The Regulation And Parameter Real-time Identification Of Deep Drawing In Square Box Of Tailor-welded Blank

With the increasingly competitive auto market, the community called for security,economy, environmental protection, tailor-welded blank (TWB) being paid more and moreattention in the automotive industry has a lot of advantages such as increasing productionefficiency, reducing cost of production, conservation environmental protection, increasingsecurity. It becomes a hot topic in industry and academia. TWB of different thickness ordifferent materials leads to uneven metal flow and prone to some problems includingweld-line movement, wrinkling, cracking and so on. Those problems are difficult tocontrol. square box of TWB is the most common, relatively geometry rules, used widelyin a class of stamping，representative in the non-symmetric parts. The study of the formingprocess of square box is critical to the technology development of TWB.With combinations of the mold parameters derived, Blank-holder force is imposed onboth sides of materials to simulate study by DYNAFORM. Better solutions are obtained,the amount of weld-line movement is reduced while making the shape as much as possibleto meet depth at the same time. The Neural network model of deep drawing qualityparameters in square box of TWB is established making use of simulation andexperimental data to train and recognize the model. The convergence of identificationmodel can reach 1‰，In a randomized sample，the identification errors are relatively small.In addition, the sample complexity and the complexity of network structure on theefficiency of the model，accuracy and generalization ability are researched. The findingsshow that BP neural network suits the identification of deep drawing parameters in squarebox of TWB.In this paper, the orthogonal test simulation and neural networks are combined toobtained the model of deep drawing quality parameters in square box of TWB. Thefoundations are laid for intelligent conformity of deep drawing of square box of TWB.