Study And Application Of Anti-misoperation System For Power Dispatch And Control Cloud Based On Machine Learning

With the increasing scale of the power grid,the number of equipment faced by operators at all levels of power grids is growing rapidly.The risk of misoperations in the daily work of writing operational tickets,daily operations,and operational mode adjustments is gradually increasing.It is too difficult to rely solely on human subjective consciousness to avoid misuse.In this context,the operation of anti-mistakes came into being,from the traditional five defenses to the current topology to prevent mistakes,anti-missing functions more and more,anti-missing is more and more wide,the current anti-missing system is mainly in the function The anti-error analysis is performed before the control command is sent.If the operation is an erroneous operation,the lock is performed to reduce the loss caused by the operation error.At the current stage,the anti-missing system is mainly based on the expert experience to establish an anti-error rule base,run the logic algorithm to prevent mis-constraints,and the anti-error rule library mainly relies on manual configuration,and the update method is difficult to adapt to the rapid development of the power grid.This paper starts from the construction needs of the new generation of anti-missing system,and studies the construction of intelligent anti-missing system.The focus is on the anti-missing analysis technology based on machine learning and the cloud-end integrated anti-error technology based on edge computing.The actual operation of the unit is a sample,and the sample training is used to form an error prevention model.Based on the advantages of regulating cloud data sharing,a cloud-wide interconnection and edge computing application system is constructed to improve the recognition accuracy and intelligence level of the control error prevention system and reduce misoperation.Risks provide support for the safe and stable operation of the grid.The main work and innovations of the thesis are as follows:1)Based on the principle of anti-error analysis and judgment,the anti-missing system is redesigned from the structure and process,and the construction of grid-based cloud anti-misoperation system based on machine learning is studied and applied.The actual power grid regulation automation main station.2)Combining the deep learning algorithm with the anti-error analysis theory,the paper introduces a stack-based sparse self-encoder to construct a neural network model based on machine learning,and uses PCA and SVM to improve the accuracy of error prevention analysis.3)Based on the theory of edge computing,the paper has built the anti-missing system to control the cloud,realize the integration of the provinces and counties of the anti-missing system,and solve the problem of the manual configuration of the units of the anti-error knowledge base in the past and the unified updating of the knowledge base.System intelligence provides a practical,efficient and safe way to prevent mistakes.