Immune Feedforward Neural Network for Fault Detection

An immune algorithm-based approach was developed to optimize a feedforward neural network. The network architecture, activation functions, and training method were encoded as individuals with an appropriate method for individual selection. The immune feedforward neural network is then applied to fault detection of water quality monitoring equipment. This gives better performance than a feedforward neural network.