A method for emergency voltage control of a utility grid in a fault scenario

Abstract

The application discloses a kind of public power grid emergency voltage control methods under fault scene, comprising the following steps: S1, with the total amount of machine and load shedding minimum as power grid emergency voltage control target, control target function and control constraint condition are established;S2, power grid emergency voltage control is modeled to Markov decision process, constructs power grid emergency voltage control intelligent agent;S3, near-end strategy optimization algorithm is used to train and solve power grid emergency voltage control intelligent agent;S4, power grid emergency voltage control intelligent agent is accessed to power grid, judges whether power grid occurs large disturbance or fault based on the electrical quantity information uploaded based on wide area measurement system;S5, if power grid occurs large disturbance or fault, obtains power grid real-time operating state, generates control strategy by power grid emergency voltage control intelligent agent, executes emergency control action, and power grid recovers stable;The application can realize the fast voltage support and stable recovery under fault scene, ensure power grid safe and stable operation.

Description

Technical Field

[0001] This invention belongs to the field of emergency voltage control technology for power grids, and specifically relates to a method for emergency voltage control of public power grids under fault scenarios. Background Technology

[0002] With the acceleration of global energy transition and the in-depth advancement of the "dual carbon" goal, the proportion of new energy sources, represented by wind and solar power, in the power system continues to increase rapidly. The integration of power generation, grid, load, and storage is a key path to achieve clean and low-carbon energy transformation and enhance the absorption capacity of new energy sources, and its grid-connected operation scale is expanding day by day. However, with a large number of new energy power plants being connected to the public power grid, the original synchronous generator units with large inertia and strong anti-interference capabilities are being replaced by new energy units, resulting in a significant decrease in the grid's voltage regulation capability and insufficient grid anti-interference capability. When the grid experiences disturbances or faults, it is easy to cause grid voltage instability. If control measures are not taken in time, voltage collapse may occur in severe cases, leading to system disconnection, which brings unprecedented challenges to the safe and stable operation of the public power grid.

[0003] Therefore, in order to solve the above problems, it is necessary to develop an emergency voltage control method for the public power grid under fault scenarios. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide an emergency voltage control method for public power grids under fault scenarios, so as to achieve rapid voltage support and stable recovery under fault scenarios and ensure the safe and stable operation of the power grid.

[0005] The objective of this invention is achieved as follows: a method for emergency voltage control of a public power grid under fault scenarios, comprising the following steps:

[0006] S1. Taking the minimum total amount of generator and load shedding as the emergency voltage control objective of the power grid, establish the control objective function and control constraints.

[0007] S2. Model the emergency voltage control of the power grid using Markov decision process and construct an intelligent agent for emergency voltage control of the power grid.

[0008] S3. The near-end strategy optimization algorithm is used to train and solve the power grid emergency voltage control agent.

[0009] S4. Connect the power grid emergency voltage control intelligent agent to the power grid, and determine whether the power grid has experienced a large disturbance or fault based on the electrical quantity information uploaded by the wide-area measurement system.

[0010] S5. If a major disturbance or fault occurs in the power grid, the real-time operating status of the power grid is obtained, and a control strategy is generated through the power grid emergency voltage control agent to execute emergency control actions, and the power grid is restored to stability.

[0011] Furthermore, the control objective function in step S1 is expressed as: In the formula, and These represent the number of switchable generators and the number of load nodes, respectively. and They represent the first The generator and the first The number of machines that can be switched off and the amount of load that can be switched off at each load node. and These are the cut-off quantity and the load shedding quantity, respectively. , for Continuous variables within.

[0012] Furthermore, the control constraints in step S1 include grid stability level constraints, power balance constraints, voltage stability constraints, and power angle stability constraints.

[0013] Furthermore, step S2 in the Markov decision process modeling includes constructing the state space, action space, and reward function respectively.

[0014] Furthermore, the state space in step S2 is composed of the generator's power angle, speed, and operating state, the grid's node voltage amplitude and phase, and the remaining shuntable load of the load nodes, specifically represented as follows: , and They represent The power angle and speed vector of the generator in the system at any given time. express The system generator operating status binary vector at any given time, where 0 indicates the generator is disconnected and 1 indicates the generator is operating normally. and They represent The voltage magnitude and phase vector of the system nodes at each instant. express The remaining detachable load vector of the load node at time step.

[0015] Furthermore, the action space in step S2 is represented as follows: In the formula, and These represent the number of switchable generators and the number of load nodes, respectively. express Time generator The excision action is indicated by 0 (no action) and 1 (operational action). express Time-based load nodes The shear load.

[0016] Furthermore, the reward function in step S2 includes a power angle instability reward. Voltage instability reward Emergency Action Rewards and illegal action rewards Specifically, it is expressed as: .

