Fuzzy-control-based transient voltage load shedding control method and device for power system containing high-proportion new energy, and medium

Abstract

The invention discloses a transient voltage load shedding control method and device for a power system containing high-proportion new energy based on fuzzy control and a medium, and relates to the technical field of power system stability analysis and control. The method comprises the following steps: predicting a voltage trajectory by using a voltage trajectory prediction model; calculating predicted voltage recovery time according to the predicted voltage track; determining a key load shedding object; and the system voltage is finally recovered to be stable based on a fuzzy control-based multi-round load shedding control strategy. A large number of samples are trained offline through artificial intelligence, a voltage track is accurately predicted according to real-time voltage response, and the accuracy of voltage stability evaluation is improved; the key load shedding object is selected according to the characteristics of the induction motor in the stall state, so that the low-voltage load shedding efficiency can be further improved; and meanwhile, according to a fuzzy control theory, an action decision is made on the basis of response information after the system is disturbed, so that theoretical support is provided, the system is effectively guaranteed to quickly recover to be stable, and the load shedding amount of the system is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power system stability analysis and control, and more specifically, to a transient voltage load reduction control method, device and medium for a power system containing a high proportion of new energy sources based on fuzzy control. Background Art

[0002] At present, the power system is gradually showing the dual characteristics of "high proportion of renewable energy" and "high proportion of power electronic equipment". The system modeling is complex under the new power system, the anti-interference ability is reduced, and the voltage response characteristics of the system after the disturbance are more complex. At the same time, the dynamic characteristics of the new energy units and induction motors are coupled with each other after the disturbance, resulting in delayed voltage recovery in the system voltage. If the system runs at a low voltage for a long time, it will cause large-scale disconnection of new energy units from the grid, threatening the stability of the system. Summary of the invention

[0003] In order to solve the above technical problems, the present invention provides a transient voltage load reduction control method, device and medium for a power system containing a high proportion of new energy based on fuzzy control. By using a short-term voltage response curve, a voltage response trajectory prediction method is proposed based on an artificial intelligence method to realize the evaluation of the voltage state; at the same time, based on the predicted voltage trajectory, the operating time of the system under a continuous low-voltage state is calculated, which provides a more accurate action basis for the system voltage stability regulation; and based on the stall characteristics of the induction motor, a system key load reduction load positioning method is proposed; finally, on the basis of the above, a voltage stable multi-round low-voltage load reduction strategy based on fuzzy control is proposed, which can realize the formulation of action decisions based on real-time response information, effectively ensure the rapid recovery of the system stability, and reduce the load reduction of the system.

[0004] In a first aspect, the present invention provides a transient voltage load reduction control method for a power system containing a high proportion of new energy sources based on fuzzy control, the method comprising:

[0005] Predicting voltage trajectory using a voltage trajectory prediction model;

[0006] Calculate the estimated voltage recovery time based on the predicted voltage trajectory;

[0007] Identify key load shedding targets;

[0008] The multi-round load shedding control strategy based on fuzzy control finally restored the system voltage to stability.

[0009] Further, predicting the voltage trajectory using the voltage trajectory prediction model includes:

[0010] Obtain historical system voltage response data and extract the dynamic response information of each bus voltage after the disturbance;

[0011] The voltage response data after the fault removal is intercepted, and the time after the fault removal is used as the criterion for dividing the training set and the test set. The short-time voltage response data after the fault removal is used as the training set, and the subsequent voltage time series response data is used as the test set.

[0012] The constructed training set and test set samples are used for offline training of the BiLSTM model to obtain a voltage trajectory prediction model; wherein the BiLSTM model stores the key features of the input information as a state vector c through a memory layer t , and read and modify the state vector at each moment through the gating mechanism. The calculation formula is as follows:

[0013]

[0014] Where: h t , x t are the output vector and input vector at time t, respectively, i t 、f t , o t They are input gate, forget gate and output gate respectively; the input gate is used to control the input information; the forget gate is used to reduce the state vector c at the previous moment t-1 The output gate determines how much information the state variable transmits as the output at the current moment; W i , W f , W o , W c are the weight matrices of the input gate, forget gate, output gate and state vector respectively; b i 、b f 、b o 、b c is the bias vector; σ and tanh are nonlinear activation functions;

[0015] The BiLSTM model obtains the output result by splicing the prediction results of the two LSTM models and coordinating the forward and backward correlations of the time series data. The calculation formula is as follows:

[0016]

[0017] Where: O t is the voltage amplitude prediction value output by the BiLSTM model at time t, are the output vectors of the forward and reverse memory layers, W1 and W2 are weight matrices, and g(·) is the fully connected layer function;

[0018] The accuracy of the voltage trajectory prediction model is evaluated using mean square error, mean absolute error, and determination coefficient.

