Power-current enhancement control method and terminal for transient voltage stability

Abstract

The present invention provides a power-current enhancement control method and terminal for transient voltage stabilization. The method includes: generating a dynamic voltage threshold sample set using a fault set; the sample set includes stable samples and unstable samples; the stable samples and unstable samples correspond to labels respectively; constructing and training a voltage threshold enhancement model based on a penalty loss function, wherein the enhancement model selects a long short-term memory network; monitoring in real time whether the power current criterion is triggered and whether the system is in a process of continuous voltage reduction; and when it is determined that the power current criterion is triggered and the system is in a process of continuous voltage reduction, time-adaptively evaluating the transient voltage and emergency control. In this way, transient voltage assessment and control can be achieved accurately, quickly, and explainably to ensure the safe and efficient operation of the actual power grid.

Description

Technical Field

[0001] The present invention belongs to the field of large power grid stability analysis and control applications, and more specifically, relates to a power-current enhancement control method and terminal for transient voltage stability. Background Art

[0002] In recent years, due to factors such as rapid load growth, increased access to renewable energy, and fluctuating electricity demand, power grid voltage safety issues have frequently emerged, primarily including voltage fluctuations, voltage imbalances, and voltage imbalances. These issues have led to significant issues with transient voltage stability in the system after a disturbance, necessitating an accurate, rapid, and real-time assessment method to ensure safe grid operation.

[0003] However, established transient voltage stability assessment methods rely on detailed system parameters and are not responsive. Consequently, a large number of model-free methods that do not rely on system parameters have emerged, including the Lyapunov exponent method and machine learning. While both methods have demonstrated high accuracy, they suffer from significant limitations in timeliness and interpretability, hindering their application to real-world power grids.

[0004] In addition, model-free methods also include transient voltage indicators designed based on voltage collapse phenomena. These indicators simply compare the indicator value with a fixed threshold set by human experience to determine system stability. However, they lack a reasonable activation mechanism, and their accuracy is highly dependent on the selection of the threshold.

[0005] Therefore, it is urgent to establish an accurate, fast and explainable power-current enhancement control technology solution for transient voltage stability for transient voltage assessment and control to ensure the safe and efficient operation of actual power grids. Summary of the Invention

[0006] In view of this, the present invention proposes a power-current enhancement control method and terminal for transient voltage stabilization, aiming to solve the technical problems of defects in accuracy, timeliness or interpretability in the prior art.

[0007] In a first aspect, the present application provides a power-current enhancement control method for transient voltage stabilization, comprising:

[0008] Generate a dynamic voltage threshold sample set using the fault set; the sample set includes stable samples and unstable samples; the stable samples and unstable samples correspond to labels respectively;

[0009] Constructing and training a voltage threshold enhancement model based on a penalty loss function, wherein the enhancement model selects a long short-term memory network;

[0010] Real-time monitoring of whether the power and current criteria are triggered and whether the system is in the process of continuous voltage reduction;

[0011] When it is determined that the power and current criterion is triggered and the system is in a process of continuous voltage reduction, transient voltage and emergency control are evaluated in a time-adaptive manner.

[0012] Furthermore, the stable samples and unstable samples respectively correspond to input features;

[0013] The voltage amplitude and phase angle of the monitoring node are selected as input features;

[0014] The voltage amplitude and phase angle of the monitoring node before the fault occurs, at the time of fault occurrence, at the time of fault removal, at the start and end time of the first criterion triggering or at the start and end time of more than one criterion triggering are selected as input features.

[0015] Furthermore, when training the voltage threshold enhancement model based on the penalty loss function,

[0016] If the predicted value of the stable sample Greater than the true value The calculation formula for the loss value corresponding to the stable sample in each batch is as follows:

[0017]

[0018] If the predicted value of the unstable sample Less than the true value The calculation formula for the loss value corresponding to the unstable samples in each batch is as follows:

[0019]

[0020] If the predicted value of the unstable sample is neither less than the true value nor the predicted value of the stable sample is greater than the true value, the loss value corresponding to the stable sample or unstable sample in each batch is calculated as follows:

[0021]

[0022] Among them, α s and α u are the penalty factors for stable samples and unstable samples, respectively, and their value range is (1, +∞); N1, N2 and N3 respectively represent the number of samples that meet the corresponding conditions in each batch, and i is a natural number that is not less than 1 and not greater than N1, N2 or N3.