[0017] Furthermore, the training process of the near-end policy optimization algorithm in step S3 includes the following steps: ① Integrating environmental state information The input is fed into the new policy network, yielding two values. These two values ​​are then used as the mean and variance, respectively, to construct a normal distribution. Actions are then sampled using this normal distribution. ② An action value is input into the environment to obtain a reward. and the next state ,Will ③ As a storage of control experience; Input the data into the new policy network, repeat steps ① and ②, until a certain amount of control experience is stored; ④ The obtained data will be processed. ⑤ Input the control experience stored in step ③ into the evaluation network and calculate the action value function; ⑥ Input all stored state combinations into the new policy and old policy networks to obtain the mean and standard deviation to construct the new and old policy distributions; ⑦ Input all stored action combinations into the new and old policy distributions to obtain the probability of each action and calculate the ratio of the corresponding probabilities; ⑧ Calculate the objective function of the action network and then perform backpropagation to update the new policy network; ⑨ Repeat steps ⑥ and ⑦ until the loop ends, and update the old policy network using the parameters of the new policy network; ⑨ Repeat all the above steps until the algorithm converges.

[0018] Furthermore, in step S4, a method based on the phase trajectory change trend is used to determine whether the power grid has experienced a large disturbance or fault. The slope of the generator rotor motion trajectory change is monitored in real time by a wide-area measurement system. When it is found that the slope is increasing, that is, the electromagnetic energy converted from the increase in the power angle difference can no longer accelerate the reduction of the speed difference, it is determined that the system will lose stability, that is, the power grid has experienced a large disturbance or fault.

[0019] Furthermore, the slope of the change in the generator rotor trajectory in step S4 is expressed as follows: In the formula, For prime mover mechanical power, The inertial time constant, The internal potential of the generator, The voltage is the infinite bus voltage. The equivalent impedance between two potentials, Due to the generator speed difference, This is the generator rotor angle.

[0020] Due to the adoption of the above technical solution, the beneficial effects of the present invention are as follows: by taking the minimum total amount of generator tripping and load shedding as the optimization objective of the control measures, and performing Markov decision process modeling, the near-end strategy optimization algorithm is used to train and solve the power grid emergency voltage control agent, and the agent is connected to the power grid. When a disturbance or fault occurs, an emergency voltage control strategy is quickly generated to achieve rapid voltage support and stable recovery under fault scenarios, thereby ensuring the safe and stable operation of the power grid. Attached Figure Description

[0021] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0023] like Figure 1 As shown, a public power grid emergency voltage control method under fault scenarios includes the following steps:

[0024] S1. Taking the minimum total amount of generator tripping and load shedding as the emergency voltage control objective of the power grid, establish the control objective function and control constraints.

[0025] Preferably, the control objective function in step S1 is expressed as: In the formula, and These represent the number of switchable generators and the number of load nodes, respectively. and They represent the first The generator and the first The number of machines that can be switched off and the amount of load that can be switched off at each load node. and These are the cut-off quantity and the load shedding quantity, respectively. , for Continuous variables within.

[0026] Preferably, the control constraints in step S1 include grid stability level constraints, power balance constraints, voltage stability constraints, and power angle stability constraints.

[0027] Preferably, the power grid stability level constraint is expressed as: In the formula, This indicates a node where stability control measures fail to activate after a fault. Voltage stability level, To set a threshold, and They represent the first The generator and the first The first load node cut-off caused the first The change in the voltage stability level of each node. , Indicates the node after the implementation of stability control measures. Voltage stability level, Indicates the first The load change at each node, Indicates the nodes before and after the activation of stability control measures after a fault. The difference in voltage stability levels.

[0028] Preferably, the power balance constraint is expressed as: In the formula, and These represent the upper and lower limits of unbalanced power when considering the frequency regulation capability of generator sets in the power grid.

[0029] Preferably, the voltage stability constraint is expressed as: , express Time Node The voltage amplitude.

[0030] Preferably, the power angle stability constraint is expressed as: , , These represent the start and end times of emergency control, respectively. express Time of the first The power angle of the generator, This represents the set of generators in the system.

[0031] S2. Model the emergency voltage control of the power grid using Markov decision process and construct an intelligent agent for emergency voltage control of the power grid.

[0032] Preferably, step S2 in the Markov decision process modeling includes constructing the state space, action space, and reward function respectively.

[0033] Preferably, the state space in step S2 is composed of the generator's power angle, speed, and operating state, the grid's node voltage amplitude and phase, and the remaining shuntable load of the load nodes, specifically represented as follows: , and They represent The power angle and speed vector of the generator in the system at any given time. express The system generator operating status binary vector at any given time, where 0 indicates the generator is disconnected and 1 indicates the generator is operating normally. and They represent The voltage magnitude and phase vector of the system nodes at each instant. express The remaining detachable load vector of the load node at time step.