[0019] In response to the input short-term real-time voltage response data, the voltage trajectory prediction model outputs a complete voltage response trajectory.

[0020] Furthermore, the calculation formulas for the mean square error, mean absolute error and determination coefficient are as follows:

[0021]

[0022] Where: E MS is the mean square error, E MA is the mean absolute error, R 2 is the coefficient of determination, y i is the true value, is the predicted value, m and n are the sample sizes, is the average value.

[0023] Further, calculating the estimated voltage recovery time according to the predicted voltage trajectory includes:

[0024] Determine the voltage stability safety operation domain according to the first voltage stability guideline and the second voltage stability guideline; wherein the first voltage stability guideline is that the load bus after the fault recovers to the first voltage stability threshold within the first set time, and the second voltage stability guideline is that the voltage amplitude after the low voltage load reduction control action recovers to the second voltage stability threshold within the second set time after the fault;

[0025] Based on the predicted voltage trajectory, the time it takes for the voltage to recover to the voltage stable and safe operating range is predicted;

[0026] Furthermore, the predicted time for the voltage to recover to the voltage stable and safe operation domain is used to calculate the time difference ΔT that the system continues to operate in the low voltage state. The calculation formula is as follows:

[0027] ΔT=T pred -T

[0028] Where: T is the initial time when the fault voltage is lower than the stable and safe operation domain, T pred is the moment when the voltage returns to the stable threshold.

[0029] Furthermore, the key load reduction objects are determined by the following method:

[0030] Extract the power response data of each load node after the disturbance;

[0031] Calculate the ratio of reactive power to active power of the load bus;

[0032] Based on the ratio of reactive power to active power of the load bus, the stalled induction motor of the system is identified, and the key load in load shedding is screened out as the key load shedding object.

[0033] Furthermore, the multi-round load reduction control strategy based on fuzzy control, which enables the system voltage to eventually return to stability, includes:

[0034] Extract the voltage response data and make a difference with the voltage stable and safe operation domain to determine the voltage difference ΔV. The calculation formula is as follows:

[0035] ΔV=V t -V th

[0036] Where: V t is the node voltage amplitude at time t, V th Represents the voltage stability safety threshold in the power system safety guidelines.

[0037] The voltage difference ΔV and the low voltage duration ΔT are used as the input of the fuzzy controller, and the low voltage load reduction amount and the load reduction time interval are used as the fuzzy control output to determine the fuzzy control strategy; the fuzzy control strategy is:

[0038] The domain range of the fuzzy controller input ΔV is [0,0.8], and the domain range of ΔT is [0,ΔT max ], where ΔT max is the maximum value that ΔT may appear, the domain range of the output low-voltage load reduction ΔP of the fuzzy controller is [0,0.5], the domain range of the next action interval time t is [0,0.1], the domain of voltage and time difference, load reduction and removal time contains 5 fuzzy subsets, and the corresponding fuzzy control language variables are VS, S, M, L, VL respectively; when the voltage difference ΔV is greater than the first threshold value and / or the low voltage duration ΔT is greater than the second threshold value, the low-voltage load reduction ΔP is increased in the output of the fuzzy control, and the action time interval t is shortened;

[0039] The voltage response data of each node is obtained, and the voltage difference ΔV is calculated in real time. At the same time, the real-time voltage data is used as input, and the voltage trajectory prediction model is used to predict the complete voltage trajectory, and the duration of low voltage ΔT is calculated;

[0040] The voltage difference ΔV and the low voltage duration ΔT are used as input information. Based on the fuzzy control strategy, the current round of low voltage load reduction and the interval time of the next load reduction action are output; at the next round of load reduction action, it is determined whether the system voltage has returned to stability. If the system voltage has returned to stability, the loop is exited;

[0041] If the system voltage does not return to stability at the next load reduction action, the voltage difference ΔV and the low voltage duration ΔT are continued to be calculated, and the fuzzy control strategy is used again to generate the corresponding load reduction action command until the system voltage returns to stability.

[0042] In a second aspect, the present invention provides a transient voltage load reduction control device for a power system containing a high proportion of new energy sources based on fuzzy control, the device comprising:

[0043] A voltage trajectory prediction module is configured to predict a voltage trajectory using a voltage trajectory prediction model;

[0044] A recovery time calculation module is configured to calculate an estimated voltage recovery time based on a predicted voltage trajectory;

[0045] A load shedding object determination module is configured to determine a key load shedding object;

[0046] The load shedding control module is configured as a multi-round load shedding control strategy based on fuzzy control, so that the system voltage can eventually return to stability.

[0047] In a third aspect, the present invention provides a readable storage medium, wherein the readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method as described above.

