This paper introduces nonlinear approaches which include neural networks and ANFIS to identify and control heat transfer within a chamber. Initially, traditional linear models are obtained using transfer functions with delays through Matlab identification tools. However, this traditional linear model failed to faithfully represent the system when the input was changed. This outcome was expected since linear models are reliable only within specific operational ranges. To create a novel model that is applicable across the entire state space, two alternative identification methods, utilizing neural networks and an adaptive neuro-fuzzy inference system were introduced. After testing them with input data not used during the training, the models were compared and all of them showed satisfiying results. In the continuation of the research, control techniques based on these techniques were presented. After assigning an arbitrary temperature as a reference signal, inverse models were made and four controllers in direct inverse control sheme were compared: three feedforward neural networks with different numbers of neurons in the hidden layer and the adaptive neuro-fuzzy inference controller. The results and possible improvements are discussed in the conclusion.