumed that the power system fault removal time is t c , which is the time when the model starts transient stability assessment, and the time when the PMU fails is t m , the longest transient stability assessment time is t max = 20 sampling cycles, the sub-classifier corresponding to the PMU combination where the faulty PMU causing data loss is located is classified as an unavailable sub-classifier, and the rest are classified as available sub-classifiers. The evaluation result obtained by the transient stability evaluation integrated model at the current moment is a weighted fusion of the evaluation result of the unavailable sub-classifier at the moment before the PMU fails and the evaluation result of the available sub-classifier at the current moment. The specific expression of this process is:

[0143]

[0144] Where S represents the remaining available transient stability assessment sub-classifier after the PMU fails, represents the evaluation result of the i-th sub-classifier at the t-th time, Indicates that the jth subclassifier is m-1 The evaluation result at time t is that if the PMU fails at time t m ≤t c ,So The intermediate value of 0.5 between 0 (indicating transient stability) and 1 (indicating transient instability) is taken, which means that the sub-classifier cannot distinguish whether the input sample belongs to transient stability or transient instability due to the failure of PMU before the transient stability assessment begins;

[0145] Step 4.2: Determine the final result of online evaluation.

[0146] In this embodiment, the observation window length of the transient stability assessment integrated model is set to T = 7 sampling periods, and prediction is performed by inputting time series features with a time length of T = 7. In order to obtain accurate prediction results, the transient instability judgment threshold τ = 0.79 is defined. If the prediction result y of the model at the current moment is t ∈[τ,1], then the final evaluation result output at the current time t is transient instability. If y t ∈[0,1-τ], then the final evaluation result output at the current time t is transient stability. These two situations are called satisfying the threshold conditions. However, if y t∈(1-τ,τ), that is, the threshold condition is not met, then it is determined that the current time t cannot give an accurate prediction result temporarily, and it is necessary to enter the next time t+1 to repeat the evaluation of step 4 until the threshold condition is met or the maximum evaluation time t is reached. max = 20 sampling cycles, the specific expression of transient stability Y predicted by online evaluation is:

[0147]

[0148] The final transient stability Y is the online transient stability assessment result under the condition of data loss caused by any number of PMU failures at any time.

[0149] In this embodiment, the accuracy index λ is selected a and stability index λ G Reflects the performance of model evaluation:

[0150]

[0151]

[0152] Where N TS 、N FS 、N FU 、N TU They represent the number of correct stable predictions, the number of incorrect stable predictions, the number of incorrect instability predictions, and the number of correct instability predictions, respectively.

[0153] The average response time is used to reflect the speed of transient stability assessment. It is the average time used for transient stability assessment on the test set. The unit is sampling period. The calculation formula is:

[0154]

[0155] Where L represents the maximum number of evaluation cycles allowed, T i represents the i-th evaluation cycle, C(T i ) represents the number of samples that have completed classification in the i-th evaluation cycle.

[0156] In order to illustrate the superiority of the proposed method, the overall performance of the proposed method and some existing methods for processing missing data (ML, DTSS and Ensemble RVFL) in all cases of missing data are compared, and the T of the above indicators is defined as ar ,λ a and G The overall expectation is expressed as:

[0157]

[0158] Where N represents the total number of PMUs, P represents the failure probability of a single PMU, It means that after the i-th PMU fails at time t, T ar The average value, T 0 represents the total evaluation time of the statistic, and They represent the accuracy λ after the i-th PMU fails at the tth moment. a The average value and stability index λ G The average value of .