[0048] The present invention has at least the following beneficial effects:

[0049] The present invention uses LSTM to predict voltage trajectory, calculate the estimated voltage recovery time, select key load shedding, and implement a multi-round load shedding control strategy based on fuzzy control. This method uses a short-term voltage response curve and proposes a voltage response trajectory prediction method based on an artificial intelligence method to achieve voltage stability evaluation; at the same time, based on the predicted voltage trajectory, the system's operating time under a continuous low-voltage state is calculated, providing a more accurate action basis for system voltage stability regulation; based on the induction motor stall characteristics, a system key load shedding positioning method is proposed; finally, based on the above work, a voltage stability multi-round low-voltage load shedding strategy based on fuzzy control is proposed, which can make action decisions based on real-time response information, effectively ensure the system quickly recovers stability, and reduce the amount of load shedding in the system. Finally, a typical example is built in DIgSILENT / PowerFactory to verify the effectiveness of the proposed control strategy, which has the advantages of scientific rationality and strong applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flow chart of a transient voltage load shedding control method for a power system containing a high proportion of new energy sources based on fuzzy control according to an embodiment of the present invention is shown.

[0051] Figure 2 A LSTM structure diagram according to an embodiment of the present invention is shown.

[0052] Figure 3 A BiLSTM structure diagram according to an embodiment of the present invention is shown.

[0053] Figure 4 A schematic diagram of voltage trajectory prediction based on BiLSTM according to an embodiment of the present invention is shown.

[0054] Figure 5 A topology diagram of a 10-machine 39-node system according to an embodiment of the present invention is shown; in the diagram, G represents a generator, G1 to G10 represent 10 different generators in the 10-machine 39-node system, and numbers 1 to 39 represent 39 different nodes in the 10-machine 39-node system.

[0055] Figure 6 A system voltage image is shown in which a short circuit fault is set for line 14-15 at 2s and the fault is removed at 2.16s according to an embodiment of the present invention. In the figure, bus7, bus8, bus12, and bus13 represent bus7, bus8, bus12, and bus13, respectively.

[0056] Figure 7 A schematic diagram of a voltage image after a load shedding action is performed according to an embodiment of the present invention is shown.

[0057] Figure 8 A schematic diagram of a voltage image after load shedding using a traditional method in a voltage delayed recovery scenario is shown.

[0058] Fig. 9 A structural diagram of a transient voltage load reduction control device for a power system containing a high proportion of new energy sources based on fuzzy control according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0059] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and specific embodiments, but are not intended to limit the present invention. For the various steps described herein, if there is no necessity for a causal relationship between each other, the order in which they are described as examples herein should not be regarded as a limitation, and those skilled in the art should know that they can be adjusted in order, as long as the logic between them is not destroyed, resulting in the inability to implement the entire process.

[0060] The embodiment of the present invention provides a transient voltage load reduction control method for a power system containing a high proportion of new energy sources based on fuzzy control. Figure 1 As shown, the method includes steps S100 to S400, which are described in detail as follows.

[0061] S100, predicting a voltage trajectory using a voltage trajectory prediction model.

[0062] In this embodiment, the voltage trajectory prediction based on LSTM can be performed by selecting the historical voltage data of each node, constructing an offline training sample set, and using the LSTM method to achieve the complete voltage trajectory prediction.

[0063] In one embodiment, the voltage trajectory prediction model can be selected as a BiLSTM model, wherein the BiLSTM model includes multiple LSTM networks, and the structure of the BiLSTM model is as follows: Figure 3 As shown, the structure of the LSTM network is as follows Figure 2 Step S100 specifically includes:

[0064] S101, collecting historical system voltage response data, and extracting the dynamic response information of each bus voltage of the system after the disturbance;

[0065] S102, intercepting the voltage response data after the fault is removed, using the time after the fault is removed as the standard for dividing the training set and the test set, using the short-term time series voltage response data after the fault is removed as the training set, and using the subsequent voltage time series response data as the test set.

[0066] S103, use the constructed training set and test set samples for offline training, set the BiLSTM model to have 200 units in each hidden layer, and the number of iterations to 400. The BiLSTM model stores the key features of the input information as a state vector c through the memory layer t , and read and modify the state vector at each moment through the gating mechanism. The calculation formula is as follows:

[0067]

[0068] Where: h t , x t are the output vector and input vector at time t, respectively, i t 、f t , o t They are input gate, forget gate and output gate. The input gate is used to control the input information; the forget gate is used to reduce the state vector c at the previous moment. t-1 The output gate determines how much information the state variable transmits as the output at the current moment; W i , W f , W o , W c are the weight matrices of the input gate, forget gate, output gate and state vector respectively; b i 、b f 、b o 、b c is the bias vector; σ and tanh are nonlinear activation functions.