[0023] Furthermore, the real-time monitoring of whether the power and current criterion is triggered and whether the system is in a process of continuous voltage reduction includes:

[0024] Once a fault is detected in real time and cleared, the active power P, apparent power S, and current I are monitored to see whether the following criteria 1 and 2 are met:

[0025] I k+1 -I k >ε&S k+1 -S k >ε

[0026] I k+1 -I k >ε&P k+1 -P k >ε

[0027] The variable ε ranges from 0.00001 to 0.0002; k is the current moment, and k+1 is the next moment;

[0028] When the voltage continuous decrease process determined by the above criterion 1 or criterion 2 is greater than or equal to 0.05s, the trigger power current criterion is determined.

[0029] Furthermore, the time-adaptive evaluation of transient voltage and emergency control includes:

[0030] After the power and current criterion is triggered for the first time, the measured data is preprocessed to form input features and then input into the trained voltage threshold enhancement model based on the penalty loss function;

[0031] The trained voltage threshold enhancement model based on penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship;

[0032] Compare the operating voltage value after the first judgment is completed with the predicted voltage threshold;

[0033] If the operating voltage value is less than the predicted voltage threshold, it is determined that the system is unstable; otherwise, it is determined that the system is not unstable.

[0034] Furthermore, the time-adaptive evaluation of transient voltage and emergency control includes:

[0035] Each time the power and current criterion is triggered, the measured data is preprocessed to form input features and then input into the trained voltage threshold enhancement model based on the penalty loss function.

[0036] The trained voltage threshold enhancement model based on penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship;

[0037] Compare the operating voltage value after each judgment with the predicted voltage threshold;

[0038] If the operating voltage value is less than the predicted voltage threshold, it is determined that the system is unstable; otherwise, it is determined that the system is not unstable.

[0039] Furthermore, once a fault is detected in real time and removed, the system continuously monitors in real time whether the power and current criterion is triggered and whether the system is in a process of continuous voltage reduction within the set maximum time of 10s.

[0040] At least once, it is determined by the above criterion 1 or criterion 2 that the voltage drop process lasts longer than or equal to 0.05s and the triggering power current criterion is determined.

[0041] Furthermore, it also includes:

[0042] Once the system is identified as unstable, emergency control measures are immediately implemented to reduce the risk of instability;

[0043] The emergency control measures include: load shedding and DC power regulation.

[0044] In a second aspect, the present application provides a terminal, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in the first aspect.

[0045] In a third aspect, the present application provides a computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the method described in the first aspect.

[0046] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0048] Figure 1 Schematic diagram of the flow of the power-current enhancement control method for transient voltage stabilization according to an embodiment of the present application;

[0049] Figure 2 A schematic diagram of a dynamic threshold calculation process according to an embodiment of the present application;

[0050] Figure 3 A grid structure diagram of an actual power grid according to an embodiment of the present application;

[0051] Figure 4is the label distribution of the actual power grid in the embodiment of the present application;

[0052] Figure 5 This is the process of distinguishing unstable samples in the embodiment of the present application;

[0053] Figure 6 Application of the embodiment of this application Figure 1 Schematic diagram of the composition of the terminal of the method shown. DETAILED DESCRIPTION

[0054] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that the embodiments of the present disclosure and the features therein can be combined with each other unless they conflict.

[0055] Transient voltage indicators designed based on voltage collapse phenomena can determine system stability simply by comparing the indicator value with a fixed threshold set by human experience. However, they lack a reasonable activation mechanism, and their accuracy is highly dependent on the selection of the threshold.

[0056] Unlike other metrics, the power-current criterion uses an increase in busbar current and a decrease in power after a fault as its triggering mechanism. Once this mechanism is triggered, it indicates that the system is in an unstable region, and a fixed threshold is used to determine stability. However, the choice of fixed threshold still affects the accuracy of this criterion to a certain extent, leaving it with significant potential for improvement.

[0057] The technical solution provided in this application enhances the power and current criterion by integrating it with deep learning technology, establishing a complementary relationship between data-driven and knowledge-driven approaches. This results in an intelligently enhanced transient voltage criterion that is both accurate, timely, and interpretable. The technical solution provided in this application also discloses an intelligently enhanced transient voltage stability criterion.

[0058] like Figure 1 As shown, the power-current enhancement control method for transient voltage stabilization of an embodiment of the present application includes the following steps:

[0059] S100: Generate a dynamic voltage threshold sample set using a fault set; the sample set includes stable samples and unstable samples; the stable samples and unstable samples correspond to labels respectively;

[0060] S200: constructing and training a voltage threshold enhancement model based on a penalty loss function, wherein the enhancement model selects a long short-term memory network;

[0061] S300: Real-time monitoring of whether the power and current criterion is triggered and whether the system is in a process of continuous voltage reduction;

[0062] S400: When it is determined that the power current criterion is triggered and the system is in a process of continuous voltage reduction, transient voltage and emergency control are evaluated in a time-adaptive manner.

