| Numerical control technology is the core for advanced manufacturing technology and its equipments. Nowadays, in order to improve manufacturing level, adaptability to the dynamic market, and possibility in rival activity, the technology has been adopted widely in the manufacturing industry all around the world. However CNC system has just been a sequent and logic executor since it came into being used. It lacks effective real-time feedback in the machining process. Even more it cannot cooperate with other intelligent decision-makers from the manufacturing system. As a result, NC programmers always make the programs in ready with his experience at advance. In making the programs they tend to use conservative machining conditions to avoid spindle overload or tool wear. The tendency brings about low cutting efficiency. Although feasible machining parameters could become available through complicate calculations, the programmer still cannot hold the idea whether the parameters are in accordance with the aim of manufacturing system or not. Holonic control architecture (HCA) is an integration of both hierarchical and heterarchical control architecture. It holds several advantages of advanced control architecture, such as adaptability, agility, autonomy and fault tolerance. The integration can provide a new development way for manufacturing system control architecture (MSCA). Because of these characters, it would be the best choice in developing MACA. The issue about NC machining intelligent control system with holonic control architecture principle is the main focus in this paper. Some key technologies are presented and developed. These contributions play an important role in improvement of NC machining level and efficiency and develop a new way in promote online process planning level. At the same time they can provide theoretic guidance in construction of HCA for low-level parts in manufacturing system.The main contributions in this dissertation are summed up as follows.(1) The principle and features of holonic architecture mode are discussed and its decision-making ability is analyzed. Based on the discussion, intelligent control system architecture is developed for CNC machining process to support holonic control. The architecture is divided into three parts, i.e. on-line, off-line and real-time planning. Then the on-line part with multiple decision objectives is mainly considered and discussed. Further a combined algorithm based genetic algorithm (GA) and metamorphic greed heuristics (MGH) is proposed to find a solution for the planning. The hybrid algorithm realizes eliminating OR nodes and sequencing AND nodes from AND/OR graph used to represent the relationship of machining features. In the algorithm, a data containment tree structure is applied to convert the graph to be available for GA chromosome coding. In calculation of the individuals' fitness value, a heuristic relation matrix is provided to determine heuristic paths according to the tree. In the end, an illustration is presented to test the applicability of the proposed approach. The developed methodologies can help decision-makers improve process planning level in dynamic CNC machining circumstances.(2) The existent problems in traditional expert system are presented and discussed. Then a new approach for intelligent determination of milling conditions with machine learning based fuzzy logic is presented. Firstly, some problems about the application of traditional expert system in selection of machining parameters for milling process are analyzed and discussed. Then a fuzzy logic reasoning model is proposed to realize acquiring suitable milling feedrate according to three machining parameters: tool diameter, machining depth and material hardness. A hybrid method composed with artificial neural network and k-means cluster is designed to acquire the rules knowledge for fuzzy logic reasoning in the model. In the end, an illustration is presented to test the applicability of the proposed approach. The result shows good performance about knowledge representation in determination of milling conditions. As a result, the fuzzy logic can provide expert system with a new way in intelligent online selection of milling conditions.(3) Optimal cutting parameters have a great influence on reducing the production cost and time and improving the product quality. Relative key technologies are designed in developing optimal cutting parameters inference agent. A multi-objective optimal model is established to obtain patterns for the agent. Genetic algorithm (GA) is applied to solve the optimization issue. With the obtained patterns, a particle swarm optimization (PSO) based artificial neural network (ANN) is proposed to obtain milling-parameter matching knowledge. An optimization model is established to gain patterns for training the network. In order to improve training performance, a new method is introduced to optimize the network's structure at the same time as evolutionary computation. The result from illustrations shows that the modified PSO-trained ANN has a better knowledge-recovering performance than conventional PSO-based ANN, BP ANN and GA-based ANN. In the end optimal machining parameters inference agent is developed with Visual C++ as developing tool, SQL Server 2000 as DBMS, CJlibrary as human machine interface developing class library.(4) According to the requirement of holonic control architecture in CNC machining process, a decision-making mode of real-time planning holon is bring forth. The cooperative agent's function principle and structure in the model is analyzed. In order to have an effective execution of the function of real-time planning, intelligent hierarchical control methodology is introduced into use in construction of the control function. For realization of this introduction, a control executor system is developed with embedded single chip computer. Spindle power is applied to be constraint objective and federate override be hierarchical control interface with NC machine. The developed system realizes functions cooperation among autonomy of machine equipment, ACC control of machining process and optimization of manufacturing system. As a result, the presented model provides a solution for the automation problem island in CNC control of traditional machining process.
|
PDF Full Text Request