Emulsifier Fault Diagnosis Based On Back Propagation Neural Network Optimized By Particle Swarm Optimization

With the development of high-speed railway, water conservancy construction and mineindustry, emulsion explosive has been widely used. After decades of development and improvement,some other kinds of industrial explosive which are not eco-friendly and low safety have beenreplaced by the emulsion explosive. So the role of emulsion explosive is more and more important.Emulsion explosive promotes economic growth in recent years, but it also causes several accidents.How to guarantee the economic benefits while ensure the safety of workplace at the same time isthe principle to producers. To solve the problem which almost accidents are caused by emulsifier,the government regulatory agencies has regulated the production safety systems and standardsgreatly; on the other hand, enterprises have enhanced the safety awareness of self-improvement. Sothe improvement of on-line monitoring and fault diagnosis system of emulsifier become more andmore important.On the basis of studying the equipment fault diagnosis technology and development trend, thepaper firstly present an emulsifier fault diagnosis system of optimizing the BP neural network basedon improved particle swarm optimization. Some studies in this thesis are listed as follows.(1) Taking the AE-HLC-III open primary emulsifier as the study object, the main fault typeswhich may occur in the production process and the fault mechanism are analyzed based on thestructure of the emulsifier. Then the paper also has researched the approach and process of theemulsifier’s vibration signals.(2) Combining with the original monitoring technology and equipment in the continuousproduction line of emulsion explosive, a set of solutions about the fault diagnosis of emulsifier ispresented. Here is the specific idea. The sensors collect the fault signal of emulsifier and transmittedto the PLC. Basing on the system design, the fault diagnosis VB software of emulsifier which couldanalyze and diagnose the failure data come from PLC is installed on the IPC. The software could bewell compatible with original systems and Kingview software.(3) Based on the study of the basic principles of particle swarm algorithm, the selection ofrelevant parameters is analyzed. Aiming at disadvantages of BP neural network which has a slowrate of convergence and is easily to fall into local minimum，a method of optimizing the BP neuralnetwork based on improved particle swarm optimization was presented. Through the simulation test,the experimental results show that the optimized BP neural network forecasts more accurately, andcould be applied in the emulsifier’s fault diagnosis which has complex nonlinear mapping betweenthe fault type and fault symptom. (4) According to the basic principles of the hardware system design, a complete hardwaresystem including the vibration sensors, signal transmitters, PLC and host computer has been set up.Realize the software development of emulsifier’s fault diagnosis system based on the completion ofthe hardware system. The fault diagnosis software is mainly based on VB software development,combined with the data exchange of Access database, mixed programming of MATLAB, and theKingview software used to monitor entire production line.(5) The rotor imbalance of emulsifier is selected as a test fault. The test results showed that thisfault diagnosis system can identify the actual fault of the emulsifier accurately. So the system couldbe applied to fault monitoring and diagnosis of emulsifier in production. Ensure the safety andreliability of the emulsifier operation, while the fault diagnosis system reduces the maintenancecosts and improve the maintenance efficiency. The system has very good application prospects.