Autobody Dimensional Fault Diagnosis And Quality Control Based On A Dynamic Assembly Variation Modeling Approach

The quality of auto body is a key index of the vehicle manufacturing capability.The variation level of auto body is one of the most important assessment criterions of the vehicle.It would affect the comfort,noise perception,driving experience and performance,which are of tremendous value in affecting the customers' purchasing intention.The production of a car includes stamping,welding and assembling process with many stations involving hundreds of incoming parts.It is a typical complex system with multiple stations,containing many product characteristics and complicated variation sources.This requires the manufacturers to integrate valuable process data and useful information considering the real manufacturing conditions.Traditional auto body variation diagnosis and control is design oriented and experience based.It could not integrate the dynamic process information and real-time data during production,resulting the low accuracy and efficiency.This paper focuses on the dynamic variation characteristics analysis related to auto body dimensional variation and presented a reliability analysis framework which is in consisting with the real manufacturing process.Combined with the theory of auto body variation control,the dynamic auto body fault diagnosis and control method are thus proposed.The interest and novelty of this paper are as follows:(1).The reliability analysis framework for the serial-parallel systemThis is driven by the dynamic characteristics of the manufacturing system,which also have great impact on the auto body dimensional variation.The key dynamic factors are investigated.The auto body variation simulation is achieved by an improved state space model which has integrated the function of serial-parallel system,process capability and production ability.With the analysis,a reliability assessment framework is proposed.It is close to the real condition of the system.(2).Assembly variation diagnosis based on dynamic Bayesian networkDriven by the difficulty of auto body fault diagnosis,this paper adopts the dynamic Bayesian network.Based the variation input-output relationship,the static and dynamic nodes of dynamic Bayesian network is defined.With the help of dynamic variation propagation model,the structural and parameter learning is completed.Compared with static Bayesian network,the diagnostic accuracy and efficiency are improved.(3).The preventive maintenance of auto body fixture systemIt is presented to solve the current over-rely on the industrial experience.Thedynamic maintenance strategy is proposed with the quality constrain of the manufacturing system aiming to saving maintenance cost.The new method would update the maintenance cycle for each locating element with the change of process parameters.Thus the functionality of the system and the products could be ensured.This paper could improve the dimensional faults diagnosis and control ability of the production system.It could support the auto body quality improvement strategy in the long run.