[0159] Table 1 Comparison of experimental results

[0160]

[0161] The overall expected statistical results of the above methods are shown in Table 1. In Table 1, since ML, DTSS and EnsembleRVFL all use fixed time observation windows for TSA, their response time is equivalent to the observation window length T=7, which is relatively long. The methods of this invention all use sliding time windows, which can adaptively adjust the response time according to the threshold and effectively use time information, thereby shortening the overall response time. Moreover, the overall accuracy of the method of the invention reaches 99.54%, and the overall λ G It reached 99.76%, which is higher than other methods. Since ML and DTSS use relatively simple shallow learning models, Ensemble RVFL does not consider the importance of different features to construct PMU combinations. Therefore, when PMU data is missing, the overall effect of transient stability assessment of these methods is poor, and the recognition rate of the two types of samples cannot be taken into account. Compared with other methods, the sub-classifier used in the method of the present invention can mine information from both the observed feature variables and the timestamps of the input data, and introduces the information weighted fusion mechanism to distribute the weight of information. It has stronger learning ability and advantages in multivariate time series classification, so its overall performance of transient stability assessment can be better than other methods.

[0162] <Example 2>

[0163] The second embodiment of the present invention provides an adaptive transient stability assessment system for a power system taking into account missing PMU data and capable of automatically implementing the above-mentioned method of the present invention, including a power system data preprocessing module, a power system optimal PMU combination construction module, a power system transient stability assessment integrated model training module, and a power system online adaptive transient stability assessment module.

[0164] The power system data preprocessing module is used to execute the content described in step 1 above and preprocess the acquired power system data.

[0165] The power system optimal PMU combination construction module is used to execute the content described in step 2 above to obtain a set of PMU combinations with the lowest fault risk. This module includes a time series feature importance calculation unit and a minimum risk PMU combination optimization unit. The feature importance calculation unit is used to execute the content described in step 2.1 above to calculate the importance of time series features. The minimum risk PMU combination optimization unit is used to execute the content described in step 2.2 above to obtain the optimal solution of the objective function of the minimum risk optimization to obtain the optimal PMU combination.

[0166] The power system transient stability assessment integrated model training module is used to execute the content described in step 3 above, train sub-classifiers to build an integrated transient stability assessment model. This module includes a transient stability assessment sub-classifier training unit and a transient stability assessment integrated model construction unit. The transient stability assessment sub-classifier training unit is used to execute the content described in step 3.1 above, and obtain the trained transient stability assessment sub-classifier. The transient stability assessment integrated model construction unit is used to execute the content described in step 3.2 above, and weightedly integrate all transient stability assessment sub-classifiers.

[0167] The power system online adaptive transient stability assessment module is used to execute the content described in step 4 above, and online assess the transient stability state of the power system when PMU data is missing. This module includes a current moment integrated model output result calculation unit and an online assessment final result determination unit. The current moment integrated model output result calculation unit is used to execute the content described in step 4.1 above, weighted fusion historical assessment results and the assessment results of the available sub-classifiers at the current moment. The online assessment final result determination unit is used to execute the content described in step 4.2 above, and determine the online transient stability assessment result in the case of missing data.

[0168] The input display module is used to display the input, output and intermediate processing data of the corresponding part in the form of text, list, or two-dimensional / three-dimensional static / dynamic graphics according to the operation instructions input by the operator.

[0169] The control module is communicatively connected with the power system data preprocessing module, the power system optimal PMU combination construction module, the power system transient stability assessment integrated model training module, the power system online adaptive transient stability assessment module, and the input display module to control their operation.

[0170] <Example 3>

[0171] Embodiment 3 provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for adaptively evaluating transient stability of a power system considering missing PMU data are implemented.

[0172] The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the above-mentioned method embodiment, which will not be repeated here. The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the system is divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit, and the above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application.

[0173] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0174] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0175] In the embodiments provided by the present invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0176] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0177] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0178] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned embodiments of the power system transient stability adaptive evaluation method considering the loss of PMU data can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium.

