Hybrid neuro-fuzzy system application to inferential sensing

As the neuro-fuzzy system is based on a feed forward network structure, it cannot effectively cope with dynamic processes such as space heating systems for the built environment. To overcome this problem, an improved version of the hybrid system is developed and presented in this paper. This system has self-feedback loops for the output that can model the dynamical behavior of the process. The developed model is used in a case study for the estimation of average air temperature in the buildings served by a forced air space heating systems. The results show the effectiveness of the self-feedback system in terms of minimization of root mean squared error.