The establishment and reasoning of knowledge base system for mechanism kinematic schemes

Abstract

As one of the key problems of computer-aided conceptual design (CACD) for mechanism kinematic schemes, the effective establishment and management of knowledge base systems of kinematic behaviours and mechanisms is still troubled by information redundancy and unreasonable structure at present. Retrieval efficiency and solving intelligence of knowledge base reasoning systems on the basis of the knowledge classification method mentioned above have many unsatisfactory factors. So in this paper, firstly, data standardization technology and combined classification methods have been applied to carry out the classification of kinematic behaviours and mechanisms in the mapping field between the kinematic behaviour level and the mechanism level of conceptual design. Also, the principle of computer coding and storing have been built to give a fast and broad selection of mechanisms that meets the requirements of basic motion characters. Then on the basis of what was mentioned above, the heuristic matching propagation principle (HMPP) of kinematic behaviours and its truth table serve as a guide to perform mechanism type selections. The knowledge base automatic reasoning system of mechanism kinematic schemes based on HMPP is developed. By means of depth first search strategy, the motion requirements are matched against objects in the knowledge base along the orientation of function unit-mechanism ID→function unit ID→kinematic behaviour ID(output/input)→motion type coding/ continuity coding/linearity coding/directionality coding at four different levels of abstraction. Top-down design process in this paper conforms to the law of innovative thinking for designers. If a direct match in a higher level was not discovered, then the design course will be transformed into the match at a low level automatically. According to the truth table of HMPP, once non-zero elements in kinematic behaviour ID (“1” or “2”) appear in the matching process, they will propagate until the solving process is over, and become one of the coding of the feasible solutions. The heuristic nature of HMPP presents a top-down solving method to decompose the complex requirements into several simpler ones. Different from traditional solving methods, the method presented in our research work has two advantages. Firstly it is easier to be operated and realized with a computer. Secondly, it is a creative process by nature. Finally, an application is given to indicate its practicability and effectiveness.