[0179] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for evaluating transient stability of a power system considering data loss, characterized in that: The method comprises: Step 1. Data preprocessing: Obtain all time series feature vectors collected within a preset period after the power system fault is removed under different situations, match them with the corresponding transient stable state, construct the original sample set, and then divide it into a training sample set and a verification sample set; Step 2. Construct the set of optimal PMU combinations: Use the training sample set in step 1, calculate the importance of the time series features of the training sample set, construct the objective function of minimum risk optimization based on the importance of the time series features, and set the constraints that all network nodes are observable and there are still available PMU combinations under severe PMU failures to solve and obtain the set of optimal PMU combinations; Step 3. Training of transient stability assessment integrated model: construct transient stability assessment sub-classifiers, obtain characteristic variable information and time information based on the transient stability assessment sub-classifiers and perform weighted fusion on them, use each optimal PMU combination in step 2 to correspond to a sub-classifier, use the time series features in the corresponding training sample set observed by the optimal PMU combination to train the corresponding sub-classifier, and then use the verification sample set to obtain the weight of each sub-classifier, and then construct a weighted integrated transient stability assessment integrated model; Step 4. Online adaptive transient stability assessment: The PMU obtains the measurement data of the power system online, and obtains the final transient stability assessment result based on the transient stability assessment integrated model obtained in step 3.

2. The method for evaluating transient stability of a power system considering data loss according to claim 1, characterized in that: The step 1 includes the following sub-steps: Step 1.1: Based on the historical operation data of the power system and the time domain simulation records after the fault is cleared, max The active power and reactive power of the generator, the voltage and phase angle of the bus, the active power and reactive power of the line transmission, and the active power and reactive power of the load in the sampling period are used to construct the feature vector x containing the time series; In step 1.2, the transient stability state y of the power system is set to 0 or 1, which respectively represents the transient stability and transient instability of the system. The transient stability state y under each operating state corresponds to the characteristic vector x, and the original sample set {x, y} is constructed. The original sample set {x, y} is randomly divided into a training sample set {x1, y1} and a verification sample set {x2, y2} according to a preset ratio.

3. The method for evaluating transient stability of a power system considering data loss according to claim 1, characterized in that: The step 2 includes the following sub-steps: Step 2.1, time series feature importance calculation: Use the training sample set {x1, y1} to perform an iterative process of calculating the importance of time series features, executing the loop m = 1 until m = m max , where m is the number of iterative sampling, m max is the maximum number of iterative sampling; Determine whether the iterative cycle of the time series feature importance calculation is completed. If so, end the iterative process and obtain the weight of each time series feature calculated at the end as the importance measure of the feature, using h α Indicates the importance of the αth time series feature; Step 2.2, minimum risk PMU combination optimization: Based on the importance of PMU, the risk of unavailability of PMU combination is measured, and the objective function of minimum risk optimization is constructed to find the optimal PMU combination, where the objective function of minimum risk optimization is expressed as: Where K represents the number of PMU combinations, N represents the number of PMUs installed in the power grid, and I is an indicator function. represents the jth combination of i PMUs, x k represents the combination mode of the kth PMU combination, H k represents the importance of the kth PMU combination, P is the probability of PMU failure, and P i It represents the probability of i PMUs failing at the same time; represents the number of all combinations of any i out of N PMUs; The indicator function I is specifically expressed as: The calculation formula for the importance of the k-th PMU combination is: Where Ω k represents all observable features of the kth PMU combination, h α represents the importance of the αth time series feature, which is calculated by sub-step 2.1; The constraints are expressed as: Where f(·) represents the PMU observation range calculation function, f(x K-N+1 ) to f(x K ) is the observation range of each PMU in the N PMUs of the power grid obtained by the observation algorithm, f(x1) to f(x K-N ) is the observation range of other combinations containing 2 or more PMUs, V k represents the observable node obtained by the kth PMU combination according to the PMU observation algorithm, q represents the number of nodes in the whole network, when the node is observable, V is 1, when the node is not observable, V is 0; The optimal solution of the objective function of minimum risk optimization is obtained. When the risk value is minimized, a set of K PMU combinations is obtained. These combinations are the optimal PMU combinations.