[0063] In this way, transient voltage assessment and control can be achieved accurately, quickly and explainably to ensure the safe and efficient operation of the actual power grid.

[0064] In some embodiments, as Figure 2 As shown, the stable samples and unstable samples correspond to input features respectively;

[0065] The voltage amplitude and phase angle of the monitoring node are selected as input features;

[0066] The voltage amplitude and phase angle of the monitoring node before the fault occurs, at the time of fault occurrence, at the time of fault removal, at the start and end time of the first criterion triggering or at the start and end time of more than one criterion triggering are selected as input features.

[0067] In this way, the voltage amplitude and phase angle of any monitoring node before the fault occurs, at the time of fault occurrence, at the time of fault removal, at the start and end time of the first criterion triggering, or at the start and end time of more criterion triggering form a voltage amplitude array and a phase angle array.

[0068] In this way, the minimum value and / or maximum value of the voltage amplitude array of the stable sample and the unstable sample can be respectively used as the threshold label.

[0069] For the above, the generation of dynamic voltage threshold sample sets using the fault set can refer to Table 2 and Table 3 described below, which will not be repeated here.

[0070] In some embodiments, when training a voltage threshold enhancement model based on a penalty loss function,

[0071] If the predicted value of the stable sample Greater than the true value The calculation formula for the loss value corresponding to the stable sample in each batch is as follows:

[0072]

[0073] If the predicted value of the unstable sample Less than the true value The calculation formula for the loss value corresponding to the unstable samples in each batch is as follows:

[0074]

[0075] If the predicted value of the unstable sample is neither less than the true value nor the predicted value of the stable sample is greater than the true value, the loss value corresponding to the stable sample or unstable sample in each batch is calculated as follows:

[0076]

[0077] Among them, α s and α u are the penalty factors for stable samples and unstable samples, respectively, and their value range is (1, +∞); N1, N2 and N3 respectively represent the number of samples that meet the corresponding conditions in each batch, and i is a natural number that is not less than 1 and not greater than N1, N2 or N3.

[0078] In some embodiments, the real-time monitoring of whether the power current criterion is triggered and whether the system is in a process of continuous voltage reduction includes:

[0079] Once a fault is detected in real time and cleared, the active power P, apparent power S, and current I are monitored to see whether the following criteria 1 and 2 are met:

[0080] I k+1 -I k >ε&S k+1 -S k >ε

[0081] I k+1 -I k >ε&P k+1 -P k >ε

[0082] The variable ε ranges from 0.00001 to 0.0002; k is the current moment, and k+1 is the next moment;

[0083] When the voltage continuous decrease process determined by the above criterion 1 or criterion 2 is greater than or equal to 0.05s, the trigger power current criterion is determined.

[0084] In this way, once a fault is detected in real time and the fault is cleared, within the set maximum time of 10s (10s can have a maximum of 200 0.05s), it may be detected multiple times that the power current criterion is triggered and the system is in a process of continuous voltage reduction.

[0085] In some embodiments, as Figure 2 As shown, the time adaptive evaluation of transient voltage and emergency control includes:

[0086] After the power and current criterion is triggered for the first time, the measured data is preprocessed to form input features and then input into the trained voltage threshold enhancement model based on the penalty loss function;

[0087] The trained voltage threshold enhancement model based on penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship;

[0088] Compare the operating voltage value after the first judgment is completed with the predicted voltage threshold;

[0089] If the operating voltage value is less than the predicted voltage threshold, it is determined that the system is unstable; otherwise, it is determined that the system is not unstable.

[0090] In some embodiments, as Figure 2 As shown, the time adaptive evaluation of transient voltage and emergency control includes:

[0091] Each time (e.g., more times as described above) the power and current criterion is triggered, the measured data is preprocessed to form input features and input into the trained voltage threshold enhancement model based on the penalty loss function;

[0092] The trained voltage threshold enhancement model based on penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship;

[0093] Compare the operating voltage value after each judgment with the predicted voltage threshold;

[0094] If the operating voltage value is less than the predicted voltage threshold, it is determined that the system is unstable; otherwise, it is determined that the system is not unstable.

