Intelligent hierarchical control method for dynamic prediction of wide-area measurement data

Abstract

The invention relates to the technical field of power system control, in particular to an intelligent hierarchical control method for dynamic prediction of wide-area measurement data, which comprises the following steps of: S1, acquiring data and constructing a discrete time state equation; s2, continuous system matrix acquisition and modal analysis are carried out; s3, model prediction optimization control is carried out; and S4, intelligent hierarchical control is carried out. According to the intelligent hierarchical control structure, wide-area measurement data can be fully utilized, coordination control is carried out from the global level, the regional level and the local level, and the control precision and efficiency are improved; the global coordination layer can grasp the overall operation situation of the system and formulate a reasonable global control strategy; the area control layer can perform local optimization control according to the actual situation of the area and work cooperatively with other areas; and the local control layer can accurately control the equipment to ensure the effective execution of a control instruction.

Description

Technical Field

[0001] This invention relates to the field of power system control technology, specifically to an intelligent hierarchical control method for dynamic prediction of wide-area measurement data. Background Technology

[0002] With the continuous expansion and increasing complexity of power systems, their stable operation faces numerous challenges. The large-scale integration of new energy sources, such as wind power and photovoltaics, significantly deteriorates the dynamic performance of power systems due to their inherent low inertia and low damping characteristics. This leads to problems such as low-frequency oscillations, subsynchronous oscillations, or supersynchronous oscillations, seriously threatening system stability, equipment safety, and power quality, and limiting the efficient absorption of renewable energy.

[0003] Traditional power system stability control methods, such as damped controller design methods based on model predictive control and wide-area controller parameter design methods based on power system model identification, largely rely on accurate power system models. However, the actual operating conditions of power systems are complex and variable, making it difficult to obtain accurate and real-time updated system models. This prevents these methods from achieving online prediction and control of new power systems. Furthermore, some existing control methods often target a single or a few controlled objects, such as controlling only wind turbines, local power system static stabilizers (LPS), and static var compensators (SVC), lacking coordinated control among multiple controlled objects such as new energy sources, traditional thermal power units, converter stations, power system stabilizers (PSS), and SVCs. Failure to consider the interrelationships between controlled objects during parameter design may lead to performance degradation of new power systems or even system instability.

[0004] The rapidly developing Wide Area Measurement System (WAMS) offers a new opportunity to address these issues. WAMS can acquire real-time dynamic trajectory information of the system. How to fully utilize this measured dynamic trajectory information to construct a safe, stable, real-time emergency control system that is independent of system parameters, possesses adaptive capabilities, and exhibits global coordination performance has become crucial for the advancement of power grid transient stability control technology. Among these challenges, the most important and primary issue to be addressed is how to quickly and accurately predict real-time transient instability in power systems and achieve effective hierarchical control. The accuracy of the prediction results directly affects the reliability of emergency control decisions, and has significant theoretical and practical engineering application value for preventing large-scale power outages.

[0005] To address the aforementioned problems, we propose an intelligent hierarchical control method for dynamic prediction of wide-area measurement data. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent hierarchical control method for dynamic prediction of wide-area measurement data, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A smart hierarchical control method for dynamic prediction of wide-area measurement data includes:

[0009] Step S1: Use a wide-area monitoring system to collect measurement data of the controlled new power system to construct discrete-time state equations;

[0010] Step S2: Perform a continuous transformation on the state equations of the discrete-time system to obtain the system matrix of the continuous system, and use numerical calculation methods to perform modal analysis on the system matrix;

[0011] Step S3: Based on the modal analysis results, perform model predictive optimization control on different control objects in the system, and use the rolling optimization algorithm to solve the optimization problem online to obtain the optimal control strategy at the current moment;

[0012] Step S4: Implement intelligent hierarchical control based on the optimal control strategy, specifically including setting up a three-layer structure: a global coordination layer, a regional control layer, and a local control layer.

[0013] The global coordination layer collects system-wide information from the wide-area measurement system, assesses power system stability, identifies dominant oscillation modes and weak links, and formulates global control objectives and coordinated control strategies.

[0014] The regional control layer receives instructions from the global coordination layer, collects wide-area measurement data for its region, performs local stability analysis, and coordinates the control of the controlled objects within its region in conjunction with the global objectives and the actual situation of the region.

[0015] The local control layer directly controls specific equipment, and performs precise control of the equipment based on the instructions of the regional control layer and the real-time operating status of the equipment.

