Neural Network Modelling

In many situations, the physical behaviour, or the characteristics, of a system are dependent upon many parameters, and relationships are often non-linear. In these circumstances, attempting to identify the relationships using classical mathematical techniques would not be a practicable approach.

However, by using Neural Network Modelling, the outputs can be predicted from prescribed inputs, enabling:

Sensitivity of the behaviour or characteristics of the system to each parameter   Relationship between all of the parameters and the physical behaviour or the characteristics of the system

Once the relationship has been identified, this can be included in Software, and a predictive tool generated.

The potential for the application for Neural Networks in engineering is enormous, ranging from the characterisation of material properties, which are dependent upon process parameters and composition, through to the behaviour of a design in service which is influenced by many environmental conditions.

Examples of the where Neural Network Modelling has been applied by Eatec include:

Prediction of loads acting on an underwater plough   Prediction of cutter forces during machining   Prediction of vibration levels for rotating equipment

Eatec Limited Address: 3 Armstrong Court, Armstrong Way, Yate, Bristol. BS37 5NG. England. E-mail: info@eatec.co.uk ©Copyright - Eatec 2001 Tel: + 44 (0) 1454 332240 Fax: + 44 (0) 1454 332249 Website designed by Web Productions