[0034] Preferably, the action space in step S2 is represented as follows: In the formula, and These represent the number of switchable generators and the number of load nodes, respectively. express Time generator The excision action is indicated by 0 (no action) and 1 (operational action). express Time-based load nodes The shear load.

[0035] Preferably, the reward function in step S2 includes a power angle instability reward. Voltage instability reward Emergency Action Rewards and illegal action rewards Specifically, it is expressed as: .

[0036] Preferably, the power angle instability reward Set to a constant, large negative number; the voltage instability reward is represented as: In the formula, and These are used to characterize the degree to which the voltage amplitude at each node of the system deviates from 0.8 pu and 1.1 pu, respectively. This represents the voltage magnitude at node i at time t. , , and These are the penalty coefficients for deviations at different stages, all of which are negative; the emergency action reward is represented as follows: In the formula, and These represent the penalty coefficients for machine cut-off and load cut-off actions, respectively. express Time generator The excision action, express Time generator The running status, express Time-based load nodes The shear load, express Time-based load nodes The remaining severable load vector; the illegal action reward is represented as: , This represents the penalty coefficient for illegal actions.

[0037] S3. The near-end strategy optimization algorithm is used to train and solve the power grid emergency voltage control agent.

[0038] Preferably, the training process of the near-end policy optimization algorithm in step S3 includes the following steps: ① Integrating environmental state information The input is fed into the new policy network, yielding two values. These two values ​​are then used as the mean and variance, respectively, to construct a normal distribution. Actions are then sampled using this normal distribution. ② An action value is input into the environment to obtain a reward. and the next state ,Will ③ As a storage of control experience; Input the data into the new policy network, repeat steps ① and ②, until a certain amount of control experience is stored; ④ The obtained data will be processed. The input is fed into the evaluation network, and the action value function is calculated; specifically, the action value function is expressed as: In the formula: Representing state The cumulative reward for the entire sequence up to the termination state. express Instant rewards earned at any time The discount factor is represented as follows: ⑤ The control experience stored in step ③ is input into the evaluation network, and the corresponding state value function and advantage function are calculated. The advantage function is used as the objective function, and the evaluation network parameters are updated through backpropagation. Specifically, the state value function is expressed as: , express Instant rewards earned at any time Indicates the discount factor. express The value of a moment; the advantage function is expressed as: ⑥ Input all stored state combinations into the new policy and old policy networks to obtain the mean and standard deviation, respectively, and construct the new and old policy distributions; input all stored action combinations into the new and old policy distributions to obtain the probability of each action and calculate the ratio of the corresponding probabilities; ⑦ Calculate the objective function of the action network, and then perform backpropagation to update the new policy network; specifically, the objective function is expressed as: In the formula, Indicates the current strategy The parameters, This indicates the state of the current policy and the old policy. Select the same action The ratio of probabilities The dominant function is indicated by the state. Take action below The subsequent advantage value, This represents the clipping function. It is a hyperparameter, usually chosen as 0.3; ⑧ Repeat steps ⑥ and ⑦ until the loop ends, and update the old policy network with the new policy network parameters; ⑨ Repeat all the above steps until the algorithm converges.

[0039] S4. Connect the power grid emergency voltage control intelligent agent to the power grid, and determine whether a major disturbance or fault has occurred in the power grid based on the electrical quantity information uploaded by the wide-area measurement system.

[0040] Preferably, in step S4, a method based on the phase trajectory change trend is used to determine whether the power grid has experienced a large disturbance or fault. The slope of the generator rotor motion trajectory change is monitored in real time by a wide-area measurement system. When it is found that the slope is increasing, that is, the electromagnetic energy converted from the increase in the power angle difference can no longer accelerate the reduction of the speed difference, it is determined that the system will lose stability, that is, the power grid has experienced a large disturbance or fault.

[0041] Preferably, the slope of the change in the generator rotor trajectory in step S4 is expressed as: In the formula, For prime mover mechanical power, The inertial time constant, The internal potential of the generator, The voltage is the infinite bus voltage. The equivalent impedance between two potentials, Due to the generator speed difference, This is the generator rotor angle.

[0042] S5. If a major disturbance or fault occurs in the power grid, the real-time operating status of the power grid is obtained, and a control strategy is generated through the power grid emergency voltage control agent to execute emergency control actions, and the power grid is restored to stability.

[0043] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.