[0095] In this way, the voltage amplitude and phase angle of the monitoring node before the fault occurs, at the time of fault occurrence, at the time of fault removal, at the start and end times of the first criterion triggering, or at the start and end times of more criterion triggering are used as input features, and the trained voltage threshold enhancement model based on the penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship, realizing time-adaptive evaluation of transient voltage and emergency control.

[0096] In some embodiments, once a fault is detected in real time and removed, the power current criterion is continuously monitored in real time within a set maximum time of 10s to see whether it is triggered and whether the system is in a process of continuous voltage reduction. Accordingly,

[0097] At least once, it is determined by the above criterion 1 or criterion 2 that the voltage drop process lasts longer than or equal to 0.05s and the triggering power current criterion is determined.

[0098] In this way, time-adaptive evaluation of transient voltage and emergency control are achieved.

[0099] In some embodiments, further comprising:

[0100] Once the system is identified as unstable, emergency control measures are immediately implemented to reduce the risk of instability;

[0101] The emergency control measures include: load shedding and DC power regulation.

[0102] In this way, time-adaptive evaluation of transient voltage and emergency control are achieved.

[0103] In some embodiments, further comprising:

[0104] Once the system is deemed unstable, emergency control measures will be implemented after a preset delay (no more than the aforementioned 0.05s and less than the aforementioned 10s) to reduce the risk of instability;

[0105] The emergency control measures include: load shedding and DC power regulation.

[0106] In this way, time-adaptive evaluation of transient voltage and emergency control are achieved.

[0107] In some embodiments, in step S100, the dynamic voltage threshold sample set is composed of input features and classification labels or threshold labels. Each sample includes input features, classification labels, and threshold labels, that is, true values.

[0108] The input feature selection strategy includes the following steps S111 to S112:

[0109] S111: Considering the training cost of the enhanced model and the practicality of the transient voltage assessment method, the feature dimension should be less than or equal to the feature dimension used for the power current criterion.

[0110] Specifically, the voltage amplitude and phase angle of the monitoring node are selected as input features. This step is the type of physical quantity of the input feature.

[0111] S112: Considering the response speed of the power and current criteria (typically no more than 0.05s and less than 10s as previously mentioned), the input features consist of the voltage amplitudes and phase angles at the monitoring nodes before the fault occurs, at the moment of fault occurrence, at the moment of fault removal, and at the start and end times of the first criterion trigger. This step spans the input features in the time domain.

[0112] The above sampling points can respectively characterize the steady-state power flow level, fault impact degree, voltage recovery degree after the fault, and voltage drop degree after entering the voltage unstable area for the first time. This is conducive to the construction of knowledge-driven transient voltage intelligent enhancement criteria.

[0113] The steps of calculating the classification label or threshold label include the following S121 to S125:

[0114] S121: After the fault is cleared, a sliding window is started on the power and current change relationship of the monitoring point. If the current increases and the power decreases at the monitoring point, the next step S122 is started;

[0115] S122: Record the monitoring point information of the triggering criterion, including the line number and the operating voltage value at the end of the criterion; and record the number of times the power and current criterion is triggered;

[0116] S123: When the system voltage recovers to above 0.80 pu within 10 seconds, the transient voltage of the system is considered stable, and the classification label is set to 1, indicating a stable sample. Otherwise, the system voltage is considered unstable, and the classification label is set to 0, indicating an unstable sample.

[0117] S124: At the same time, the dimension of the operating voltage set in the simulation process is changed into a three-dimensional array, where the dimensions are the number of samples, the line number, and the number of times the power and current criterion is triggered.

[0118] S125: The minimum and maximum values ​​of the arrays are taken as threshold labels for stable samples and unstable samples respectively.

[0119] Threshold labels and classification labels are different. The threshold label is the voltage threshold that the intelligent model needs to predict. Therefore, the threshold label is used to evaluate the accuracy of the model's predicted threshold. The classification label represents whether the power system voltage is stable or unstable after a transient event and is used to verify whether the proposed method correctly identifies system stability.

[0120] The principle of this step is as follows Figure 2 shown. Figure 2 In the example, sample_1, sample_2, sample_3, and sample_n are used to indicate n samples, label_1, label_2, label_3, and label_n are used to indicate the labels corresponding to n samples; line_1, line_2, line_3, and line_k are used to indicate k lines; trigger_1, trigger_2, trigger_3, and trigger_m are used to indicate the mth power and current criteria trigger. S / P / I / U indicate apparent power / active power / effective current / voltage amplitude respectively; U e (n,1),U e (n,2),U e (n,3),U e (n,m) respectively indicate the bus voltage amplitude corresponding to the mth power and current criterion being triggered.