[0069] Different from the LSTM network, the BiLSTM model is a bidirectional loop structure consisting of a forward and backward LSTM network. BiLSTM can extract feature information in the forward and backward directions at the same time. By splicing the prediction results of the two LSTM networks and coordinating the forward and backward correlations of the data, the output result of BiLSTM is finally obtained. The calculation formula is as follows:

[0070]

[0071] Where: O t is the voltage amplitude prediction value output by BiLSTM at time t, are the output vectors of the forward and reverse memory layers, W1 and W2 are weight matrices, and g(·) is the fully connected layer function;

[0072] S104, using mean square error (MSE), mean absolute error (MAE), and coefficient of determination to evaluate the accuracy of the artificial intelligence model in predicting voltage trajectory;

[0073]

[0074] Where: E MS is the mean square error, E MA is the mean absolute error, R 2 is the coefficient of determination, y i is the true value, is the predicted value, m and n are the sample sizes, is the average value.

[0075] S105, in the online application stage, the short-term real-time voltage response data is input into the trained BiLSTM model (i.e., the voltage trajectory prediction model), and the complete voltage response trajectory can be predicted using artificial intelligence, as shown in the schematic diagram. Figure 4 shown.

[0076] S200: Calculate an estimated voltage recovery time according to the predicted voltage trajectory.

[0077] In this embodiment, in the calculation part of the estimated voltage recovery time, the duration of the system under low voltage operation is calculated based on the predicted voltage trajectory and combined with the safety and stability guidelines of the power system.

[0078] In one embodiment, step S200 specifically includes the following steps:

[0079] S201, determining a voltage stability safety operation domain according to a first voltage stability guideline and a second voltage stability guideline; wherein, the first voltage stability guideline is that the load bus after a fault recovers to a first voltage stability threshold within a first set time, and the second voltage stability guideline is that the voltage amplitude after the low-voltage load reduction control action recovers to a second voltage stability threshold within a second set time after the fault.

[0080] For example, the existing domestic "Guidelines for Voltage Stability Evaluation of Power Systems" is investigated, which stipulates that the load bus after a fault should be restored to above 0.8pu within 10s. At the same time, considering the lack of detail in the current guidelines, the voltage amplitude after the low-voltage load reduction control action mentioned in the North American Electricity Council should be restored to 0.8pu within 4s after the fault. Combining the two voltage stability guidelines, the voltage stability safe operation domain is depicted, and the determined voltage stability safe operation domain is 0.8pu.

[0081] S202, predicting the time for the voltage to recover to a voltage stable and safe operation domain according to the predicted voltage trajectory.

[0082] In this embodiment, when the determined voltage stable and safe operation domain is 0.8 pu, the time for the voltage to recover to 0.8 pu is predicted based on the predicted complete voltage response trajectory.

[0083] S203, using the predicted time for the voltage to recover to the voltage stable and safe operation region, calculate the time difference ΔT during which the system continuously operates in the low voltage state.

[0084] In one embodiment, the predicted time for the voltage to recover to the voltage threshold is used, minus the time when the fault occurs, to calculate the time ΔT that the system continues to operate in the low voltage state. The calculation formula is as follows:

[0085] ΔT=T pred -T

[0086] Where: T is the initial time when the fault is lower than 0.8pu, T pred is the moment when the voltage returns to the stable threshold.

[0087] S300: Determine key load reduction targets.

[0088] In this embodiment, in the part of selecting the key load shedding object, a method for positioning the key load of the system is proposed according to the stall characteristics of the induction motor.

[0089] In one embodiment, step S300 specifically includes the following steps:

[0090] S301, extracting power response data of each load node after being disturbed.

[0091] S302, considering that the induction motor load absorbs a large amount of reactive power after stalling, while the change in active power consumed is relatively small compared to the reactive power, the ratio of reactive power to active power of the load bus is calculated based on this characteristic.

[0092] S303, according to the real-time calculated ratio of reactive power to active power of the load bus, the system stalled induction motor is identified, thereby screening out the key load in load shedding.

[0093] S400, a multi-round load reduction control strategy based on fuzzy control, finally restores the system voltage to stability.

[0094] In this embodiment, in the multi-round load shedding control strategy based on fuzzy control, the voltage difference and time difference that can reflect the voltage stability state are used as input, and the load shedding amount and load shedding action time interval are used as output to formulate corresponding fuzzy control rules. The actual voltage response is used as input, the system voltage trajectory is predicted using the trained model, the estimated voltage recovery time is calculated, and the transient voltage load shedding control strategy is generated online.

[0095] In one embodiment, step S400 specifically includes the following steps:

[0096] S401, extracting voltage response data, and subtracting it from the voltage stability safety threshold to obtain a voltage difference ΔV.

[0097] In one embodiment, the voltage difference ΔV is calculated by the following formula:

[0098] ΔV=V t -V th

[0099] Where: V t is the node voltage amplitude at time t, V th Represents the voltage stability safety threshold in the power system safety guidelines.

