Research On Combustion Fault Diagnosis Of Marine Auxiliary Boiler Based On Neural Network

Marine auxiliary boiler is an important part of the vessel.As one of the important equipment to ensure the normal operation of the ship,it is mainly used to produce saturated water vapor.The safe,stable and efficient operation of marine auxiliary boiler is of great significance to ship’s safety and economic benefits.With the continuous development of artificial intelligence and ship automation technology,the field of fault diagnosis and state recognition of ship equipment has gradually become a hot research topic.However,the structure of the ship system is complex and the fault features are various.It is difficult to obtain a comprehensive fault sample at this stage,so the research and application of the combustion.fault diagnosis of the ship auxiliary boiler is still in the immature research stage.In recent years,neural network method has made important achievements in many research fields.It has very important research value to introduce it into the combustion fault diagnosis of marine auxiliary boiler.In this paper,the self-organizing feature map(SOM)neural network is selected to study the combustion fault diagnosis of marine auxiliary boiler.And the corresponding neural network is improved to improve the accuracy of diagnosis,so as to improve the safety of ship operation.The paper selects DMSVLCC,the large oil tanker engine simulator,developed by Dalian Maritime University as a test platform to simulate the normal operation of D-type water pipe auxiliary boiler,fuel supply pump wear,dirty fuel preheater,ignition pump failure,and fan failure.The sample data was extracted and preprocessed by principal component analysis(PCA),which was used as a validation study of the subsequent fault diagnosis methods.Using the experimental data to train SOM neural network for preliminary diagnosis,because of the limitations and shortcomings of neural network,the diagnosis results are not ideal.In order to improve the accuracy of fault diagnosis,on the basis of neural network,particle swarm optimization(PSO)is used to optimize the updating process of its weight vector and improve the accuracy of diagnosis.By analyzing the diagnosis results of the experiment and the advantages and disadvantages of the optimization algorithm,this paper combined with learning vector quantization neural network(LVQ)to make up for the shortcomings of the algorithm in the process of competition,and carried out further algorithm research.Finally,the model of PSO-SOM-LVQ hybrid neural network algorithm fault diagnosis is verified by using the sample data of system running simulation.According to the comparison of three experimental results,it is shown that the fault diagnosis results of the hybrid neural network model are consistent with the actual state,and the accuracy of the hybrid neural network model is significantly improved compared with the former two algorithms,which fully verifies the reliability and accuracy of the algorithm in the combustion fault diagnosis of the ship’s auxiliary boiler,and provides a new idea for the development of the intelligent diagnosis of the combustion fault of the ship’s auxiliary boiler.