Based on reliability design theory of mechanical system and robust design method, the reliability-based robust design of mechanical system with correlated failure modes is extensively discussed and a numerical method for reliability-based robust design of mechanical system with correlated failure modes is proposed. The reliability sensitivity is integrated with the reliability-based optimization design model and the reliability-based robust design of mechanical system is described as a multi-objection optimization. On the condition of knowing first two moments of basic random variables, the respective program can be used to obtain the reliability-based robust design information of mechanical system accurately and quickly．

By applying reliability theory of mechanism kinematics accuracy (MKA) in design of main motion mechanism of internal combustion engine (ICE), analysis on the reliability of main motion MKA of ICE was carried out. Accuracy error expression was derived by the first order Taylor expansion, and then the limit state function expression according with engineering practice was established combining with reliability theory. In the case of given characteristics of random parameters distribution, analysis on the reliability of main motion MKA of ICE by computer program was done, which could be utilized to obtain reliability information of planar linkage machine accurately and quickly. The results demonstrate that the proposed method is effective and practical, and has important theoretical value and practical significance．

Implementation of transforming membership degrees is the kernel computation in multi-level fuzzy evaluation of mechanism selection; but the existing algorithms contain redundancy and the redundant parts of the index membership degree, which are useless in objective classification, are involved in calculation of objective membership. Therefore, by using the entropy-based data-mining method, through mining the knowledge information hiding in each index membership degree about the objective classification, index-distinguishing weight was defined. By using this weight, the redundant values in index membership degree but useless to objective classification were eliminated and the effective values were extracted. The effective ones were transformed into comparable values through importance weights, which were utilized to calculate objective membership degree to realize membership degree transformation. Based on the above, the improved fuzzy evaluation model of mechanism selection was established.

After analyzing the disadvantages of the existing evaluation methods of maintainability design, an evaluation method integrating quality function deployment(QFD) and D-S based on requirement analysis was proposed. The method combined the advantages of QFD and D-S in dealing with uncertainty and could closely reflect the military demand. The total process was represented by diagrams, which had some engineering practicability．

Rectangular optimal packing problem is commonly faced by the production practice in manufacturing field. This paper intends to solve the problem by utilizing the lowest horizontal search algorithm. First, the paper analysis the defects of the original lowest horizontal search algorithm, and then design an evaluation function for it. During the packing procedure, the one with highest value will be chosen and packed in the current position after all the items to be packed being evaluated by the function, which overcomes the randomness of the search process and optimizes the search direction. The experimental results show that the improved algorithm gives better placement pattern. In addition, the larger the problem is, the better the algorithm will perform. Therefore, the improved algorithm is suitable for solving large scale packing problems.

The application of genetic algorithm in construction with steel structure not only changes the traditional design concept, but also enriches the theories and methods of steel structural optimization. A series of improvement of fundamental genetic algorithm was proposed based on compare, analysis and integration of the existing improving methods, and finally an effective and comprehensive genetic algorithm was developed. The improved algorithm adopted the thought of combining the optimal saving strategy and the rate choice, and the crossover probability relevant to the evolution algebra and the mutation probability related to individual adaptability was designed. This algorithm has been applied to the optimization design which takes the least weight as an objective in construction with steel structure and the result shows that it achieves satisfactory effect and effectively increases calculation speed．

Based on solid model of double wishbone independent suspension with twin torsion bar, the finite element model was established, and simulation of connection relation among parts was carried out. Then finite element analysis on suspension structure was conducted and the reason which caused crack of longitudinal beam on vehicle frame was obtained and correlative structural optimization of the suspension structure was done. The results show that it overcomes the problem of crack of longitudinal beam and has been applied in real vehicle, which demonstrates that it has practical value in engineering．

Adhesion is the most common failure in micro-machine, and general micro-machines all try to avoid it. Surface tension is the most crucial reason causing adhesion. Deep research on physical principles of adhesion under surface tension was carried out. By taking polycrystalline silicon as an instance and considering its adhesion under surface tension, structure parameter design formula of anti-adhesion polycrystalline silicon micro-gear under surface tension was derived. The results show that the generation of adhesion of micro-gear has relation with material characteristics (elasticity, surface properties), micro-gear's structural parameters (radius, tooth width) and the gap between micro-gear and basic body．