[0121] In some embodiments, in step S200, when constructing and training a voltage threshold enhancement model based on a penalty loss function, a long short-term memory network (LSTM) is selected as the enhancement model.

[0122] The LSTM network model contains a forget gate, an input gate, and an output gate. During the training process, the input X t With the state memory unit S t-1 、Intermediate output h t-1 Together, they determine the amount of state memory the cell forgets. A sigmoid function is used to calculate the output value, which ranges from 0 to 1. An output close to 0 indicates forgetting the previous state; a value close to 1 indicates retaining the previous state. The input gate is responsible for updating the cell state. It calculates the current input and the previous hidden state, then calculates the new cell state using a sigmoid function and a tanh function. This allows LSTM to selectively update cell states.

[0123] X in the input gate t After the sigmoid and tanh functions are applied, the vector retained in the state memory unit is determined. The output gate determines the output of the current time step. Based on the current input and hidden state calculations, the final output is generated through the sigmoid and tanh functions. The intermediate output h t After the update S t With output O t Together, we determine that the cell state update formula is as follows:

[0124] f t =σ(W fx x t +W th h t-1 +b f ) (1)

[0125] i t =σ(W ix x t +W ih h t-1 +b i ) (2)

[0126] g t =φ(W gx x t +W gh h t-1 +b g ) (3)

[0127] o t =σ(W ox x t +Woh h t-1 +b o ) (4)

[0128] S t =g t ⊙i t +S t-1 ⊙f t (5)

[0129] h t =φ(S t )⊙o t (6)

[0130] Where, f t ,i t , g t , O t , h t and S t are the forget gate, input gate, input node, output gate, intermediate output and state unit states respectively; W fx , W th , W ix , W ih , W gx , W gh , W ox and W oh are the matrix weights of the corresponding gate multiplied by the input and intermediate output; b f , b i , b g , b o are the bias terms of the corresponding gates respectively; ⊙ represents the bitwise multiplication of the elements in the vector; σ represents the change of the sigmoid function; φ represents the change of the tanh function.

[0131] During the training of the threshold enhancement model, the loss function calculation logic is as follows:

[0132] 1) If the predicted value of the stable sample Greater than the true value The calculation formula for the loss value corresponding to the stable sample in each batch is as follows:

[0133]

[0134] 2) If the predicted value of the unstable sample Less than the true value The calculation formula for the loss value corresponding to the unstable samples in each batch is as follows:

[0135]

[0136] 3) If the above situations or conditions 1) or 2) are not met, the loss value corresponding to the stable or unstable samples in each batch is calculated as follows:

[0137]

[0138] Where, α s and α u are the penalty factors for stable samples and unstable samples, respectively, and their value range is (1, +∞); N1, N2 and N3 respectively represent the number of samples in each batch that meet the corresponding situations or conditions 1) or 2) or 3) above, and i is a natural number not less than 1 and not greater than N1, N2 or N3.

[0139] The dynamic voltage threshold sample set generated in step S100 is divided into a training set and a test set.

[0140] Set hyperparameters such as learning rate, batch size, and dropout. According to the above calculation formula, train a voltage threshold enhancement model based on a penalty loss function on the training set to learn the mapping relationship between each electrical quantity in the input features and the voltage threshold.

[0141] Finally, based on the missing alarm rate (MAR), false alarm rate (FAR), accuracy (ACC), and average response time (ART) of the test model, the hyperparameters are repeatedly adjusted until the evaluation performance reaches the optimal level.

[0142] Specifically, when missing alarm rate (MAR), false alarm rate (FAR), accuracy (ACC) and average response time (ART) are used as evaluation indicators, the confusion matrix is ​​first given, as shown in Table 1.

[0143] Table 1 Confusion matrix

[0144]

[0145] Among them, MAR reflects the probability that an unstable sample is missed as a stable sample, and its formula is:

[0146]

[0147] Among them, FAR reflects the probability that a stable sample is misclassified as an unstable sample, and its formula is:

[0148]

[0149] Among them, ACC reflects the overall evaluation accuracy of the sample, and its formula is:

[0150]

[0151] Among them, ART reflects the recognition time of unstable samples after fault removal, and its formula is:

[0152]

[0153] Where: T i Indicates the time when the i-th unstable sample is correctly identified, N u Indicates the number of unstable samples.

[0154] In this way, the voltage threshold enhancement model based on the penalty loss function is trained using deep learning technology. The trained voltage threshold enhancement model based on the penalty loss function is used to construct a data-driven transient voltage intelligent enhancement criterion in the following step S400.