Claims

1. A method for emergency voltage control of a public power grid under fault scenarios, characterized in that: Includes the following steps: S1. Taking the minimum total amount of generator tripping and load tripping as the emergency voltage control objective of the power grid, establish the control objective function and control constraints. S2. Model the emergency voltage control of the power grid using Markov decision process and construct an intelligent agent for emergency voltage control of the power grid. S3. The near-end strategy optimization algorithm is used to train and solve the power grid emergency voltage control agent. S4. Connect the power grid emergency voltage control intelligent agent to the power grid, and determine whether the power grid has experienced a large disturbance or fault based on the electrical quantity information uploaded by the wide-area measurement system. S5. If a major disturbance or fault occurs in the power grid, the real-time operating status of the power grid is obtained, and a control strategy is generated through the power grid emergency voltage control agent to execute emergency control actions, and the power grid is restored to stability.

2. The emergency voltage control method for a public power grid under fault scenarios according to claim 1, characterized in that: The control objective function in step S1 is expressed as follows: In the formula, and These represent the number of switchable generators and the number of load nodes, respectively. and They represent the first The generator and the first The number of machines that can be switched off and the amount of load that can be switched off at each load node. and These are the cut-off quantity and the load shedding quantity, respectively. , for Continuous variables within.

3. The emergency voltage control method for a public power grid under fault scenarios according to claim 1, characterized in that: The control constraints in step S1 include grid stability level constraints, power balance constraints, voltage stability constraints, and power angle stability constraints.

4. The emergency voltage control method for a public power grid under fault scenarios according to claim 1, characterized in that: Step S2 in the Markov decision process modeling includes constructing the state space, action space, and reward function respectively.

5. The emergency voltage control method for a public power grid under a fault scenario according to claim 4, characterized in that: In step S2, the state space is composed of the generator's power angle, speed, and operating state, the grid's node voltage amplitude and phase, and the remaining shuntable load at the load nodes, specifically represented as follows: , and They represent The power angle and speed vector of the generator in the system at any given time. express The system generator operating status binary vector at any given time, where 0 indicates the generator is disconnected and 1 indicates the generator is operating normally. and They represent The voltage magnitude and phase vector of the system nodes at each instant. express The remaining detachable load vector of the load node at time step.

6. The emergency voltage control method for a public power grid under a fault scenario according to claim 4, characterized in that: The action space in step S2 is represented as follows: In the formula, and These represent the number of switchable generators and the number of load nodes, respectively. express Time generator The excision action is indicated by 0 (no action) and 1 (operational action). express Time-based load nodes The shear load.

7. The emergency voltage control method for a public power grid under fault scenarios according to claim 4, characterized in that: The reward function in step S2 includes a power angle instability reward. Voltage instability reward Emergency Action Rewards and illegal action rewards Specifically, it is expressed as: .

8. The emergency voltage control method for a public power grid under fault scenarios according to claim 1, characterized in that: The training process of the near-end policy optimization algorithm in step S3 includes the following steps: ① Inputting environmental state information The input is fed into the new policy network, yielding two values. These two values ​​are then used as the mean and variance, respectively, to construct a normal distribution. Actions are then sampled using this normal distribution. ② An action value is input into the environment to obtain a reward. and the next state ,Will ③ As a storage of control experience; Input the data into the new policy network, repeat steps ① and ②, until a certain amount of control experience is stored; ④ The obtained data will be processed. ⑤ Input the control experience stored in step ③ into the evaluation network and calculate the action value function; ⑥ Input all stored state combinations into the new policy and old policy networks to obtain the mean and standard deviation to construct the new and old policy distributions; ⑦ Input all stored action combinations into the new and old policy distributions to obtain the probability of each action and calculate the ratio of the corresponding probabilities; ⑧ Calculate the objective function of the action network and then perform backpropagation to update the new policy network; ⑨ Repeat steps ⑥ and ⑦ until the loop ends, and update the old policy network using the parameters of the new policy network; ⑨ Repeat all the above steps until the algorithm converges.

9. The emergency voltage control method for a public power grid under fault scenarios according to claim 1, characterized in that: In step S4, a method based on the phase trajectory change trend is used to determine whether the power grid has experienced a large disturbance or fault. The slope of the generator rotor motion trajectory change is monitored in real time by a wide-area measurement system. When it is found that the slope is increasing, that is, the electromagnetic energy converted from the increased power angle difference can no longer accelerate the reduction of the speed difference, it is determined that the system will lose stability, that is, the power grid has experienced a large disturbance or fault.

10. The emergency voltage control method for a public power grid under a fault scenario according to claim 9, characterized in that: The slope of the change in the generator rotor trajectory in step S4 is expressed as: In the formula, The mechanical power of the prime mover. The inertial time constant, The internal potential of the generator, The voltage is the infinite bus voltage. The equivalent impedance between two potentials, Due to the generator speed difference, This is the generator rotor angle.

Images