Research On Comprehensive Evaluation Of Aircraft Assembly Process

Assembly process is one of the major tasks in aircraft manufacturing process planning. Based on the fussy mathematics and support vector machine(SVM) theories, evaluation on aircraft assembly process was investigated. The main research contents and results are as follows:Aircraft assembly is characteristics of huge amount of parts and complicate process. To evaluate the assembly process systematically, 27 influencing factors(sub-indices) were firstly identified to apply to the process evaluation, from the four aspects of efficiency, precision of assembly, assembly costs, and ergonomics(main indices). Therefore, a two-level index-system is established. Furthermore,the various strategies of mathematical evaluation were developed for the sub-indices with different attributes. To be specific, membership function-based method is adopted to evaluate quantitative and qualitative indices with poor subjectivity, and questionnaire-based method is adopted to evaluate the indices with strong subjectivity.According to the characteristics of small samples available in assembly process evaluation, a SVM-based method of assembly process evaluation was proposed. Firstly, with the prior knowledge such as experiences and cases in aircraft assembly, a fuzzy comprehensive evaluation method is used to acquire the preliminary result on the assembly process, and thus, a training sample set could be obtained to train the SVM. To improve the performance of poorly-trained SVM without enough training samples, an evolutionary mechanism of SVM is introduced. If the prediction results from the SVM differ a lot from the fussy evaluation, the SVM should be re-trained with the expanded sample set with new selected samples. Therefore, the SVM has good scalability in classification with this evolutionary mechanism, and the applicability and accuracy of the proposed method of assembly process evaluation could be improved.Based on the research, a prototype of aircraft assembly process evaluation was developed by using the secondary development tool of CATIA- CAA. A case study was also conducted to validate the proposed methodology in this prototype.