A model-data joint driven real-time operation risk assessment method for power systems

Abstract

The application provides a model-data combined driving real-time operation risk assessment method of a power system, and belongs to the field of power systems. The method firstly samples a large number of power grid real-time operation scenes considering multi-source uncertainty by using a Monte Carlo method; calculates the risk of each real-time operation scene and accumulates the risk index by using an optimal load shedding and optimal power flow method for abandoning wind and light; trains and learns each risk scene sample by using a convolutional neural network, extracts meaningful and relevant features in the original data, inputs the extracted features into a support vector machine model for regression training, thereby greatly improving the real-time operation risk assessment efficiency, and providing an important reference basis for real-time operation risk assessment of a power system with a high proportion of new energy access.

Description

Technical Field

[0001] This invention relates to the field of power systems, and more specifically to a model-data jointly driven method for real-time operation risk assessment of power systems. Background Technology

[0002] As the proportion of renewable energy sources gradually increases, the risks faced by the power system are also exhibiting unprecedented complexity and challenges. Renewable energy generation is characterized by significant randomness, volatility, and intermittency, which profoundly impacts the stability of the power grid. A major power system failure could trigger severe economic losses and social problems. Simultaneously, the continuous expansion of the power system's scale and increasing structural complexity undoubtedly exacerbates the difficulty of grid risk assessment. Traditional risk assessment methods, typically based on the stable output of conventional energy sources and the smooth changes in load demand, are ill-suited to addressing the uncertainties and volatility introduced by renewable energy sources.

[0003] Furthermore, traditional indicators and models used for real-time risk assessment in power systems have significant limitations when dealing with new power systems. Their computational efficiency and speed are also unsatisfactory in supporting large-scale real-time operational risk assessments. In particular, with the large-scale integration of renewable energy sources, how to comprehensively and accurately assess the system's security and stability has become a critical issue that urgently needs to be addressed. Summary of the Invention

[0004] To address these issues, this invention proposes a model-data jointly driven method for real-time operation risk assessment of power systems. The aim is to overcome the limitations of traditional risk assessment methods in effectively addressing the uncertainties and volatility brought about by new energy sources in the context of high-proportion renewable energy integration in current power systems. This method not only overcomes the shortcomings of traditional assessment methods but also better addresses the uncertainties and volatility brought about by new energy sources. By integrating accurate modeling and data analysis methods, it enables real-time monitoring and assessment of the power system's operating status, timely detection of potential risks, and early warning.

[0005] The technical solution adopted in this invention is as follows: A model-data jointly driven method for real-time operation risk assessment of power systems, comprising the following steps:

[0006] S1. Use the Monte Carlo method to sample and form a large number of real-time power grid operation scenarios that take into account multi-source uncertainties;

[0007] S2. Calculate the risk and accumulate risk indicators for each real-time operation scenario by using the optimal load shedding and optimal power flow method for wind and solar curtailment.

[0008] S3. Use convolutional neural networks to train and learn from samples of various risk scenarios, extract meaningful and relevant features from the original data, and input the extracted features into the support vector machine model for regression training to achieve a significant improvement in the efficiency of real-time risk assessment.

[0009] Furthermore, the specific steps in step S1 of using the Monte Carlo method to sample and form a large number of real-time power grid operation scenarios that consider multi-source uncertainties are as follows:

[0010] S11. Sampling the operating status of all system components to obtain data including... State vector of a component system ;

[0011] S12. The output uncertainty of wind power, photovoltaic power and load is simulated by sampling in the form of predicted value plus prediction error.

[0012] Furthermore, in step S11, the state of each component is typically divided into two types: normal and shutdown. Assuming that each component's shutdown is independent, the component only has two states: normal operation and fault shutdown. Indicator element state, This indicates its failure probability. The component shutdown model is as follows, using a uniformly distributed random number interval [0,1]:

[0013] (1),

[0014] In the formula, This indicates that the component is in operation; This indicates that the component is inactive; the status includes... The state of a component system is represented by a vector. :

[0015] (2),

[0016] Furthermore, in step S12, the output uncertainty of wind power, photovoltaic power, and load is simulated by sampling in the form of predicted values ​​plus prediction errors. The calculation method is as follows:

[0017] (3),

[0018] In the formula, This is the actual value; These are predicted values, obtained based on statistical or physical models. This represents the prediction error.

[0019] Furthermore, step S2 utilizes the optimal load shedding and optimal power flow method for wind and solar curtailment to calculate risks and accumulate risk indicators for each real-time operation scenario. The specific steps are as follows:

[0020] S21. Perform power flow calculation and analysis on the real-time operating status obtained by sampling, and determine whether the system has experienced power flow over-limit or node voltage over-limit. If abnormal operating status occurs, perform optimal load shedding and power flow correction for wind and solar power curtailment.

