The Establishment Of Diesel Engine Reliability Spectrum Based On Sample Set

Reliability engineering has been developed rapidly in recent years,and has been paid great attention and promoted in the field of vehicle engineering.For the whole vehicle,the reliability of the automobile engine has a great impact on it.The same type of engine can be used for different models and different engineering backgrounds,such as tractors,road transport vehicles,construction machinery vehicles and dump trucks.In the design stage of the product,it can be used for different use backgrounds according to different power,and the applicability division of a certain type of engine for different use backgrounds needs field test data for analysis and evaluation.In order to study this topic,a large number of field test data of the research group are summarized and sorted out,which are divided into four use backgrounds:tractor,road transport vehicle,construction machinery and dump truck.According to the description of the failure mode,the subsystem failure of each model and different use background is divided into six main subsystems:crank linkage mechanism,gas distribution,air intake and exhaust system,fuel system,supercharging system,cooling and lubrication system and starting system.The fault data distribution model is estimated for the samples with sufficient data volume,which have the best fitting degree with Weibull distribution through estimation.The reliability function of Weibull distribution is used to calculate the reliability directly,and then to judge which application background it is more suitable for.However,for the engine data with few fault samples,it is impossible to use the data to estimate its fault distribution,and then to obtain its reliability.Therefore,this paper makes a statistical analysis of the fault proportion of the subsystems of each type of engine in different use backgrounds,estimates the fault distribution model and calculates the reliability of the subsystem fault data under the large sample data,uses the sufficient data to establish the neural network for training,takes the statistical subsystem fault proportion as the input,and calculates the obtained subsystem fault proportion through the calculation The system reliability is trained as output.The trained neural network model is used to predict the subsystem reliability without reliability function under the small sample,and the subsystem reliability of 50 days and 100 days is predicted respectively.Here,the subsystem fault under the small sample data is considered to conform to the Weibull distribution,and the subsystem reliability function is deduced with the help of two groups of reliability data values.Then,the neural network model is established again,which takes the subsystem reliability as input and the whole machine reliability as output for training.The reliability of the sub-system predicted by the data and neural network is used to predict the reliability of the whole engine and the background of its application,so as to establish a complete reliability spectrum.The completed reliability spectrum is used to predict the reliability of the diesel engine under the small sample data,and the reliability of the whole engine and the subsystem of the two types of diesel engine is predicted respectively.The applicability of these types of diesel engine under different use background is analyzed,and the predicted results are added to the reliability spectrum again.When a new type of engine appears,the established reliability spectrum can be used to predict the reliability under different use backgrounds,and then a use background and appropriate prediction can be made for it.Finally,a reliability analysis interface based on SQL Server database system is developed to screen the engine model and use background,input the use days,and through the background program calculation,the reliability of the whole machine and its subsystems and the proportion of subsystem failure can be obtained under this screening condition.