This thesis is concerned with dynamic and transient analysis of MSF desalination plants. The technique is developed using artificial neural networks (ANN) approach for the purpose of prediction, analysis, modelling, and control of MSF desalination plant. The applicability of the method to predict an approximation of the transient operating conditions as well as the control action are shown satisfactory. The network architecture and learning algorithm are developed based on the Multilayered Feed forward Networks (MFN) with the Back Propagation (BP) learning algorithm. It was shown that the approach could intelligently capture the dynamics of the system. An improved technique is developed for the BP learning algorithm based on Global Error Node Evaluation (GENE) approach for MFN to retains the function approximation requirements for a nonlinear dynamic behaviour. However, by using this approach considerable improvement for the generalization capability could be obtained for the case study under consideration. The technique provides the necessary dynamic learning, behaviour required for MFN. This approach appears to be effective for the input - output dynamic modelling of complex process systems and therefore on-line adaptation is possible (when the characteristic of the system is changing or when more test data are available for another operating range). The developed algorithm is used for the development and validation of an empirical multi-controller structure for MSF desalination plant. Satisfactory results are obtained from practical examples with the additional training ability.