Research on accurately fitting method of spatial circle is carried out. Linear and nonlinear least square fitting are common methods for fitting regular curve and surface. As a regular quadric curve, spatial circle has no specific curve formula and therefore linear and nonlinear least square methods cannot be directly applied to it. Due to the fact that spatial circle can be regarded as the intersection line of a plane and a spherical surface, spatial circle eigenvalue can be solved by eigenvalues of the plane and the spherical surface. The two-step spatial circle fitting method based on projection is proposed. The spatial circle fitting principle is presented, and the correctness, feasibility and accuracy are validated. The method is implemented by programming. The method has superiority in accuracy comparing with Beziert and B-spline curves fitting methods, and it can improve spatial circle fitting accuracy in reverse engineering．

To overcome the shortcomings in model accuracy of the existent NN training algorithms, a new algorithm to train feed-forward networks, i.e. GA-BP Bayesian algorithm, was proposed, and it was special for establishing approximate model. This algorithm was developed to improve the generalization of neural networks, and its objective is to obtain the weights corresponding to the maximum posterior probability. And it adopts both genetic algorithm and L-M (Levenberg-Marquardt) back-propagation (BP) to search optimal weights. L-M BP is the most popular algorithm to train feed-forward networks. Combined with typical computing instance, GA-BP Bayesian algorithm was compared with L-M BP and the result indicated that the approximate model established based on it has higher and more stable accuracy.

High precision measurement of dynamic rotational speed is the key problem for high performance motion control systems, especially for the low speed cases. For this problem, a novel real time estimation method of rotation speed was proposed. A disturbance observer was adopted to estimate the equivalent external disturbance and modeling error, and its output was introduced to the rotation speed observer as an input, which guaranteed convergence of estimation and enhanced the robustness. Finally simulation and experiments were conducted for a DC motor servo system. The results show this method is effective and robust, and it is insensitive to quantization noise．

Vehicle weighing module is the core part of vehicle weighing system. The working principle, flow and implementing scheme of vehicle weighing module were presented. 3D model of the vehicle frame was established by Pro/Engineer. In ANSYS, necessary simplification of this model was done and load was applied to the frame according to the actual situations; then the strain sensitive point was found and the relation between strain and load was obtained. The algorithm of weighing was set up based on support vector machines (SVM) and its feasibility and correctness in different conditions was verified by using cross-validation method.

he bending vibrations in a very wide textile machine frame can be reduced by filling both walls with cast mineral. Results of FEM calculations and measurements will be presented.

To solve the problem that positive string will fall down when cannon tube rifle becomes wider between the beginning segment and the changing tangle segment by using fixed cathode of ECM to machine big caliber and deep string groove cannon tube, the method by separating work teeth as discrete component and regarding it as a structure that connects but exists relative movement with cathode was adopted. Solid geometric model of cathode was established and work teeth, flow-guiding system and driving system were designed. When cathode was hauled by pull rod to axial feeding and radial direction, micro-motor embedded in cathode will drive independent work teeth according to program simultaneously. According to the change in trajectory equation, the work teeth keep consistent with the tangle. This method can realize highaccurately machining of cathode teeth．

The final settlement deformation of soft soil foundation has great impact on design safety and normal usage of constructing project such as highway, railway, airport and port terminal. In light of the settlement prediction result of grey model (GM), the selected value of GM directly determines the correctness of prediction result. Therefore how to select value and the rule to select value is the core of accuracy of GM. Based on optimized GM(1,1), grey-visualization software for settlement prediction was developed by VB. Through analysis on accuracy of different time intervals as well as compare between prediction result and actual inspected data, it is demonstrated that this method can predict settlement of soft soil foundation more accurately and has practical engineering value．

In order to evaluate the risk of structural design of construction engineering， technique accidents or problems caused by negligence or mistakes as well as other kinds of subjective or objective reasons during construction engineering structure design were comprehensively analyzed by the methods of Delphi and statistic data. AHP was adopted to evaluate the arisen technique accidents or problems; and the importance and rank of possible technique accidents or problems during structure design were determined， which had great significance to effectively prevent engineering structure design risk.