[0016] Preferably, step S1, data acquisition and discrete-time state equation construction, includes:

[0017] Using a wide-area monitoring system to collect measurement data from the controlled new power system, including bus voltage amplitude, phase angle, and line power flow, to construct past state measurement moments. Current state measurement matrix Control measurement matrix By constructing the matrix, the discrete-time state equation of the new power system is obtained using a system identification algorithm:

[0018] ,

[0019] in For the first The state vector at time t, For the first Control vector at time, and These are the discrete-time system matrix and the input matrix, respectively. This is the process noise vector.

[0020] Preferably, step S2, continuous system matrix acquisition and modal analysis, includes:

[0021] The system matrix of the continuous system is obtained by performing a continuous transformation on the state equation of the discrete-time system. The system matrix is ​​then subjected to modal analysis using numerical calculation methods to calculate the eigenvalues ​​and eigenvectors of the system. Based on the modal analysis results, the nodes with larger participation factors are selected as the installation locations of the controller, and the signals that have a greater impact on the oscillation modes are selected as the control signals.

[0022] Preferably, step S3, model prediction optimization control, includes:

[0023] Based on the novel dynamic model of the power system obtained through data-driven identification, the objective function is to minimize the weighted sum of the oscillation amplitude of key system measurements and the adjustment amplitude of controller actions.

[0024] ,

[0025] in For the first The oscillation amplitude of a key measurement For the first The adjustment range of the controller action, and Using weighted coefficients, model predictive optimization control is performed on different controlled objects within the system. The rolling optimization algorithm is used to solve the optimization problem online and obtain the optimal control strategy at the current moment.

[0026] Preferably, step S3 includes: in the continuous system matrix acquisition and modal analysis, the modal analysis results are also used to evaluate the stability of the system. If the real part of the eigenvalue is greater than zero, the system is unstable and corresponding control measures need to be taken.

[0027] Preferably, step S4 includes: in intelligent hierarchical control, the global coordination layer and the regional control layer, and the regional control layer and the local control layer interact with each other through a communication network, and the communication network adopts a redundant design.

[0028] Preferably, the method further includes: evaluating the effectiveness of the control method by comparing the oscillation amplitude of key system measurements and the damping ratio of the system before and after control; if the control effect does not meet expectations, adjusting the control parameters or control strategy.

[0029] Preferably, the data acquisition includes: using multimodal sensors to acquire data in order to obtain more comprehensive power system operation information, wherein the multimodal sensors include voltage sensors, current sensors, and power sensors.

[0030] Preferably, step S3 further includes: in model prediction optimization control, using artificial intelligence algorithms to train and optimize the model to improve the model's prediction accuracy and control performance, wherein the artificial intelligence algorithms include neural network algorithms and genetic algorithms.

[0031] Preferably, step S4 further includes: in intelligent hierarchical control, the regional control layer dynamically adjusts the control strategy according to the load changes in its region. When the load increases, the active power output of the generator is increased, and the compensation capacity of the SVC is adjusted to maintain voltage stability.

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] This invention proposes and designs a method for predictive control of different objects within a system using a system model identified from wide-area measurement data. This method enables online identification and prediction of the system model through data-driven approaches, eliminating the reliance on precise power system models and improving the adaptability and flexibility of the control method.

[0034] This invention realizes the coordinated control of different control objects in a novel power system. It comprehensively considers the mutual influence between various control objects such as new energy sources, traditional thermal power units, converter stations, PSS and SVC. By optimizing the control strategy, it enhances the damping level of the system, effectively suppresses the oscillation of the power system, and improves the stability and reliability of the power system.

[0035] The intelligent hierarchical control structure of this invention can fully utilize wide-area measurement data to coordinate control at three levels: global, regional, and local, thereby improving control accuracy and efficiency. The global coordination layer can grasp the overall operating status of the system and formulate reasonable global control strategies; the regional control layer can perform local optimization control based on the actual situation of its region and work in coordination with other regions; the local control layer can precisely control the equipment and ensure the effective execution of control commands. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] like Figure 1 As shown, the intelligent hierarchical control method for dynamic prediction of wide-area measurement data includes the following steps:

[0039] Step S1: Data Acquisition and Discrete-Time State Equation Construction

[0040] Considering that power system sampling measurement data is discrete and includes control inputs from different stability controllers, the wide-area monitoring system collects measurement data from the controlled new power system, including bus voltage amplitude, phase angle, and line power flow. A past state quantity measurement matrix is ​​constructed. Current state measurement matrix Control measurement matrix .

[0041] By constructing the matrix, the discrete-time state equations of the novel power system are obtained using a system identification algorithm:

[0042] ,

[0043] in For the first The state vector at time t, For the first Control vector at time, and These are the discrete-time system matrix and the input matrix, respectively. This is the process noise vector.