[0100] S402, taking the voltage difference ΔV and the low voltage duration ΔT as fuzzy control inputs, and taking the low voltage load reduction amount and load reduction time interval as fuzzy control outputs to formulate a fuzzy control strategy. The fuzzy control strategy is:

[0101] The domain range of the fuzzy controller input ΔV is [0,0.8], and the domain range of ΔT is [0,ΔT max ], where ΔT maxis the maximum value that ΔT may appear. The domain range of the output low-voltage load reduction ΔP of the fuzzy controller is [0,0.5], and the domain range of the next action interval t is [0,0.1]. The domains of voltage and time difference, load reduction and cut-off time contain 5 fuzzy subsets, and the corresponding fuzzy control language variables are VS, S, M, L, and VL respectively. At the same time, it is stipulated that when the voltage difference ΔV and the low voltage duration ΔT are large, the low-voltage load reduction ΔP is increased in the output of the fuzzy control, and the action time interval t is shortened.

[0102] S403, using a wide area measurement device to collect voltage response data of each node in the system, and calculating the voltage difference ΔV in real time. At the same time, using the real-time voltage data as input, using BiLSTM to predict the complete voltage trajectory, and calculating the duration of low voltage ΔT.

[0103] S404, determining the low voltage load shedding control strategy: taking the voltage difference ΔV and the low voltage duration ΔT as input information, using the constructed fuzzy control logic, outputting the low voltage load shedding amount of this round and the interval time of the next load shedding action.

[0104] S405, executing the formulated low-voltage load reduction control strategy, and at the same time judging whether the system voltage has recovered to be stable at the next load reduction action moment, if the system voltage has recovered to be stable, exiting the loop.

[0105] S406, if the system voltage does not return to stability at the next load shedding action, continue to calculate the voltage difference ΔV and the low voltage duration ΔT, and use fuzzy control again to generate a corresponding load shedding action instruction, and repeat the above steps.

[0106] S407: After multiple rounds of load shedding, the system voltage finally returns to stability.

[0107] After knowing the basic implementation steps and working principle of the method proposed in the present invention, the feasibility and advancement of the present invention will be further illustrated in conjunction with specific implementation cases.

[0108] Implementation Cases:

[0109] A classic IEEE 10-machine 39-node calculation system is built. To build a voltage stability scenario under the access of new energy, the static loads under nodes 4, 7, 8, 12, 15, and 39 in the system are changed to a comprehensive load model in which the induction motor and the static load each account for 50%. At the same time, generators 8 and 10 are changed to direct-drive wind turbines. The specific topology is as follows: Figure 5 shown.

[0110] A short circuit fault is set for line 14-15 at 2s and the fault is cleared at 2.16s. The system voltage image is as follows: Figure 6 As shown. Figure 6It can be inferred that the voltage amplitude drops after the system is disturbed. At this time, the wind turbine enters the low voltage ride-through state and generates additional reactive power. However, at this time, the wind turbine is still smaller than the original reactive support capacity of the traditional synchronous machine. In addition, the torque is unbalanced because the terminal voltage of the induction motor is lower than the rated value, which causes a sudden increase in reactive power absorption. The coupling effect between the source side and the load side ultimately causes the voltage to be unable to recover to the 0.8pu threshold in a short period of time. The system runs in a low voltage state for a long time. At this time, there is a risk of wind turbines being disconnected from the grid, and voltage stability control measures need to be taken.

[0111] The transient voltage multi-round load reduction strategy proposed in this application is used to solve the system problem. First, the real-time voltage response trajectory of the system after the disturbance is extracted and transmitted to the trained BiLSTM model to predict the complete voltage response curve of the system, so as to determine that the system will experience voltage delay recovery. At the same time, according to the predicted complete voltage response curve, the voltage difference ΔT is calculated. The reactive and active power ratios of each load bus are calculated in real time. After real-time calculation by the system, it is found that the induction motors 04, 07, 08, and 15 of the system are all at risk of instability. Therefore, they are given priority in the load reduction plan. Finally, the voltage difference Δ and time difference ΔT are used with a fuzzy controller to obtain the corresponding load reduction amount and load reduction action interval, and the partition coordination control plan is formed in combination with the key load reduction load selection results. The voltage image after the load reduction action is executed is as follows: Figure 7 As shown in the figure, the simulation results show that the proposed control strategy has good adaptability in this scenario and can quickly and accurately ensure the system voltage recovery.

[0112] At the same time, compared with traditional load reduction measures, the results of traditional load reduction methods are as follows Figure 8 The comparison results are shown in Table 1. It can be seen that compared with the traditional method, the load reduction method proposed in this paper and the traditional method can quickly restore the system voltage, but the voltage recovery is faster after regulation by the method of the present invention, and its load reduction amount has obvious advantages over the traditional method.