[0021] S22. Establish risk indicators for key risk factors in real-time operation to more accurately perceive and quantify the risk level in the real-time operation of the power grid.

[0022] Furthermore, the power flow calculation and analysis equations in step S21 are as follows:

[0023] (4),

[0024] In the formula: It is the number of system nodes; and The nodes are listed in order. The injected active power and injected reactive power; and The nodes are listed in order. Voltage amplitude and phase angle; , and These are, in order, the real and imaginary parts of the admittance in the nodal admittance matrix;

[0025] The optimal load shedding and wind / solar curtailment power flow model is as follows:

[0026] Objective function:

[0027] (5),

[0028] Constraints:

[0029] (6),

[0030] In the formula, It is a node The load shearing power at the location; and These are nodes The amount of wind and solar power curtailed at the location; , These are the penalty factors for load shedding and power abandonment, respectively. For scheduling intervals; and These are the node voltage magnitude vector and the phase angle vector, respectively. yes Element; and These are nodes The active and reactive power of the load; , , and These are nodes The lower and upper limits of injected active power and injected reactive power; and These are nodes The upward and downward ramp rates of the unit assembly; It is a side road The trend; It is a side road The maximum conveying capacity; and The nodes are listed in order. The lower and upper limits of voltage amplitude; , , , and These are the load nodes, new energy power station nodes, power supply nodes, all branches, and all nodes in the system.

[0031] Furthermore, step S22 includes the following indicators:

[0032] The increased flexibility is insufficient to meet expectations; parameters This refers to the expected value of the difference between the increased reserve and actual demand that conventional units can provide during an operating day. The calculation method is as follows:

[0033] (7),

[0034] In the formula: and They represent the system at time t. The increase in reserve deficit and reserve capacity; and They are time points and Net load; and They are nodes Units are assembled in The maximum output and actual output at any given moment;

[0035] The reduced flexibility is insufficient to meet expectations; parameters This refers to the expected value of the difference between the reduced reserve and actual demand that conventional units can provide during an operating day. The calculation method is as follows:

[0036] (8),

[0037] In the formula: and They represent the system at time t. The reduction of reserve deficit and reserve capacity; For nodes Units are assembled in The minimum output at any given moment;

[0038] Wind curtailment expectation, parameters The calculation method is as follows:

[0039] (9),

[0040] In the formula: This represents the probability of wind curtailment occurring.

[0041] Discarding expectation, parameters The calculation method is as follows:

[0042] (10)

[0043] In the formula: This represents the probability of a state of light abandonment occurring.

[0044] Expected power shortage, parameters The calculation method is as follows:

[0045] (11),

[0046] In the formula: This represents the probability of a load shedding state occurring.

[0047] Furthermore, in step S3, the ReLU activation function, RMSE loss function, and Adam optimization algorithm are selected, and the feature matrix is ​​constructed based on the real-time operating status of the power system as follows:

[0048] (12)

[0049] In the formula, This represents the number of system nodes. For nodes Load power; For nodes New energy power; For nodes Power of conventional generating units.

[0050] Another objective of this invention is to provide a model-data co-driven real-time operation risk assessment system for power systems, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements a model-data co-driven real-time operation risk assessment method for power systems as described above.

[0051] Another object of the present invention is to provide a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, implements a model-data jointly driven real-time operation risk assessment method for power systems as described above.

[0052] Advantages and benefits of this invention: This invention aims to address the randomness, volatility, and intermittency of new energy power generation by combining advanced and accurate modeling and data analysis methods to enhance the risk identification and early warning capabilities of power systems in complex operating environments. Through an innovative risk assessment model, it accurately evaluates the operational safety and stability of the power system. This invention balances the economic efficiency and safety of the power system, possessing significant theoretical importance and broad application prospects, and can provide scientific support and decision-making basis for the safe operation of high-proportion new energy power systems. Attached Figure Description

[0053] Figure 1 This is a flowchart of the present invention;

[0054] Figure 2 The flowchart shows the support vector machine using the radial basis kernel function as the kernel function in this invention. Detailed Implementation

[0055] The invention will be further described below with reference to the accompanying drawings:

[0056] Example 1

[0057] like Figure 1 As shown, a model-data jointly driven method for real-time operation risk assessment of power systems includes the following steps:

[0058] S1. Using the Monte Carlo method, a large number of real-time power grid operation scenarios are sampled to consider multi-source uncertainties. The specific steps are as follows:

[0059] S11. Sampling the operating status of all system components to obtain data including... State vector of a component system In this context, each component's state is typically divided into two types: normal and shutdown. Assuming that each component's shutdown is independent, the component only has two states: normal operation and fault shutdown. Indicator element state, This indicates its failure probability. The component shutdown model is as follows, using a uniformly distributed random number interval [0,1]:

