Smart air-conditioning system using multilayer perceptron neural network with a modular approach

The project involved the design of a smart air-conditioning system to control the temperature in an air-conditioned bus more effectively. The system comprised a modular network using a multilayer perceptron. A backpropagation algorithm with momentum term was used to train the network. There are 4 neural network modules in the whole system. The input parameters for training were broken down into 3 categories which are temperature, number of passengers and time of the day. A final data fusion module was used for decision making to set the system to an optimum state of operations such that passengers will feel comfortable. Test results obtained were encouraging with fast training time.