[0113] Table 1 Comparison results of different load reduction strategies under voltage delay recovery scenario

[0114]

[0115] The embodiment of the present invention also provides a transient voltage load reduction control device for a high-proportion new energy power system based on fuzzy control, such as Fig. 9 As shown, the device 900 includes:

[0116] A voltage trajectory prediction module 901 is configured to predict a voltage trajectory using a voltage trajectory prediction model;

[0117] A recovery time calculation module 902 is configured to calculate an estimated voltage recovery time according to a predicted voltage trajectory;

[0118] A load shedding object determination module 903 is configured to determine a key load shedding object;

[0119] The load shedding control module 904 is configured as a multi-round load shedding control strategy based on fuzzy control, so that the system voltage eventually returns to stability.

[0120] In some embodiments, the voltage trajectory prediction module is further configured to:

[0121] Obtain historical system voltage response data and extract the dynamic response information of each bus voltage after the disturbance;

[0122] The voltage response data after the fault removal is intercepted, and the time after the fault removal is used as the criterion for dividing the training set and the test set. The short-time voltage response data after the fault removal is used as the training set, and the subsequent voltage time series response data is used as the test set.

[0123] The constructed training set and test set samples are used for offline training of the BiLSTM model to obtain a voltage trajectory prediction model; wherein the BiLSTM model stores the key features of the input information as a state vector c through a memory layer t , and read and modify the state vector at each moment through the gating mechanism. The calculation formula is as follows:

[0124]

[0125] Where: h t , x t are the output vector and input vector at time t, respectively, i t 、f t , o t They are input gate, forget gate and output gate respectively; the input gate is used to control the input information; the forget gate is used to reduce the state vector c at the previous moment t-1 The output gate determines how much information the state variable transmits as the output at the current moment; W i , W f , W o , W c are the weight matrices of the input gate, forget gate, output gate and state vector respectively; b i 、b f 、b o 、b c is the bias vector; σ and tanh are nonlinear activation functions;

[0126] The BiLSTM model obtains the output result by splicing the prediction results of the two LSTM models and coordinating the forward and backward correlations of the time series data. The calculation formula is as follows:

[0127]

[0128] Where: O t is the voltage amplitude prediction value output by the BiLSTM model at time t, are the output vectors of the forward and reverse memory layers, W1 and W2 are weight matrices, and g(·) is the fully connected layer function;

[0129] The accuracy of the voltage trajectory prediction model is evaluated using mean square error, mean absolute error, and determination coefficient.

[0130] In response to the input short-term real-time voltage response data, the voltage trajectory prediction model outputs a complete voltage response trajectory.

[0131] In some embodiments, the calculation formulas for the mean square error, mean absolute error and determination coefficient are:

[0132]

[0133] Where: E MS is the mean square error, E MA is the mean absolute error, R 2 is the coefficient of determination, y i is the true value, is the predicted value, m and n are the sample sizes, is the average value.

[0134] In some embodiments, the recovery time calculation module is further configured to:

[0135] Determine the voltage stability safety operation domain according to the first voltage stability guideline and the second voltage stability guideline; wherein the first voltage stability guideline is that the load bus after the fault recovers to the first voltage stability threshold within the first set time, and the second voltage stability guideline is that the voltage amplitude after the low voltage load reduction control action recovers to the second voltage stability threshold within the second set time after the fault;

[0136] Based on the predicted voltage trajectory, the time it takes for the voltage to recover to the voltage stable and safe operating range is predicted;

[0137] The predicted time for the voltage to recover to the voltage stable and safe operation domain is used to calculate the time difference ΔT that the system continuously operates in a low voltage state.

[0138] In some embodiments, the recovery time calculation module is further configured to use the predicted time for the voltage to recover to the voltage stable and safe operation domain to calculate the time difference ΔT during which the system continues to operate in a low voltage state, and the calculation formula is as follows:

[0139] ΔT=T pred -T

[0140] Where: T is the initial time when the fault voltage is lower than the stable and safe operation domain, T pred is the moment when the voltage returns to the stable threshold.

[0141] In some embodiments, the load shedding object determination module is further configured to:

[0142] Extract the power response data of each load node after the disturbance;

[0143] Calculate the ratio of reactive power to active power of the load bus;

[0144] Based on the ratio of reactive power to active power of the load bus, the stalled induction motor of the system is identified, and the key load in load shedding is screened out as the key load shedding object.