[0044] Step S2, Obtaining the Matrix and Modal Analysis of the Continuous System:

[0045] The discrete-time system state equations identified from measurement data can be transformed into continuous-time equations to obtain the system matrix of the continuous system. Numerical methods, such as the bilinear transformation, are used for the transformation. The system matrix is ​​then processed. Modal analysis is performed to calculate the eigenvalues ​​and eigenvectors of the system. Based on the modal analysis results, nodes with larger participation factors are selected as the installation locations for the controller, and signals that have a greater impact on the oscillation modes are selected as the control signals.

[0046] The discrete-time system state equations identified based on measurement data are as follows:

[0047] By performing a continuous transformation, the system matrix of the continuous system can be obtained. Numerical calculation methods, such as the bilinear transform method, yield the following transformation formula:

[0048] ,

[0049] in, The sampling period is It is an identity matrix.

[0050] For system matrix Perform modal analysis and solve the characteristic equation:

[0051] Thus, the eigenvalues ​​of the system can be calculated. and the corresponding feature vector .

[0052] This formula is for solving matrices. Eigenvalues The characteristic equation of . Where, This indicates finding the determinant. It is the identity matrix. Let be the eigenvalue variables to be solved. It is a known matrix. The equation is calculated by working through the matrix. The determinant of the matrix is ​​set to zero, and we obtain the information about... The polynomial equation, the solution of which is a matrix. The eigenvalues ​​of the system. Eigenvalues ​​have important applications in system stability analysis, dynamic system modeling, and other fields. By solving the eigenvalues ​​obtained from this equation, the dynamic characteristics of the system can be analyzed, and whether the system is stable can be determined.

[0053] Based on the modality analysis results, the participation factors were utilized. Evaluation Node For the The degree of participation of each oscillation mode, The calculation formula is:

[0054] ,

[0055] in, and These are the corresponding elements of the left and right eigenvectors, respectively. The node with the larger participation factor is selected as the controller installation location, and the signal with the greater influence on the oscillation mode is selected as the control signal.

[0056] This formula is used to calculate indicators. Its meaning is: molecule Representing vectors and The absolute value of the product, the denominator Yes From 1 to Within the range, and The absolute value of the sum of products. The final calculation result is the ratio of the numerator to the denominator, which is usually used to measure the proportion, weight or correlation of a certain relationship based on vector multiplication. The specific meaning needs to be further clarified in conjunction with the actual application scenarios of the formula in the intelligent hierarchical control method for dynamic prediction of wide-area measurement data.

[0057] Step S3, Model Prediction Optimization Control:

[0058] Based on the novel dynamic model of the power system obtained through data-driven identification, the objective function is to minimize the weighted sum of the oscillation amplitude of key system measurements and the adjustment amplitude of controller actions.

[0059] ,

[0060] in, and These are the measurement error weighting coefficient and the control cost weighting coefficient, respectively, used to adjust the relative importance of the two optimization objectives;

[0061] Indicates to Sum of the absolute values ​​of the prediction errors of each measurement value. Representing the The difference between the predicted and actual values ​​of a measurement is used to minimize the prediction bias of the measurement data.

[0062] That is to Sum of the absolute values ​​of the changes in each control variable For the first The adjustment amount of each control variable is used to control the cost of the control behavior. This is achieved by minimizing the objective function. It can ensure prediction accuracy while reasonably constraining the resource consumption of control operations, and achieve dynamic optimization of intelligent hierarchical control.

[0063] Step S4, Intelligent Layered Control:

[0064] The first layer, the global coordination layer, is responsible for collecting system-wide information from the wide-area measurement system, including the operating status of each regional power grid and electrical quantities at key nodes. Based on this information, it conducts a global assessment of the stability of the entire power system, identifying the dominant oscillation modes and weak points. According to the overall system operation, it formulates global control objectives and coordinated control strategies to guide lower-level control. For example, when low-frequency inter-regional oscillations occur, the global coordination layer identifies the regions involved in the oscillations and the key control objects requiring adjustment.

[0065] The second layer: Regional control layer. This layer receives instructions from the global coordination layer and simultaneously collects wide-area measurement data within its region, performing local stability analysis of the power system within that region. Combining global control objectives with the actual conditions of the region, it coordinates and controls the controlled objects within its region. For example, in a regional power grid, based on the generator output, load changes, and power flow through interconnections with other regions, it adjusts the excitation control parameters of generators and the compensation capacity of SVCs within the region to suppress regional oscillations and, in conjunction with global control strategies, achieves coordinated operation with other regions.

