Use this URL to cite or link to this record in EThOS:
Title: Improving the cost model development process using neural networks
Author: Wang, Qing
ISNI:       0000 0001 3560 9105
Awarding Body: De Montfort University
Current Institution: De Montfort University
Date of Award: 2000
Availability of Full Text:
Access from EThOS:
Access from Institution:
In order to achieve success in export markets through provision of high levels of product choice, manufacturing industry will need to develop and economically use many new materials and manqfýcturing processes. To support this development, it is expected that the quantity, type, accuracy and complexity of cost models will need to be greatly increased. It is essential under these changing conditions that the process of developing cost models is able to remain responsive to user requirements and effective in terms of the resources required to generate models. This research investigates existing methods of establishing 'cost estimating relationships'and identifies their relative benefits and limitations in terms of their effects on the overall cost model development process. The basic tasks involved in the cost model development process and the basic characteristics of cost models have been identified and used to evaluate the use of Artificial Neural Networks (ANNs) as alternative methods of establishing cost models. The problem of identifying suitable ANN structures has been resolved by the use of the Taguchi Methodology. Experiments to identify the influence of varying the number of layers and number ofprocessing elements per layer within an ANN structure have shown that in general, increasing the number of processing elements per layer and decreasing the number of layers leads to increased estimating accuracy. Experiments to examine the effects of varying the amount of data used to develop the model and varying the number of variables within the model have indicated that substantial benefits, in terms of simplifying data identification and collection tasks can be realised when compared with regression analysis.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available