[0145] In some embodiments, the load shedding control module is further configured to:

[0146] Extract voltage response data and make a difference with the voltage stable and safe operation domain to determine the voltage difference ΔV;

[0147] The voltage difference ΔV and the low voltage duration ΔT are used as the input of the fuzzy controller, and the low voltage load reduction amount and the load reduction time interval are used as the fuzzy control output to determine the fuzzy control strategy; the fuzzy control strategy is:

[0148] The domain range of the fuzzy controller input ΔV is [0,0.8], and the domain range of ΔT is [0,ΔT max ], where ΔT max is the maximum value that ΔT may appear, the domain range of the output low-voltage load reduction ΔP of the fuzzy controller is [0,0.5], the domain range of the next action interval time t is [0,0.1], the domain of voltage and time difference, load reduction and removal time contains 5 fuzzy subsets, and the corresponding fuzzy control language variables are VS, S, M, L, VL respectively; when the voltage difference ΔV is greater than the first threshold value and / or the low voltage duration ΔT is greater than the second threshold value, the low-voltage load reduction ΔP is increased in the output of the fuzzy control, and the action time interval t is shortened;

[0149] The voltage response data of each node is obtained, and the voltage difference ΔV is calculated in real time. At the same time, the real-time voltage data is used as input, and the voltage trajectory prediction model is used to predict the complete voltage trajectory, and the duration of low voltage ΔT is calculated;

[0150] The voltage difference ΔV and the low voltage duration ΔT are used as input information. Based on the fuzzy control strategy, the current round of low voltage load reduction and the interval time of the next load reduction action are output; at the next round of load reduction action, it is determined whether the system voltage has returned to stability. If the system voltage has returned to stability, the loop is exited;

[0151] If the system voltage does not return to stability at the next load reduction action, the voltage difference ΔV and the low voltage duration ΔT are continued to be calculated, and the fuzzy control strategy is used again to generate the corresponding load reduction action command until the system voltage returns to stability.

[0152] In some embodiments, the load reduction control module is further configured to calculate the voltage difference ΔV by the following formula:

[0153] ΔV=V t -V th

[0154] Where: V t is the node voltage amplitude at time t, V th Represents the voltage stability safety threshold in the power system safety guidelines.

[0155] It should be noted that the various device structures described in this embodiment belong to the same technical concept as the previously described method, and achieve the same technical effect through the same principle, which will not be repeated here.

[0156] An embodiment of the present invention further provides a readable storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the method described in any of the above embodiments.

[0157] In addition, although exemplary embodiments have been described herein, the scope includes any and all embodiments based on the present invention with equivalent elements, modifications, omissions, combinations (e.g., various embodiments intersecting schemes), adaptations or changes. The elements in the claims will be interpreted broadly based on the language adopted in the claims, and are not limited to the examples described in this specification or during the implementation of this application, and the examples will be interpreted as non-exclusive. Therefore, this specification and examples are intended to be considered as examples only, and the true scope and spirit are indicated by the following claims and the full scope of their equivalents.

[0158] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of them) can be used in combination with each other. For example, those of ordinary skill in the art can use other embodiments when reading the above description. In addition, in the above-mentioned specific embodiments, various features can be grouped together to simplify the present invention. This should not be interpreted as a feature of an invention that is not claimed for protection being necessary for any claim. On the contrary, the subject matter of the present invention may be less than all the features of the embodiments of a specific invention. Thus, the following claims are incorporated into the specific embodiments as examples or embodiments, wherein each claim is independently used as a separate embodiment, and it is considered that these embodiments can be combined with each other in various combinations or arrangements. The scope of the present invention should be determined with reference to the full scope of the equivalent forms of the attached claims and these claims.

Claims

1. A transient voltage load reduction control method for a power system containing a high proportion of new energy based on fuzzy control, characterized in that: The method comprises: Predicting voltage trajectory using a voltage trajectory prediction model; Calculate the estimated voltage recovery time based on the predicted voltage trajectory; Identify key load shedding targets; The multi-round load shedding control strategy based on fuzzy control finally restored the system voltage to stability.

2. The method according to claim 1, characterized in that The method of predicting the voltage trajectory by using the voltage trajectory prediction model includes: Obtain historical system voltage response data and extract the dynamic response information of each bus voltage after the disturbance; The voltage response data after the fault removal is intercepted, and the time after the fault removal is used as the criterion for dividing the training set and the test set. The short-time voltage response data after the fault removal is used as the training set, and the subsequent voltage time series response data is used as the test set. The constructed training set and test set samples are used for offline training of the BiLSTM model to obtain a voltage trajectory prediction model; wherein the BiLSTM model stores the key features of the input information as a state vector c through a memory layer t , and read and modify the state vector at each moment through the gating mechanism. The calculation formula is as follows: Where: h t , x t are the output vector and input vector at time t, respectively, i t 、f t , o t They are input gate, forget gate and output gate respectively; the input gate is used to control the input information; the forget gate is used to reduce the state vector c at the previous moment t-1 The output gate determines how much information the state variable transmits as the output at the current moment; W i , W f , W o , W c are the weight matrices of the input gate, forget gate, output gate and state vector respectively; b i 、b f 、b o 、b c is the bias vector; σ and tanh are nonlinear activation functions; The BiLSTM model obtains the output result by splicing the prediction results of the two LSTM models and coordinating the forward and backward correlations of the time series data. The calculation formula is as follows: Where: O t is the voltage amplitude prediction value output by the BiLSTM model at time t, are the output vectors of the forward and reverse memory layers, W1 and W2 are weight matrices, and g(·) is the fully connected layer function; The accuracy of the voltage trajectory prediction model is evaluated using mean square error, mean absolute error, and determination coefficient. In response to the input short-term real-time voltage response data, the voltage trajectory prediction model outputs a complete voltage response trajectory; The calculation formulas for the mean square error, mean absolute error and determination coefficient are: Where: E MS is the mean square error, E MA is the mean absolute error, R 2 is the coefficient of determination, y i is the true value, is the predicted value, m and n are the sample sizes, is the average value.

