A Study On Marine Diesel Engine Abnormal Thermal Fault Simulation And Diagnosis

With the development of modern technology,the marine diesel engine has been changed over to the large-scale,comprehensive and intelligent one,which needs increasingly stringent requirements in its operation and maintenance.The marine diesel engine as the main power source of ship,and when it fails,will directly affect the ship's steering performance and the safety of equipment and personnel.Therefore,the study of marine diesel engine fault diagnosis has been a hot research topic.This paper has simulated the normal working condition and abnormal working condition of the diesel engine with AVL-BOOST,and identified abnormal thermal fault characteristic parameters of the diesel engine with BP neural network,then diagnosed the fault type and degree.Major work of this paper is as follows:(1)According to the perspective of the mechanism of diesel modeling,this article get a detailed study on the diesel engine works and mathematical model of each sub-module On these bases,build a four-stroke diesel zero-dimensional numerical model with AVL-BOOST,the debugging process model's problem should be explained and give solutions.The final calculation results with the experimental results of the model were compared to verify the accuracy of the model.the results show that the simulation model with high precision on the thermal parameters,meets the research needs of this article.(2)From the diesel engine functions and composition of these two perspectives,to analyze the type of thermal breakdown of diesel and extract the diesel characteristic parameters to reflect the current operating status.Select several typical diesel engine thermal breakdown,simulate these types of failures by changing the parameters of the model,and analysis the simulation results through the working principle of diesel engine.The analysis results showed that the fault simulation results trend the actual situation is consistent.(3)Then introduce the theory and implementation of artificial neural network technology,expatiate the establishment of neurons mathematical models,analyze the neurons characteristics of the transfer function.Several typical network model structure is introduced to study the rules of mathematical learning principles,on the bases of these theories,focus on the algorithm of BP neural network,and how to use programming tools to achieve it.These works guarantee to solve the pattern recognition problem of this paper.With full use of the extracted characteristic parameter fault simulation results,create the appropriate network model for pattern recognition to solve the problem of this paper.In order to improve the efficiency of network,train the sample data preprocessing.Choose the way to be the training network algorithm which is based on first order gradient decreased and numerical optimization,and analyze the characteristics and effects of the two algorithm.different levels of noise are added to the samples to test the network which is trained by optimal algorithm.The test results showed that there are good classification results and capacities of it,which could diagnose the fault type and degree accurately.