Use this URL to cite or link to this record in EThOS:
Title: An approach for the causal analysis in casting processes based on probabilistic analysis, neural network and design optimisation
Author: Ransing, R. S.
Awarding Body: University College of Swansea
Current Institution: Swansea University
Date of Award: 1995
Availability of Full Text:
Access from EThOS:
For the causal analysis in a manufacturing process, many statistical techniques are available to assist an engineer in the decision making. Later, expert systems techniques were also proposed to attempt some routine decision making problems. However, these systems faced a criticism that they do not generate any new information i.e. the user only gets that information which is already encoded in the database and rulebase. The algorithms proposed in this thesis are expected to generate new information even for an expert user. A two pronged strategy has been adopted for the causal analysis. Analysing the influence of causes on the occurrence of defects by recognising patterns in the rejection data and also using the previous diagnostic examples. The program learns from examples as well a mistakes and at the same time explains or justifies its results. In the second strategy, the causal relationship is analysed by the numerical simulation of the process. If a defect is predicted, the algorithm will automatically redesign the process or die until the casting is simulated as defect free. A new scheme has been proposed for the causal relationship. Causes are categorised into two groups viz. Rootcauses and Metacauses. Rootcauses are the actual process, design and material parameters which can be directly controlled. Metacauses denote the scientific rationale which relate rootcauses to defects. Representation of the causal knowledge in such a network form opened two research directions for the diagnostic analysis. The first one based on the probabilistic analysis and the second one on neural network. In both cases the network was constrained to the defect-metacause-rootcause relationship and necessary modifications were done in the conventional approaches.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available