Design And Implementation Of Equipment Fault Early Warning System Based On Storm

Power plants are one of the important infrastructures in any country.In order to ensure that the power plant equipment can run smoothly for a long time,the equipment of the power plant is monitored through the fault warning system,and the faults are early warning and thus maintained in advance.However,there are nearly 10,000 measuring points in a basic thermal power generating unit.In the face of a large amount of monitoring data generated continuously,it is necessary to design a system that can process real-time data streams and find fault measuring points.In order to deal with the real-time data flow generated by power plant equipment,this paper designs a device fault early warning system based on Storm.Firstly,the model is built to maintain a set of interrelated measuring points,the historical data of the normal running time of these measuring points is selected,the data model is established by Gaussian mixture model clustering algorithm,and the similarity between the data and the feature data of date model in each dimension is calculated by Storm.If the similarity is too low,it is considered to be an abnormal point for alarm.In order to avoid data matching speed and Storm processing speed mismatch,Kafka is added as a message queue to cache data,and Web services are used to manage the whole system,which enables technicians to better manage and reduce the operation difficulty.Finally,the simulated power plant generates a large amount of monitoring data for Storm to perform real-time calculation processing and analyze the results.The results show that the equipment fault early warning system can cope with a large amount of real-time data and accurately predict abnormal points,thus achieving early warning of equipment failure.