3. The method according to claim 2, characterized in that The calculating the estimated voltage recovery time according to the predicted voltage trajectory includes: Determine the voltage stability safety operation domain according to the first voltage stability guideline and the second voltage stability guideline; wherein the first voltage stability guideline is that the load bus after the fault recovers to the first voltage stability threshold within the first set time, and the second voltage stability guideline is that the voltage amplitude after the low voltage load reduction control action recovers to the second voltage stability threshold within the second set time after the fault; According to the predicted voltage trajectory, the time for the voltage to recover to the voltage stable and safe operation domain is predicted; using the predicted time for the voltage to recover to the voltage stable and safe operation domain, the time difference ΔT that the system continues to operate in the low voltage state is calculated, and the calculation formula is as follows: ΔT=T pred -T Where: T is the initial time when the fault voltage is lower than the stable and safe operation domain, T pred is the moment when the voltage returns to the stable threshold.

4. The method according to claim 3, characterized in that The key load shedding objects are determined by the following methods: Extract the power response data of each load node after the disturbance; Calculate the ratio of reactive power to active power of the load bus; Based on the ratio of reactive power to active power of the load bus, the stalled induction motor of the system is identified, and the key load in load shedding is screened out as the key load shedding object.

5. The method according to claim 4, characterized in that The multi-round load reduction control strategy based on fuzzy control makes the system voltage finally return to stability, including: Extract the voltage response data and make a difference with the voltage stable and safe operation domain to determine the voltage difference ΔV. The calculation formula is as follows: ΔV=V t -V th Where: V t is the node voltage amplitude at time t, V th Represents the voltage stability safety threshold in the power system safety guidelines; The voltage difference ΔV and the low voltage duration ΔT are used as the input of the fuzzy controller, and the low voltage load reduction amount and the load reduction time interval are used as the fuzzy control output to determine the fuzzy control strategy. The fuzzy control strategy is: The domain range of the fuzzy controller input ΔV is [0,0.8], and the domain range of ΔT is [0,ΔT max ], where ΔT max is the maximum value that ΔT may appear, the domain range of the output low-voltage load reduction ΔP of the fuzzy controller is [0,0.5], the domain range of the next action interval time t is [0,0.1], the domain of voltage and time difference, load reduction and removal time contains 5 fuzzy subsets, and the corresponding fuzzy control language variables are VS, S, M, L, VL respectively; when the voltage difference ΔV is greater than the first threshold value and / or the low voltage duration ΔT is greater than the second threshold value, the low-voltage load reduction ΔP is increased in the output of the fuzzy control, and the action time interval t is shortened; Obtain the voltage response data of each node, calculate the voltage difference ΔV in real time, and use the real-time voltage data as input to predict the complete voltage trajectory using the voltage trajectory prediction model to calculate the system low voltage duration ΔT; The voltage difference ΔV and the low voltage duration ΔT are used as input information. Based on the fuzzy control strategy, the current round of low voltage load reduction and the interval time of the next load reduction action are output; at the next round of load reduction action, it is determined whether the system voltage has returned to stability. If the system voltage has returned to stability, the loop is exited; If the system voltage does not return to stability at the next load reduction action, the voltage difference ΔV and the low voltage duration ΔT are continued to be calculated, and the fuzzy control strategy is used again to generate the corresponding load reduction action command until the system voltage returns to stability.

6. A transient voltage load reduction control device for a power system containing a high proportion of new energy based on fuzzy control, characterized in that: The device comprises: A voltage trajectory prediction module is configured to predict a voltage trajectory using a voltage trajectory prediction model; A recovery time calculation module is configured to calculate an estimated voltage recovery time based on a predicted voltage trajectory; A load shedding object determination module is configured to determine a key load shedding object; The load shedding control module is configured as a multi-round load shedding control strategy based on fuzzy control, so that the system voltage can eventually return to stability. 7 . A non-transitory computer-readable storage medium storing instructions, which, when executed by a processor, executes the method according to claim 1 .

Images