Research On Intelligent Decision And Process Reuse For Process Design Of Machining

Process design of machining is the transition from design phase to manufacture stage,which is the key to the output of design achievement.At present,how to improve the quality and efficiency of process design has become one of the major bottleneck problems.In the current context of manufacturing upgrading its intelligence,the process design capability not matching the intelligent manufacturing capacity has become one of the major restrained factors of the promotion of the company manufacturing level.Although many enterprises have been applied CAPP(Computer Aided Process Planning)systems,these systems is basically about some predefined simple rules in process decision and need a lot of man-machine interactive operations,thus reducing the validity and automatic level of process decision.Meanwhile the existing process planning systems lack of excavation and recycling of a large number of process cases accumulated by enterprises.This can cause more repeated design,which reduces the efficiency of process planning.The process planning of mechanical products is taken as the research object,some technologies of intelligent decision and process reuse are researched.By improving the mode of process planning,this research is conducive to shortening the development period of the products and promoting the enterprises' ability of rapidly responding to market changes.It also plays an important role in realizing the intelligent join between process planning stage and manufacturing system and driving the production mode innovation based on intelligent manufacturing paradigm.The dissertation is supported by the National Natural Science Foundation of China(project number: 51375395,51475381),and the technologies of intelligent decision and process reuse are researched.The key contents include the following four aspects: decision making of feature process chain,decision making of process route,multidimensional similarity measure of process route and discovery and reuse of typical process.The main contributions and innovations in the dissertation are as follows:The factors affecting selection of a feature process chain are explored,and a fuzzy comprehensive evaluation method of feature operation chain selection using hole feature as an example is developed,thus realizing the selection of possible feature operation chains based on processing requirement constraints.An enterprise processing network model is built by applying complex network theory,and then a manufacturing resource instability index based on the network model is established.The final operation chain is decided by calculating the manufacturing resource instability index of possible feature operation chains.When the feature operation chains are determined,A generating mechanism of potential process routes is presented.The similarity between potential process routes and typical process route is analyzed by using Needleman-Wunsch algorithm,and a process reasonable index is established based on matching rate of typical process.The potential process route's replacement rate of manufacturing resources is calculated,and then a process economical index is given.Considering the above two indexes,a multifactor decision model based on Set Pair Analysis is discussed to select the final process route from potential process routes.A method for measuring the similarity between two process paths based on Needleman-Wunsch algorithm is presented,which realize calculating the similarity of process paths defined by machined process routes.Geometry change that corresponds to machining operation is extracted and represented,and then geometry variation sequence that corresponds to process route is established.An improved Levenshtein distance(Ld)is used to calculate the similarity of three dimensional geometry entities,and KuhnMunkres algorithm is adopted to calculate the similarity of geometry changes.On this basis,learning from the idea of Blast algorithm “fragments retrieval replaces whole retrieval”,the similarity of geometry variation sequences can be calculated.A multidimensional similarity measure method,blending with the similarity of process path and the imilarity of geometry variation sequence,is proposed.A typical process route discovery method based on intelligent clustering is presented,and reusing ways of the typical process route are discussed.An intelligent clustering model based on ant colony algorithm is built on the basis of calculating the multidimensional similarity of process route.Then every typical process route in each clustering cluster is extracted by using maximum average similarity.Three reusing ways of the typical process route are given as follows: direct revision method,indirect matching method and case reasoning method.