Research On Power Oscillation Monitoring And Warning And Disturbance Source Location Of The New Power System Driven By Response

With the continuous deepening of the national energy green transformation,a large amount of new energy is integrated into the power system to build a new type of power system,which is one of the important tasks to achieve the dual carbon goals of "peak carbon emissions" and "carbon neutrality." The power system presents a trend of "double high"(i.e.,high proportion of new energy integration and high proportion of power electronics,abbreviated as "double high"),resulting in a more complex topology of the power system and more factors affecting the stability of the power system.Thus,various preventive and control measures need to be further improved.The traditional model-based power system stability assessment and corresponding preventive and control measures for power oscillations are no longer suitable for the operational requirements of the new power system.With the rapid development of the wide-area measurement system,research on power oscillation monitoring and disturbance source localization by response has received increasing attention.The dynamic response of the system contains rich and easily extractable data characteristics,which can reflect the dynamic characteristics of the system under normal operating conditions.This article mainly studies the parameter identification,evaluation,stability,and disturbance source localization of power oscillation patterns in the power system.First,the main mode parameters of power oscillation in the system are identified by the power system’s random response data.The potential weak damping mode of the system is recognized,and artificial intelligence is used to enhance the intelligent identification of mode parameters.Second,the location identification of disturbance sources is studied using an AI-enhanced method.The real-time tracking of the power system oscillation state is achieved,and the monitoring and early warning of power oscillations are carried out.The disturbance source’s location is accurately determined,and potential safety issues are discovered,providing real-time guidance for subsequent control strategies.The innovative research contents of this article include:(1)A numerical analysis method based on response-driven power oscillation monitoring and warning is investigated,utilizing random perturbation-driven responses.Randomized Dynamic Mode Decomposition(Randomized-DMD)and Compressed Dynamic Mode Decomposition(Compressed-DMD)techniques are introduced to study power oscillations in electrical power systems.The innovation lies in the dimensionality reduction of the original input data using different types of random matrices.Randomized-DMD employs randomized independent identically distributed Gaussian matrices,while Compressed-DMD utilizes sparse row matrices.Both methods ensure accurate calculation of mode parameters while improving computational efficiency,demonstrating robustness and addressing issues such as the curse of dimensionality and low computational efficiency associated with traditional DMD methods and other numerical analysis approaches.(2)Applying AI-enhanced methods for online monitoring and early warning of power oscillations in the power system,an accurate real-time online evaluation of power oscillation monitoring in the power system is achieved.A multi-layer artificial neural network is introduced to calculate randomized dynamic mode decomposition to obtain a model trained with mode parameters,realizing real-time online evaluation of power oscillation monitoring in the power system.The innovation of this method is to establish a mapping relationship between the system response signal and mode parameters through a multi-layer artificial neural network,achieving real-time monitoring and early warning of the power system’s power oscillation through offline training and online application,solving the problem that numerical analysis methods require a certain amount of accumulated data for calculation and achieving accurate real-time online monitoring.(3)A combined energy function and normal distribution identification method is applied to study the location of power oscillation disturbance sources in power systems.Based on the mechanism of forced power oscillation in power systems,a numerical analysis disturbance source location method combining energy function and normal distribution identification is proposed,using system response data to locate potential disturbance sources in the system.This method locates the disturbance sources of minor load random fluctuations and large-amplitude forced power oscillations.The innovation of this method is the application of the 3 Sigma criterion to identify the location of disturbance sources,which solves the problem of manual decision-making required by traditional energy function methods,quantifies the location of power oscillation disturbance sources,and improves the efficiency and accuracy of the location.(4)A quasi-real-time localization and analysis study of disturbance sources in power oscillations in power systems based on logistic regression is conducted.Building upon the energy function and normal distribution identification method for disturbance source localization,an AI-enhanced method based on energy function is proposed.It employs logistic regression to perform regression learning and modeling on the computed results of the energy function,guided by the 3 Sigma model of the normal distribution.This method enables quasi-real-time localization and analysis of disturbance sources in power oscillations in power systems,providing corresponding guidance for subsequent control measures.The innovation lies in utilizing logistic regression to establish an AI classifier model and guiding the labeling of energy function computation results with the 3 Sigma criterion.This approach addresses the need for a large volume of data in the identification of power oscillation disturbance sources in power systems.