Abstract:This paper proposes a method of constructing a microservice-based post disturbance review system that supports multi-source information fusion on cloud computing platform, effectively solves the problems of high system coupling, poor flexibility, and difficult expansion of the current system, as well as satisfy the requirements of analyzing large-scale fault data within few minutes. According to the functional requirements, the system is reconstructed into four fine-grained microservices of grid model, accident record, operation record and human-computer interaction; on this basis, each data source is separately connected to a microservice module to avoid coupling between services and stacked. Combining the load parameters of the container cluster collected by the cloud platform monitoring system, a resource prediction model based on long short-term memory neural network is proposed to predict and schedule container resources in advance, avoiding the impact of sudden load changes on system efficiency and improve the container resource scheduling level. The result analysis shows that the reliability of the microservice-based post disturbance review system realized by the method reaches 99.999980%, and has pretty response efficiency.