4. The method for evaluating transient stability of a power system considering data loss according to claim 3 is characterized in that , the step 3 includes the following sub-steps: Step 3.1, training a transient stability assessment sub-classifier: using a convolutional neural network for time series feature classification as a sub-classifier for transient stability assessment, wherein the one-dimensional convolutional neural network and the two-dimensional convolutional neural network in the sub-classifier are represented by the following functions: Where X represents the input feature, x1 is the timestamp-related information obtained by the one-dimensional convolutional neural network C1(·), and x D is a two-dimensional convolutional neural network, C2(·) obtains information related to the observed feature variables; θ1 and θ2 are the parameters of the one-dimensional and two-dimensional convolutional neural networks, respectively; Introducing the information weighted fusion mechanism, accessing the weighted module, firstly merging the timestamp-related information x1 and the observed feature variable-related information x D The information after splicing is represented as x D+1 =[x D |x1], then x D+1 The global time information is compressed into the input channel through the global average pooling layer, and then the information passes through the first fully connected layer with ReLU activation function and the second fully connected layer with Sigmoid activation function to obtain the information weight s D+1 ; Using the information weights obtained D+1 Calculate the output of the attention module It is expressed as: at last The final output result of the entire network is obtained through the global average pooling layer and the softmax operator; The optimizer learning rate of each neural network of the sub-classifier is set to η, and the process of updating each network parameter is: Where L1, L2, L W1 and L W2 They are the loss functions of the one-dimensional convolutional neural network, the two-dimensional convolutional neural network, the first fully connected layer in the weighted module, and the first fully connected layer, and all use the cross entropy loss function; According to the optimal K PMU combinations obtained in step 2, each PMU combination corresponds to a sub-classifier, and the features of the training sample set {x1, y1} observed by the PMU combination are used as the input features of the sub-classifier; Execute loop i t =1 until i t =i t_max , iteratively train each sub-classifier separately; Determine whether the iterative cycle of sub-classifier training is completed, if so, end the training iteration process to obtain K trained transient stability assessment sub-classifiers; Step 3.2, construct the transient stability assessment integrated model: use the verification sample set {x2, y2} to obtain the weight of each sub-classifier in sub-step 3.1, and search for the corresponding weight of each sub-classifier when the prediction result of the verification sample set is closest to the actual result. The specific process is expressed as follows using the optimization solution process: Where M represents the number of samples in the validation set, y i represents the true category of the i-th sample in the validation sample set, Represents the predicted category of the i-th sample in the validation sample set, ω k represents the weight of the k-th subclassifier, P(x ik ) represents the probabilistic output result of the i-th sample in the validation sample set in the k-th sub-classifier; After the optimization solution is completed, the corresponding weight ω of each sub-classifier is obtained, and then the integrated model for transient stability assessment is constructed. The weights of each sub-classifier are used for weighted integration. In the case of no missing data, the final transient stability prediction result y given by the integrated model at time t is t ′ can be expressed as: In the formula represents the transient stability assessment output of the kth sub-classifier at the tth time.