[0060] (1),

[0061] In the formula, This indicates that the component is in operation; This indicates that the component is inactive; the status includes... The state of a component system is represented by a vector. :

[0062] (2),

[0063] S12. The output uncertainty of wind power, photovoltaic power, and load is simulated by sampling in the form of predicted values ​​plus prediction errors. The calculation method is as follows:

[0064] (3),

[0065] In the formula, This is the actual value; These are predicted values, obtained based on statistical or physical models. This represents the prediction error;

[0066] S2. Utilize the optimal load shedding and optimal power flow method for wind and solar curtailment to calculate risks and accumulate risk indicators for each real-time operation scenario. The specific steps are as follows:

[0067] S21. Perform power flow calculation and analysis on the sampled real-time operating status to determine whether the system has experienced power flow overruns or node voltage overruns. If abnormal operating conditions are found, perform optimal load shedding and power flow correction for wind and solar power curtailment. The power flow calculation and analysis equations are as follows:

[0068] (4),

[0069] In the formula: It is the number of system nodes; and The nodes are listed in order. The injected active power and injected reactive power; and The nodes are listed in order. Voltage amplitude and phase angle; , and These are, in order, the real and imaginary parts of the admittance in the nodal admittance matrix;

[0070] The optimal load shedding and wind / solar curtailment power flow model is as follows:

[0071] Objective function:

[0072] (5),

[0073] Constraints:

[0074] (6),

[0075] In the formula, It is a node The load shearing power at the location; and These are nodes The amount of wind and solar power curtailed at the location; , These are the penalty factors for load shedding and power abandonment, respectively. For scheduling intervals; and These are the node voltage magnitude vector and the phase angle vector, respectively. yes Element; and These are nodes The active and reactive power of the load; , , and These are nodes The lower and upper limits of injected active power and injected reactive power; and These are nodes The upward and downward ramp rates of the unit assembly; It is a side road The trend; It is a side road The maximum conveying capacity; and The nodes are listed in order. The lower and upper limits of voltage amplitude; , , , and These are the load nodes, new energy power station nodes, power supply nodes, and the collection of all branches and nodes in the system.

[0076] S22. Establish risk indicators for key risk factors in real-time operation to more accurately perceive and quantify the risk level during real-time power grid operation, including the following indicators:

[0077] The increased flexibility is insufficient to meet expectations; parameters This refers to the expected value of the difference between the increased reserve and actual demand that conventional units can provide during an operating day. The calculation method is as follows:

[0078] (7),

[0079] In the formula: and They represent the system at time t. The increase in reserve deficit and reserve capacity; and They are time points and Net load; and They are nodes Units are assembled in The maximum output and actual output at any given moment;

[0080] The reduced flexibility is insufficient to meet expectations; parameters This refers to the expected value of the difference between the reduced reserve and actual demand that conventional units can provide during an operating day. The calculation method is as follows:

[0081] (8),

[0082] In the formula: and They represent the system at time t. The reduction of reserve deficit and reserve capacity; For nodes Units are assembled in The minimum output at any given moment;

[0083] Wind curtailment expectation, parameters The calculation method is as follows:

[0084] (9),

[0085] In the formula: This represents the probability of wind curtailment occurring.

[0086] Discarding expectation, parameters The calculation method is as follows:

[0087] (10)

[0088] In the formula: This represents the probability of a state of light abandonment occurring.

[0089] Expected power shortage, parameters The calculation method is as follows:

[0090] (11),

[0091] In the formula: This represents the probability of a load shedding state occurring.

[0092] S3. A convolutional neural network is used to train and learn from samples of various risk scenarios, extracting meaningful and relevant features from the original data. These extracted features are then input into a support vector machine model for regression training, significantly improving the efficiency of real-time risk assessment. In this embodiment, the support vector machine selects a radial basis function kernel as the kernel function. The specific process is as follows: Figure 2 As shown, this embodiment uses the ReLU activation function, RMSE loss function, and Adam optimization algorithm to construct the feature matrix based on the real-time operating status of the power system, as follows:

[0093] (12)

[0094] In the formula, This represents the number of system nodes. For nodes Load power; For nodes New energy power; For nodes Power of conventional generating units.

[0095] In summary, this invention balances the economy and safety of power systems, has significant theoretical implications and broad application prospects, and can provide scientific support and decision-making basis for the safe operation of high-proportion renewable energy power systems.

