Research And Application Of Intelligent Assembly Technology For Cabin Inner Structure

The rapid development of digital, intelligent computing, virtual reality, and machine vision technologies provide new technologies and tools for improving design, production and assembly of the cabin products. This thesis builds an intelligent assembly system by the use of computational intelligence, human intelligence and machine intelligence to solve the assembly problems of cabin inner structure. Assembly sequence of the inner parts is programmed and optimized by ant colony intelligence algorithm which simplifies the difficulty of the assembly planning of the complex structure. The augmented interactive assembly project based on the mixed reality technology improves the user's awareness in the aspects of vision, information and operation. Moreover, the combination of human-computer intelligence carries out virtual verification for the assembly process of the module. Method of visual assembly quality inspection provides an effective guidance for the inspectors and help them find remainders quickly, reducing the chance of missing reminders. An assembly quality model is proposed on the basis of Markov Chain to realize quantitative assessment of cabin assembly quality.A method about information acquisition and reuse is introduced for the rapid construction of virtual assembly environment. Firstly, the augmented information model for the virtual assembly system is established to realize the connection between the information sources and virtual assembly system. According to the different sources, information can be obtained through CAD system, user input, and assembly planning module. Then the corresponding application strategy of differently obtained information to the virtual assembly system is provided. The whole assembly scene is managed as a scenario tree structure, with which the rapid construction of virtual assembly environment and the data support for the virtual assembly system are realized.Bionic algorithms and virtual reality is combined in this thesis to solve the problem of assembly planning, from which feasible assembly sequence is obtained through the ant colony algorithm, and assembly path planning is competed by human-computer interaction in a virtual reality. On the basis of assembly sequence planning method, a bionic paradigm of assembly sequence planning is proposed. Aiming at the problem of cabin inner structure assembly sequence planning, a solution of creating implicitly solving space by assembly complete graph model is put forward, and then an ant colony algorithm is constructed to search the best optical assembly sequence of cabin structure. Moreover, coupled with the assembly characteristics of cabin inner structure, an assembly path planning method is proposed via human-computer interaction in virtual assembly environment.A novel augmented interaction technology based on the ideology of uniting mixed reality, augmented information and constraining proxy is presented, which can improve interacting efficiency of virtual assembly environment in terms of vision, information and operation. Technology for uniting virtual and reality inherits all interaction of virtual reality, and it also provides additional scenes of the real assembly to enrich the perception of assembly process at the same time. Augmented information provides users with assembly instruction such as detailed description and features. Constraint proxy and its match rules ensure visual restriction relationship for accessories reconstruction, and avoid complex calculation of accessories constraint's match. Augmented human-computer interaction technology provides a convenient interaction method for the application of virtual assembly.Machine visual technology is used for assembly quality inspection, which detects assembly redundant objects and measures gap between the parts with the saved panoramic picture and video of collected information for the cabin inner module. DirectShow technology is used to capture the video from camera and restore the distorted image caused by lens distortion. The image data are analyzed by sub-pixel technology and the distances between parts are measured by the method of edge position of gray moment. Then, the corner points of the counter are extracted, and he Hausdorff distance is improved as the matching criterion to find the target region by the counter model of the redundant objects.According to the assembly quality requirements of cabin, a three-level structural evaluation system of cabin is proposed by adopting the analytic hierarchy process. After discussing the factors influencing the assembly quality of cabin, an analyzing model of assembly quality is proposed on the basis of Markov chain. The transfer matrix of cabin assembly process state is built, which provides the basis for quantitative evaluation of the cabin assembly quality. Through the calculation of the assessed value of the underlying index by the quality loss function, an integrated appraisal of cabin assembly quality is obtained by synthetically sum of weight, and thus the assembly quality of cabin can be comprehensively evaluated.