5. The method for evaluating transient stability of a power system considering data loss according to claim 1, characterized in that: The step 4 includes the following sub-steps: Step 4.1, calculation of the output result of the integrated model at the current moment: Use the phasor measurement unit PMU in the wide-area measurement system to obtain the real-time measurement data of the power system during operation, and input it into the transient stability assessment integrated model obtained in step 3 for online assessment. When data missing occurs, it is assumed that the power system fault removal time is t c , which is the time when the model starts transient stability assessment, and the time when the PMU fails is t m , the longest transient stability assessment time is t max , the sub-classifier corresponding to the PMU combination where the faulty PMU causing data loss is located is classified as an unavailable sub-classifier, and the rest are classified as available sub-classifiers. The evaluation result obtained by the transient stability evaluation integrated model at the current moment is the weighted fusion of the evaluation result of the unavailable sub-classifier at the moment before the PMU failure and the evaluation result of the available sub-classifier at the current moment. The specific expression of this process is y t for: Where S represents the remaining available transient stability assessment sub-classifier after the PMU fails, represents the evaluation result of the i-th sub-classifier at the t-th time, Indicates that the jth subclassifier is m-1 The evaluation result at time t is that if the PMU fails at time t m ≤t c ,So The intermediate value of 0 and 1 is 0.5, which means that the sub-classifier cannot distinguish whether the input sample belongs to transient stability or transient instability because of the failure of PMU before the transient stability assessment begins; Step 4.2, online evaluation final result determination: Set the observation window length of the transient stability evaluation integrated model to T, make predictions by inputting time series features with a time length of T, and define the transient instability judgment threshold τ∈(0.5,1). If the prediction result y of the model at the current moment is t ∈[τ,1], then the final evaluation result output at the current time t is transient instability. If y t ∈[0,1-τ], then the final evaluation result output at the current time t is transient stability. These two situations are called satisfying the threshold conditions. However, if y t ∈(1-τ,τ), that is, the threshold condition is not met, then it is determined that the current time t cannot give an accurate prediction result temporarily, and it is necessary to enter the next time t+1 to repeat the evaluation of step 4 until the threshold condition is met or the maximum evaluation time t is reached. max , the specific expression of transient stability Y predicted by online evaluation is: The final transient stability Y is the online transient stability assessment result under the condition of data loss caused by any number of PMU failures at any time.

6. A power system transient stability assessment system considering data loss, characterized in that: The system comprises: The power system data preprocessing module is used to obtain all time series feature vectors collected after the power system fault is removed under different situations, match them with the corresponding transient stable state, construct the original sample set, and then divide it into a training sample set and a verification sample set; The module for constructing the optimal PMU combination of the power system is used to obtain the set of PMU combinations with the lowest fault risk: the training sample set features in the data preprocessing module are used to calculate the importance of the time series features of the training sample set, and then based on the feature importance, a set of optimal PMU combinations that can be observed by all network nodes and still have available PMU combinations under severe PMU failures is constructed to minimize the fault risk; The power system transient stability assessment integrated model training module is used to train sub-classifiers to build an integrated transient stability assessment model: a convolutional neural network for time series feature classification is used to construct a transient stability assessment sub-classifier, which can identify the observed characteristic variable information and the time information of the input data that is very important for prediction, and perform weighted fusion on the information. Each PMU combination obtained by the optimal PMU combination construction module corresponds to a sub-classifier, and the corresponding sub-classifier is trained using the time series features in the corresponding training sample set observed by the PMU combination. Then, the validation sample set is used to obtain the weight of each sub-classifier, and then a weighted integrated transient stability assessment integrated model is constructed; The power system online adaptive transient stability assessment module is used to online assess the transient stability state of the power system when PMU data is missing: the PMU obtains the measurement data of the power system online, and the transient stability assessment integrated model obtained by the transient stability assessment integrated model training module performs online adaptive fusion of the output results of the unaffected sub-classifiers and the historical output results of the affected sub-classifiers according to the data missing caused by the PMU fault, thereby obtaining the final transient stability assessment result; The power system transient stability assessment system considering data missing is used to execute the steps in the power system transient stability assessment method considering data missing described in any one of claims 1-5.

7. The power system transient stability assessment system considering data loss according to claim 6, characterized in that: Also includes: An input display module is used to display the input, output and intermediate processing data of the corresponding part in the form of text, list, or two-dimensional / three-dimensional static / dynamic graph according to the operation instructions input by the operator; The control module is used to communicate with the power system data preprocessing module, the power system optimal PMU combination construction module, the power system transient stability assessment integrated model training module, the power system online adaptive transient stability assessment module, and the input display module to control their operation.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for evaluating transient stability of a power system considering data loss as described in any one of claims 1 to 5 are implemented.

Images