Research On The Theory, Method And Application Of Performance Enhancement Design For Complex Product Under Uncertain Conditions

According to the hierarchy, coupling, fuzziness, iteration, cognition etc. of the characteristic of complex product design, the performance design is deeply integrated to product design process, and the performance enhancement design theory and approach of complex product under uncertain conditions is proposed systematically.The thorough research is conducted and the proposed approaches mainly consist of the following part, desired performance identification, behavior performance balance,structure performance adaptation, and prediction performance assessment. The approaches have been employed in the real product design with positive feedback,which verify the effectiveness and feasibilities of our proposed methods.The main contents of this dissertation are as follows:Chapter 1 illustrates the connotation and the research status of the performance design. The research status of the performance enhancement design and uncertainty design are also addressed. The key technologies of the performance enhancement design under uncertain conditions are proposed,and the research contents and the organization of the paper are given.Chapter 2 proposes a novel recursive analysis and identification approach for desired performance of product based on state perception. Firstly,a closed loop model of desired performance identification is conducted according to the evolution characteristic of performance in product lifecycle. Secondly, an integrated computation approach is proposed to forward recursive analysis of performance intent under incomplete information and reverse calculation of desired performance based on state perception. Meanwhile, fuzzy integral is applied to compensate the performance importance function, which realize the nonlinear fusion recognition and measurement of performance importance under the influence of demand domain and process domain.Chapter 3 proposes a novel balanced solution approach for behavior performance of product based on decoupling calculation. Firstly, the strength of coupling is computed according to the fuzzy behavior association-rule data of product, and design structure matrix clustering is used to initial decoupling planning of complex product,at the meantime,the behavior uit performance index is calculated. Based on the initial decoupling planning units, a hybrid multi-objective behavior performance model is established for further optimization to minimize maintenance cost, maximize quality stability, and maximize system availability. A discrete particle swarm optimization algorithm based on random search is proposed to solve the aforementioned model and a decision-making method based on entropy weight ideal point is put forward to acquire the optimal balanced outcome.Chapter 4 proposes a novel fuzzy adaptation approach for structure performance of product based on constraint transferring. First of all, information consistency transformation approach based on membership function is applied to performance constraint space which is simplified by rough set. An integrated filtering technology for constraint processing is proposed based on correlation similarity to realize the constraint-driven association mapping from function domain to structure domain.Then a hybrid multi-objective mathematic model with cost, quality and compatibility is established. A discrete differential evolution algorithm based on constraint level is proposed to solve the aforementioned model and a decision-making approach with minimum deviation of constraint satisfaction is put forward to obtain the final outcome which owns the best behavior performance.Chapter 5 proposes a novel trust assessment approach for prediction performance of product based on robust learning. First of all, a data preprocessing technology based on mutual information estimation and k-nearest neighbor analysis is applied to choose the design variables and identify the outliers in order to improve the precision of sample. A robust performance proofread prediction model is acquired by regressing fitting the hidden performance-design variables model. A reliability analysis method based on Bootstrap is used to performance analysis in order to conduct the incomplete prediction outcome. At the meantime, the crucial design variables is identified by sensitivity analysis.Chapter 6 develops the integrated platform of the perofoemance design systems for complex according to the fusion of the application and our achievements. The applications of the product design of forging equipment show the effectiveness and feasibility of the approaches we propose.Chapter 7 summarizes the main research contents and innovations, and gives the conclusions along with the future researches..