Design and Modelling

In system level decision making applications, computer simulation was utilized in combination along with neural network to generate data utilized by neural network to implicitly model such system. In cellular manufacturing applications, neural networks utilized to classify part and machines allow easy recognition of machine cells, part families and exceptional elements. Innovative parts could be allocated to existing classifications via the employ of neural networks. Feed-forward neural networks trained utilizing the back propagation or BP algorithm is admired for this application. Other kinds of neural networks comprise ART or Adaptive Response Theory networks, SOM or Self-Organizing Maps and Hopfield networks neural networks. Limitation of neural networks for design and modelling of manufacturing systems conclude from neural network themselves as several parameters are needed to be found on a trial and error basis. Additionally, neural networks do not take guarantee an optimal solution, and are frequently applied along with several other searches to enhance the quality of the solution. Conversely, the high computational efficiency as like offered by the neural network composes them a promising designing choice.

The capability of neural networks to learn from inherent parallel processing and experience, the approach based upon them permits for understood modeling of complex systems utilizing representative data, hence eliminating the require for modelling and explicit mathematical analysis. Among other abilities of neural network are their capabilities to solve problems along with incomplete or noisy data and keep uninfluenced by the problems of large dimensions serve as global computing is completed in parallel and local computation in all neurons is very easy. Neural networks, currently days, are being utilized for resolving large industrial problems. Along with emerging offered neuro-computing devices, neural networks are becoming more helpful for tackling large scale manufacturing design and modeling applications.