[0155] In some embodiments, in step S300, real-time monitoring of whether the power current criterion is triggered and whether the system is in a process of continuous voltage reduction includes:

[0156] S310: Based on the observability of the power grid, the monitoring range can be divided into full network measurement or local measurement. In online applications, the voltage amplitude and phase angle, power and current measurement data of the monitoring point are transmitted to the server or Figure 6 Terminal 8 is shown.

[0157] S320: Once a fault occurs and is cleared, the active power P, apparent power S, and current I of the selected monitoring point are monitored to see whether the following power and current criteria 1 and 2 are met:

[0158] I k+1 -I k >ε&S k+1 -S k >ε (10)

[0159] I k+1 -I k >ε&P k+1 -P k >ε (11)

[0160] Wherein, the variable ε is a very small positive number ranging from 0.00001 to 0.0002, such as 0.0001, to avoid misjudgment caused by calculation errors; k is the current moment, and k+1 is the next moment.

[0161] The power current criterion is triggered and the voltage is in a continuous reduction process. When the voltage is continuously reduced as determined by the above criterion 1 or criterion 2 for greater than or equal to 0.05s, it is determined that the power current criterion is triggered and the voltage is in a continuous reduction process.

[0162] Typically, the duration of the voltage drop determined by the power and current criterion 1 (Equation (10)) or criterion 2 (Equation (11)) should be greater than or equal to 0.05 seconds, which helps filter out interference points that may be caused by switching operations or data acquisition errors. Once the voltage drop is triggered, the process proceeds to step S400.

[0163] Step S400: Time-adaptive evaluation of transient voltage and emergency control

[0164] S410: After the power current criterion is triggered for the first time or more times, the measured data is preprocessed to form input features, and the input features are input into the trained voltage threshold enhancement model based on the penalty loss function.

[0165] S420: The trained voltage threshold enhancement model based on the penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship;

[0166] S430: Compare the operating voltage value of the selected monitoring point after each judgment with the predicted voltage threshold. If the operating voltage value is less than the predicted voltage threshold, the system is deemed unstable; otherwise, the system is deemed stable. The system returns to step S410 and continues monitoring for the set maximum time of 10 seconds.

[0167] S440: Once the system is determined to be unstable, emergency control measures are immediately implemented to reduce the risk of instability. Step S400 is then terminated or the process returns to step S410. Emergency control measures may include load shedding, DC power regulation, etc.

[0168] As described above, by comparing the operating voltage value after each judgment with the predicted voltage threshold, time-adaptive transient voltage assessment can be achieved.

[0169] Above, the power-current enhancement control method for transient voltage stability of the embodiment of the present application uses the voltage threshold as the enhancement parameter and proposes a data set or sample set construction process for the dynamic threshold. According to the principle of the criterion, a penalty loss function is designed, and the dynamic prediction of the voltage threshold is realized in combination with LSTM. The enhanced criterion based on the dynamic threshold has both accuracy and evaluation speed. At the same time, the method uses real-time quantity measurement (the aforementioned measurement data) as the input feature, has good timeliness, and meets the response requirements of large-scale power grids. In addition, the method only requires real-time quantity measurement, has good applicability, and has engineering application value.

[0170] The following combination Figure 3 、 Figure 4 and Figure 5 , specifically describes the power-current enhancement control method for transient voltage stability of the embodiment of the present application. A certain actual power grid is used as a test system. The power grid consists of a sending-end system and a receiving-end system. The details are as follows: a total of 10,575 nodes; about 550 synchronous power supplies; 912 RES sites mainly based on wind power and photovoltaic power; the total installed power capacity is 289,580MW, the installed capacity of conventional power units is 178,960MW, and the installed capacity of new energy is 110,620MW; 6 DC loops, of which the sending end and the receiving end are connected by 3 DC loops, 1 DC loop at the sending end is sent to other power grids, and 2 DC loops at the receiving end are fed into by other power grids. The grid structure is as follows Figure 3 The monitoring scope includes global measurement and local measurement. Global measurement refers to the 525kV line at the receiving end, while local measurement refers to the 24 525kV AC interconnection lines in the four zones (zones 1R, 2R, 3R, and 4R) at the receiving end.

[0171] The power grid has 14 typical modes, including 6 peak load modes, 4 low peak load modes, and 4 valley load modes. The details are shown in Table 2. The fault set includes deterministic conventional fault sets and unexpected fault sets. The details are shown in Table 3.

