Task Distribution And Dynamic Control Strategy In Knowledgeable Manufacturing Environment

Aiming at task distribution and dynamic control problems existing in knowledgeable manufacturing system (KMS), this dissertation firstly gives a study on the system structure of KMS. Also, the concept of character control and an optimization algorithm based on Tabu search are presented to solve the task distribution problems in KMS. Then three representative organization patterns of development teams are analyzed in detail for the purpose of product development task distribution, and a fuzzy neural network is trained to work as the decision-making system of development task distribution. A concept of the state transition system is proposed to analyze the knowledgeable manufacturing cell with an unreliable agent, and the optimal task control strategy of manufacturing cell is obtained. At last, aiming at the control problem of the knowledgeable manufacturing cell with multi-agents and multi-workpieces, the dissertation proposes a new matrix automaton to model and control the cell.The main content of this dissertation is introduced in detail as follows:1. Based on the system structure of KMS, the concept of character control and an optimization algorithm based on Tabu search are proposed to realize the decision-making of task distribution in the KMS. The running procedure of character control is described in detail, and some keyword definitions and detailed steps of the algorithm are given. The integration of this algorithm and character control enables the decision-making system of task distribution to respond to the variation of tasks and agents dynamically. The simulation results testify the validity of this method.2. The fast product development speed is an important factor for enterprises to improve their competitive capacity. To shorten product development time, the organization pattern of a team should be adaptive to the product; that is to say, the different product development tasks should be distributed to different development teams with suitable organization patterns. Three representative patterns are analyzed in detail, and a trained fuzzy neural network (FNN) model, whose inputs are some product character factors, is chosen to construct a decision support system. Then the decision results are used to make choice ofdevelopment team organization patterns. Several instances are given to describe how to use the proposed model and method in practice, and a satisfied result is obtained.3. A concept of the state transition system is proposed to solve the task control problems existing in the knowledgeable manufacturing system, which is used to analyze an unreliable knowledgeable manufacturing cell with uncertain factors. The uncertain factors of the knowledgeable manufacturing cell are included in the task control model by utilizing a self-study method of probability distribution parameters of stochastic events. The optimal task control strategy of the manufacturing cell is obtained by the combination of the uniform technology and the stochastic dynamic programming. The simulation results tesity the feasibility and validity of the control strategy.4. In order to solve the task control problems existing in the knowledgeable manufacturing cell (KMC) with multi-agents and multi-workpieces, a new matrix automaton is proposed to model the KMC. The automaton is of structured property and can be used to analyze the dynamic performance of the KMC. In this paper the stochastic dynamic programming is used to derive the optimal task control strategy of the automata, and a simulation method and some programs are proposed to simulate the running of the KMC. Compared to the random control principle, the objective function value of this control strategy is obviously lower.