août 2017 :

Rien pour ce mois

juillet 2017 | septembre 2017

Lifecycle Systems Improvement based on Maturity Evaluation and Reputation Analysis

Accueil > Communauté GOSPI > Thèses

Doctorant : Zhang Haiqing

  • Directeur : BOURAS Aziz
  • Laboratoire : DISP

In this thesis, a modern company is considered as a complex system which has its own life cycle and can be compared to a tree structure that contains neural networks and multiple child nodes. The Lifecycle system approach is usually used to improve data sharing, creation of correct product development processes, and reduction of energy and time through the whole process of the product’s life. The products are the main outputs to represent the competitiveness of the company. In order to improve the overall satisfaction and maturity of the company lifecycle system, we concentrate on figuring out four research issues by studying the elements of lifecycle products/systems. A TIFOS framework is proposed to conclude the fundamental functionalities of a lifecycle system. In addition, a PCMA maturity model is proposed to evaluate the AS-IS situation. A conception of fuzzy maturity is proposed. Fuzzy extended PCMA maturity model predicts the TO-BE situation. The basic research questions are validated and applied in three companies, which are located in Chengdu, Shanghai, and Shenzhen, respectively. The case studies show that the PCMA maturity model can efficiently achieve the overall maturity situation. Fuzzy extended PCMA maturity model is used to predict the TO-BE situation. The experiments of maturity analysis show that the predicted maturity levels are very close to the true values. After the maturity situation is clarified, the decision making issues should be addressed to answer how to select the right lifecycle system solutions or components to enhance the maturity of lifecycle system. These issues can be depicted into a hierarchy structure : objective, criteria, sub-criteria, and alternatives, which is typically multi-criteria decision making (MCDM) issues. We have studied the MCDM approaches and deeply analyzed five approaches. The real experiments are conducted and reveal that lambda-max, FPP, and integrated fuzzy AHP and VIKOR can credibly achieve the reasonable optimum selection. The inconsistency repairing algorithm has been presented to obtain a new modified consistent PCM (pairwise comparision matrix) for the related inconsistent one. The algorithm sets a linear programming problem based on all of the constraints. To obtain the optimum eigenvector of the middle value of the new PCM, the segment tree is used to gradually approach the greatest lower bound with the original PCM. The experiments for crisp elements show that the proposed approach can preserve more of the original information than references. The experiments for fuzzy elements indicate that the proposed algorithm can effectively reduce inconsistency and obtain suitable modified fuzzy PCMs. In order to solve the information extraction issue, this thesis jointly extracts three elements related to the products. This thesis proposed an opinion mining extraction algorithm. In detail, the algorithm automatically builds kernel structures to combine closet words into new terms from word level to phrase level, and we ensure the accuracy of opinion expressions and polarity based on three main parts. The performance of experiments demonstrates that the proposed algorithm outperforms the baseline approaches. Some interesting phenomena are discovered when studying 3,458 reviews in different domains. In conclusion, the thesis focuses on enhancing the overall satisfaction and maturity of the enterprise lifecycle system. In order to achieve this goal, several main contributions are proposed. The remaining issues related to lifecycle system and product information will be further studied.