[0096] The above are merely embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A model-data jointly driven method for real-time operation risk assessment of power systems, characterized in that, Includes the following steps: S1. Using the Monte Carlo method to sample and form a large number of real-time power grid operation scenarios that consider multi-source uncertainties, the specific steps of using the Monte Carlo method to sample and form a large number of real-time power grid operation scenarios that consider multi-source uncertainties are as follows: S11. Sample the operating status of all system components to obtain data including... State vector of a component system Each component typically has two states: normal and stopped. Assuming each component's stopped state is independent, the component only has two states: normal operation and fault-induced stopped operation. Indicator element state, This indicates its failure probability. The component shutdown model is as follows, using a uniformly distributed random number interval [0,1]: （1）， In the formula, This indicates that the component is in operation; This indicates that the component is inactive; the status includes... The state of a component system is represented by a vector. : （2）， S12. The output uncertainty of wind power, photovoltaic power, and load is simulated by sampling using the form of predicted values ​​plus prediction errors. The calculation method for simulating the output uncertainty of wind power, photovoltaic power, and load using the form of predicted values ​​plus prediction errors is as follows: （3）， In the formula, This is the actual value; These are predicted values, obtained based on statistical or physical models. This represents the prediction error; S2. Calculate the risk and accumulate risk indicators for each real-time operation scenario using the optimal load shedding and optimal power flow method for wind and solar curtailment. The specific steps for calculating the risk and accumulating risk indicators for each real-time operation scenario using the optimal load shedding and optimal power flow method for wind and solar curtailment are as follows: S21. Perform power flow calculation and analysis on the sampled real-time operating status to determine whether the system has experienced power flow overruns or node voltage overruns. If abnormal operating conditions are found, perform optimal load shedding and power flow correction for wind and solar power curtailment. The power flow calculation and analysis equations are as follows: （4）， In the formula: It is the number of system nodes; and The nodes are listed in order. The injected active power and injected reactive power; and The nodes are listed in order. Voltage amplitude and phase angle; , and These are, in order, the real and imaginary parts of the admittance in the nodal admittance matrix; The optimal load shedding and wind / solar curtailment power flow model is as follows: Objective function: （5）， Constraints: （6）， In the formula, It is a node The load shearing power at the location; and These are nodes The amount of wind and solar power curtailed at the location; , These are the penalty factors for load shedding and power abandonment, respectively. For scheduling intervals; and These are the node voltage magnitude vector and the phase angle vector, respectively. yes Element; and These are nodes The active and reactive power of the load; , , and These are nodes The lower and upper limits of injected active power and injected reactive power; and These are nodes The upward and downward ramp rates of the unit assembly; It is a side road The trend; It is a side road The maximum conveying capacity; and The nodes are listed in order. The lower and upper limits of voltage amplitude; , , , and These are the load nodes, new energy power station nodes, power supply nodes, and the collection of all branches and nodes in the system. S22. Establish risk indicators for key risk factors in real-time operation to more accurately perceive and quantify the risk level during real-time power grid operation. These indicators include the following: The increased flexibility is insufficient to meet expectations; parameters This refers to the expected value of the difference between the increased reserve and actual demand that conventional units can provide during an operating day. The calculation method is as follows: （7）， In the formula: and They represent the system at time t. The increase in reserve deficit and reserve capacity; and They are time points and Net load; and They are nodes Units are assembled in The maximum output and actual output at any given moment; The reduced flexibility is insufficient to meet expectations; parameters This refers to the expected value of the difference between the reduced reserve and actual demand that conventional units can provide during an operating day. The calculation method is as follows: （8）， In the formula: and They represent the system at time t. The reduction of reserve deficit and reserve capacity; For nodes Units are assembled in The minimum output at any given moment; Wind curtailment expectation, parameters The calculation method is as follows: （9）， In the formula: This represents the probability of wind curtailment occurring. Discarding expectation, parameters The calculation method is as follows: （10）， In the formula: This represents the probability of a state of light abandonment occurring. Expected power shortage, parameters The calculation method is as follows: （11）， In the formula: This represents the probability of a load shedding state occurring. S3. A convolutional neural network is used to train and learn from samples of various risk scenarios, extracting meaningful and relevant features from the original data. These extracted features are then input into a support vector machine model for regression training, significantly improving the efficiency of real-time risk assessment. Specifically, the ReLU activation function, RMSE loss function, and Adam optimization algorithm are selected. Based on the real-time operating status of the power system, the feature matrix is ​​constructed as follows: （12）， In the formula, This represents the number of system nodes. For nodes The load shearing power at the location; For nodes New energy power; For nodes Power of conventional generating units.

2. A model-data jointly driven real-time operation risk assessment system for power systems, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements a model-data co-driven real-time operation risk assessment method for power systems as described in claim 1.

3. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an information transmission implementation program, which, when executed by a processor, implements a model-data jointly driven real-time operation risk assessment method for power systems as described in claim 1.

Images