[0066] The third layer: Local control layer. This layer directly controls specific equipment, such as the generator's speed governor, excitation regulator, and SVC control device. Based on control commands issued by the regional control layer and combined with the real-time operating status of the equipment, it precisely controls the equipment. For example, the generator's speed governor adjusts the valve opening according to commands from the local control layer to change the generator's active power output; the excitation regulator adjusts the excitation current according to commands to control the generator's reactive power output and terminal voltage.

[0067] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart hierarchical control method for dynamic prediction of wide-area measurement data, characterized in that, include: Step S1: Use a wide-area monitoring system to collect measurement data of the controlled new power system to construct discrete-time state equations; Step S2: Perform a continuous transformation on the state equations of the discrete-time system to obtain the system matrix of the continuous system, and use numerical calculation methods to perform modal analysis on the system matrix; Step S3: Based on the modal analysis results, perform model predictive optimization control on different control objects in the system, and use the rolling optimization algorithm to solve the optimization problem online to obtain the optimal control strategy at the current moment; Step S4: Implement intelligent hierarchical control based on the optimal control strategy, specifically including setting up a three-layer structure: a global coordination layer, a regional control layer, and a local control layer. The global coordination layer collects system-wide information from the wide-area measurement system, assesses power system stability, identifies dominant oscillation modes and weak links, and formulates global control objectives and coordinated control strategies. The regional control layer receives instructions from the global coordination layer, collects wide-area measurement data for its region, performs local stability analysis, and coordinates the control of the controlled objects within its region in conjunction with the global objectives and the actual situation of the region. The local control layer directly controls specific equipment, and performs precise control of the equipment based on the instructions of the regional control layer and the real-time operating status of the equipment.

2. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Step S1, data acquisition and discrete-time state equation construction, includes: Using a wide-area monitoring system to collect measurement data from the controlled new power system, including bus voltage amplitude, phase angle, and line power flow, to construct past state measurement moments. Current state measurement matrix Control measurement matrix By constructing the matrix, the discrete-time state equation of the new power system is obtained using a system identification algorithm: ， in For the first The state vector at time t, For the first Control vector at time, and These are the discrete-time system matrix and the input matrix, respectively. This is the process noise vector.

3. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Step S2, continuous system matrix acquisition and modal analysis, includes: The system matrix of the continuous system is obtained by performing a continuous transformation on the state equation of the discrete-time system. The system matrix is ​​then subjected to modal analysis using numerical calculation methods to calculate the eigenvalues ​​and eigenvectors of the system. Based on the modal analysis results, the nodes with larger participation factors are selected as the installation locations of the controller, and the signals that have a greater impact on the oscillation modes are selected as the control signals.

4. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Step S3, model prediction optimization control, includes: Based on the novel dynamic model of the power system obtained through data-driven identification, the objective function is to minimize the weighted sum of the oscillation amplitude of key system measurements and the adjustment amplitude of controller actions. ， in For the first The oscillation amplitude of a key measurement For the first The adjustment range of the controller action, and Using weighted coefficients, model predictive optimization control is performed on different controlled objects within the system. The rolling optimization algorithm is used to solve the optimization problem online and obtain the optimal control strategy at the current moment.

5. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Step S3 includes: In the continuous system matrix acquisition and modal analysis, the modal analysis results are also used to evaluate the stability of the system. If the real part of the eigenvalue is greater than zero, the system is unstable and corresponding control measures need to be taken.

6. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Step S4 includes: In intelligent hierarchical control, the global coordination layer and the regional control layer, and the regional control layer and the local control layer interact with each other through a communication network, and the communication network adopts a redundant design.

7. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Also includes: The effectiveness of the control method is evaluated by comparing the oscillation amplitude and damping ratio of key system measurements before and after control. If the control effect does not meet expectations, the control parameters or control strategy are adjusted.

8. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, The data acquisition includes: using multimodal sensors to acquire data in order to obtain more comprehensive power system operation information. The multimodal sensors include voltage sensors, current sensors, and power sensors.

9. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Step S3 also includes: in model predictive optimization control, using artificial intelligence algorithms to train and optimize the model to improve the model's prediction accuracy and control performance. The artificial intelligence algorithms include neural network algorithms and genetic algorithms.

10. The intelligent hierarchical control method for dynamic prediction of wide-area measurement data according to claim 1, characterized in that, Step S4 also includes: In intelligent hierarchical control, the regional control layer dynamically adjusts the control strategy according to the load changes in its region. When the load increases, the active power output of the generator is increased, and the compensation capacity of the SVC is adjusted to maintain voltage stability.

Images