Research On Data-driven Assembly Quality Prediction And Quality Control Of Diesel Engine

The development of the diesel engine industry is closely related to our country’s automobile industry,agricultural machinery,construction machinery,power generation equipment and other industries.In recent years,with the development of artificial intelligence and big data technology,diesel engine manufacturing plants are gradually transforming to intelligent,informative,digital,and high-end.Data-driven diesel engine intelligent manufacturing play an important role in the plan of “Made in China 2025”.As a typical complex mechanical product,the assembly process of a diesel engine product involves multiple stages and a large number of manufacturing process parameters,which directly determine the final product performance and quality.In order to improve the assembly quality of diesel engines,it is necessary to reasonably set quality control points during the assembly process to ensure that various assembly parameters meet production requirements.Due to the complexity of the production process of diesel engine enterprises,the monitoring and control of assembly quality needs to consider setting many quality control points,which increases the difficulty of quality monitoring and control.In addition,when all the process parameters of the diesel engine meet the process specifications,the engine assembly quality may still cannot meet the requirements,and the quality situation is unpredictable.Therefore,it is of great significance to realize the intelligent prediction and control of diesel engine assembly quality.The research of this paper comes from the product quality improvement demand of Yu chai Group,which is a large domestic diesel engine manufacturer.The main research contents are as follows:(1)Analysis of diesel engine manufacturing process and data preprocessing.Firstly,it specifically introduces the complete machine assembly process and bench test process in the production of diesel engines,and then visually analyzes the typical parameters of diesel engines.According to the results of the analysis,the data is preprocessed with missing values,abnormal values,and standardization to prepare data for the subsequent establishment of a diesel engine assembly quality prediction model.(2)Established diesel engines assembly quality prediction model.Aiming at the problem of imbalanced diesel engine data set and high-dimensional manufacturing parameters,an imbalance learning prediction method and a manufacturing process parameter selection method are designed respectively.In the imbalanced learning prediction method,the samples that are difficult to be classified correctly are identified based on the classification confidence,synthesizing new samples adaptively,and combined with integrated learning to improve the model performance.In the manufacturing process parameter selection method,the maximum information coefficient is used to measure the correlation degree between the parameters and assembly quality,and combined with particle swarm algorithm for feature selection.Finally,the two methods are wrapped to form a prediction model for the assembly quality of the whole machine.(3)The process parameter optimization method of diesel engine assembly process is studied.Considering the multi-stage diesel engine assembly process,on the basis of data discretization,a Bayesian network model is established based on expert experience and search-evaluation algorithms.The Bayesian network is used to infer the assembly quality of diesel engines,and the quality inference results are used as the fitness function of the intelligent algorithm to search in the process parameter space to select the optimal process parameters,so as to achieve high quality level in the assembly process of diesel engines.