The research background of this dissertation is the Sales Management Decision Support System which is the subsystem of the national"key-95"project -Tianjin Dagu Seamless Steeltube Plant's Management-Control Integration Project. The related academic problems in this subsystem are considered in this dissertation, and there are three main parts: 1. The research of contact sequencing optimization This part is aimed at the contact sequencing optimization module in this system, the main content are as follow: · Considering a bicriterion single-machine scheduling problem with mean flow time and tardy number of jobs. An NP-hardness proof is provided for the problem of minimizing the mean flow time subject to a limited number of tardy jobs. After one theom of Emmons' is extended, a dynamic programming algorithm for this problem is presented; · Considering a bicriterion single-machine scheduling problem with maximum tardiness and tardy number of jobs. An NP-hardness proof is provided for the problem of minimizing the maximum tardiness subject to a limited number of tardy jobs. According to some rules, a heuristic algorithm for this problem is presented; · Considering the minimizing the number of late orders with multiple job classes, and also discussing the complexity of the two different versions of this problem which are that the jobs in the same class should be processed together and the jobs in the same class can be processed disjointedly respectively. Also the pseudo- polynomial algorithms to these two versions of this problem are presented. 2. The application study of neural networks model in DSS The main contents of this part are as follow: · Studying the optimal structure of multilayed neural networks using statistical method. Under the assumption that the output error of the final node has a normal distribution, we educe the networks structure information criterion (NSIC), and apply the principal component analysis (PCA) to determine the initial structure of multilayered neural networks. · Studying multiple objective decision problems though use of Hopfield Neural Network The network mapping relations and optimal algorithms are presented for finite and infinite plans' multiobjective decision problems, also the computational procedures and examples are given. 3. The application study of Object-oriented technology in decision support system The main contents of this part are as follow: · Based on the structural method, this part mainly discusses the application study of object-oriented technology for system analysis and design in CIMS practice, and also the procedures and case study in system object-oriented analysis and design are presented; · User interface is an important part of DSS. The traditional user interface development methods have long developing period and poor software reusability, and leading object-oriented softw
修改评论