[0172] Table 2 Operation mode of actual power grid

[0173]

[0174] Table 3 Fault Set

[0175]

[0176]

[0177] A total of 22,986 samples were generated by scanning the fault set in all modes, including 19,439 stable samples and 3,547 unstable samples. The label distribution of the statistical sample set is as follows: Figure 4 As shown. Figure 4 It can be seen that compared with the empirical threshold (0.75pu), the dynamic threshold has a wider distribution range, which can help the judgment criteria adapt to various scenarios and avoid misjudgment. In addition, for different monitoring ranges, there are differences in the threshold distribution of the two types of samples.

[0178] In order to demonstrate the effect of the voltage threshold enhancement criterion (shown as the enhancement criterion in Table 4), the original criterion and the data-driven method are used as comparison methods. Figure 4 As shown in Figure 4, the original criterion uses a power and current criterion with a voltage threshold of 0.75 pu, also known as the empirical threshold. The data-driven method directly uses LSTM to perform binary classification of transients. Three methods were tested on the test set, and the results are shown in Table 4. The ART for the data-driven method is calculated based on the end time of the first criterion for the unstable sample.

[0179] Table 4 Comparison of enhancement effects

[0180]

[0181] In Table 4, the evaluation performance of the original criterion is significantly affected by the monitoring scope, while the evaluation performance of the data-driven method is consistently high. However, the class imbalance in real power grids is more serious (the number of stable samples is 5.5 times the number of unstable samples), resulting in the data-driven method's MAR as high as 4.0% and 5.31%. Compared with other methods, the enhanced criterion's ACC is consistently above 98%, with a particularly low MAR of around 1%. Especially for power grids with a scale of 10,000 nodes, the enhanced criterion only needs to monitor information from 24 lines to quickly and accurately determine system stability.

[0182] To further analyze the working principle of the enhanced threshold, taking global measurement as an example, the number of times the corresponding criterion is triggered when all unstable samples are correctly classified is counted, such as Figure 5 As shown in , the horizontal axis represents the number of times the criterion is triggered, and the vertical axis represents the number of correctly classified unstable samples. Figure 5 As can be seen, the enhanced criterion can identify 3,400 unstable samples upon initial triggering, while the original criterion can only identify 1,165 unstable samples. As the number of triggers increases, the voltage collapse of the unstable samples becomes more pronounced, allowing the original criterion to identify more unstable samples. Unlike the original criterion, the enhanced criterion can identify a large number of unstable samples with fewer triggers, thereby improving the ART. Furthermore, by the end of the monitoring period, the number of unstable samples that the original criterion failed to identify was far higher than that of the enhanced criterion. This demonstrates that the enhanced threshold effectively improves the criterion's ability to detect unstable samples.

[0183] like Figure 6 As shown, the embodiment of the present application also provides a terminal to execute the method described above. Figure 6 FIG2 shows a schematic diagram of a terminal 8 provided in some embodiments of the present application. Figure 6 As shown, the terminal 8 includes: a processor 800, a memory 801, a bus 802 and a communication interface 803, and the processor 800, the communication interface 803 and the memory 801 are connected via the bus 802; the memory 801 stores a computer program that can be run on the processor 800, and the processor 800 executes the method described in any of the aforementioned embodiments of the present application when running the computer program.

[0184] The memory 801 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. The communication connection between the device network element and at least one other network element is achieved through at least one communication interface 803 (which may be wired or wireless), and may use the Internet, a wide area network, a local area network, a metropolitan area network, etc.

[0185] The bus 802 may be an ISA bus, a PCI bus, or an EISA bus. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory 801 is used to store programs. The processor 800 executes the programs upon receiving execution instructions. The method disclosed in any of the aforementioned embodiments of the present invention may be applied to the processor 800 or implemented by the processor 800.

[0186] The processor 800 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method may be performed by hardware integrated logic circuits or software instructions within the processor 800. The processor 800 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It may implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention may be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules within the decoding processor. The software modules may be located in storage media well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or the like. The storage medium is located in the memory 801 , and the processor 800 reads the information in the memory 801 and completes the steps of the above method in combination with its hardware.

[0187] The terminal provided by the embodiment of the present invention and the method provided by the embodiment of the present invention are based on the same inventive concept and have the same beneficial effects as the method adopted, operated or implemented by them.

[0188] An embodiment of the present application also provides a computer-readable storage medium corresponding to the method described in the aforementioned embodiment. The computer-readable storage medium is a CD-ROM on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it will execute the method described in any of the aforementioned embodiments.

[0189] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0190] The computer-readable storage medium provided by the above-mentioned embodiment of the present application is based on the same inventive concept as the method of the embodiment of the present invention, and has the same beneficial effects as the method adopted, run or implemented by the application program stored therein.

