Loading...

Tolerance Design of Mechanical Systems based on Reliability Modeling under Bayesian Inference

Ghaderi, Aref | 2018

714 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 51253 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Khodaygan, Saeed; Assempour, Ahmad
  7. Abstract:
  8. Mechanical production centers are seeking to produce the highest quality products at the lowest cost. Intrinsic processes of manufacturing are not precise processes, and factors such as tool wear, tool vibration and fixation, fixing defects, and other factors that occur during production, make the pieces deviate from the designer's desirable geometry. Usually, due to deviations of parts from their size, their dimensional and geometric characteristics change. Because the components are rarely just a part, often in the majority of parts of the assembly, the operation of the set may be impaired due to the accumulation of changes. Errors that are usually caused during component assembly due to dimensional and geometric changes of the parts can create a lot of problems, including loose collection, additional operations to correct errors, increase production costs and warranty. Ultimately, the customer's dissatisfaction will be sought. Tolerance design is an applied process consisting of two main stages of analysis and allocation. The product's tolerance, which the designer uses to measure the dimensional and geometric distortion created in the construction of the parts on the overall function. The assembly system estimates and commensurate with the performance of the set and the cost of producing dimensional and geometric product tolerances. Tolerance design plays a major role in industrial production as a key tool in improving quality, reducing costs and increasing profits from product manufacturing. The main purpose of this research is to examine the analysis and allocation of tolerance using statistical viewpoint and reliability using Bayesian statistics and inference. In this research, first, by presenting a tolerance analysis algorithm, firstly, using a given data, a probabilistic model is created for the modeling of tolerance using the Bayesian regression. Then, based on first order reliability and Monte Carlo analysis methods, the amount Reliability of the product in order to meet the functional and qualitative requirements of the qualitative control phase. Further, if the reliability of the product is not desirable, new optimal tolerances are allocated according to the optimal tolerance of the optimal design of the product. To achieve this goal, a new framework is formulated in the form of a multi-objective optimization problem using the concept of sensitivity vector, minimizing the cost of production and enhancing the level of product reliability. An NSGA-II multi-objective algorithm is used to extract the optimal set of data in the form of pareto front graphs. Then, to select the best optimal tolerances, the optimal answer set in the Pareto Front Figures is utilized by the upgraded TOPSIS method. In order to validate the proposed method, two examples of one-way clutch, which have a nonlinear relationship and a low number of parameters, and a transmission system that has a linear relationship and a large number of parameters, are studied and the results of the method are compared with the results of classical methods
  9. Keywords:
  10. Tolerance Analysis ; Tolerance Allocation ; Reliability ; Statistical Methods ; Optimization ; Monte Carlo Method ; Genetic Algorithm ; Bayesian Inference

 Digital Object List

 Bookmark

No TOC