Innovative quality process control via Cellular Neural Networks

The continuous detection and correction of unnatural process behaviours, owing to special causes of variations, is a basic task in manufacturing to keep any process stable and predictable, as well as to improve it continuously. In this paper, a contribution to increase quality process control from an innovative point of view is presented, which consists in designing Cellular Neural Networks (CNNs) to behave as associative memories for recognising unnatural behaviours. As an example, a test case is developed by considering abnormal cyclic behaviours given by sinusoidal signals. For this purpose, a CNN is synthesised for an associative memory, to recognise these unnatural situations. A robustness analysis of the synthesised network is then developed in the presence of input noise. The performances of the designed circuit have been illustrated in detail.