[0191] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A power-current enhancement control method for transient voltage stabilization, characterized in that: include: Generate a dynamic voltage threshold sample set using the fault set; the sample set includes stable samples and unstable samples; The stable samples and unstable samples correspond to labels respectively; Construct and train a voltage threshold enhancement model based on a penalty loss function, wherein the enhancement model selects a long short-term memory network; when training the voltage threshold enhancement model based on a penalty loss function, If the predicted value of the stable sample Greater than the true value The calculation formula for the loss value corresponding to the stable sample in each batch is as follows: If the predicted value of the unstable sample Less than the true value The calculation formula for the loss value corresponding to the unstable samples in each batch is as follows: If it is not the predicted value of the unstable sample Less than the true value And it is not the predicted value of the stable sample Greater than the true value In the case of , the calculation formula for the loss value corresponding to the stable sample or unstable sample in each batch is as follows: Among them, α s and α u are the penalty factors for stable samples and unstable samples, respectively, and their value range is (1, +∞); N1, N2, and N3 represent the number of samples in each batch that meet the corresponding conditions, and i is a natural number that is not less than 1 and not greater than N1, N2, and N3; Real-time monitoring of whether the power and current criteria are triggered and whether the system is in the process of continuous voltage reduction; When the power and current criteria are triggered and the system is in a process of continuous voltage reduction, transient voltage and emergency control are adaptively evaluated in time, including: After the power and current criterion is triggered for the first time, the measured data is preprocessed to form input features and then input into the trained voltage threshold enhancement model based on the penalty loss function; The trained voltage threshold enhancement model based on penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship; Compare the operating voltage value after the first judgment is completed with the predicted voltage threshold; If the operating voltage value is less than the predicted voltage threshold, it is determined that the system is unstable; otherwise, it is determined that the system is not unstable.

2. The power-current enhancement control method for transient voltage stabilization according to claim 1, characterized in that: The stable samples and unstable samples respectively correspond to input features; The voltage amplitude and phase angle of the monitoring node are selected as input features; The voltage amplitude and phase angle of the monitoring node before the fault occurs, at the time of fault occurrence, at the time of fault removal, at the start and end time of the first criterion triggering or at the start and end time of more than one criterion triggering are selected as input features.

3. The power-current enhancement control method for transient voltage stabilization according to claim 1, characterized in that: The real-time monitoring of whether the power and current criterion is triggered and whether the system is in a process of continuous voltage reduction includes: Once a fault is detected in real time and cleared, the active power P, apparent power S, and current I are monitored to see whether the following criteria 1 and 2 are met: I k+1 -I k >e&S k+1 -S k >e I k+1 -I k >e&P k+1 -P k >e The variable ε ranges from 0.00001 to 0.0002; k is the current moment, and k+1 is the next moment; When the voltage continuous decrease process determined by the above criterion 1 or criterion 2 is greater than or equal to 0.05s, the trigger power current criterion is determined.

4. The power-current enhancement control method for transient voltage stabilization according to claim 3, characterized in that: The time-adaptive evaluation of transient voltage and emergency control includes: Each time the power and current criterion is triggered, the measured data is preprocessed to form input features and then input into the trained voltage threshold enhancement model based on the penalty loss function. The trained voltage threshold enhancement model based on penalty loss function quickly predicts the voltage threshold according to the learned mapping relationship; Compare the operating voltage value after each judgment with the predicted voltage threshold; If the operating voltage value is less than the predicted voltage threshold, it is determined that the system is unstable; otherwise, it is determined that the system is not unstable.

5. The power-current enhancement control method for transient voltage stabilization according to claim 3, characterized in that: Once a fault is detected in real time and removed, the system will continuously monitor in real time whether the power and current criterion is triggered and whether the system is in a process of continuous voltage reduction within the set maximum time of 10s. At least once, it is determined by the above criterion 1 or criterion 2 that the voltage drop process lasts longer than or equal to 0.05s and the triggering power current criterion is determined.

6. The power-current enhancement control method for transient voltage stabilization according to claim 1 or 4, characterized in that: Also includes: Once the system is identified as unstable, emergency control measures are immediately implemented to reduce the risk of instability; The emergency control measures include: load shedding and DC power regulation.

7. A terminal comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method according to any one of claims 1 to 6.

8. A computer storage medium, characterized in that Computer-executable instructions are stored, and the computer-executable instructions are used to execute the method according to any one of claims 1 to 6.

Images