Development of data fusion algorithm for multisensored-systems

Data fusion is a method that combines data coming from multisensor. It provides a better analyzing and giving more succesfull decisions. The critical problem in the process of DF is not only comparing the data which come from sensors or practising on complex structures; but also analyzing the problem using complicated algorithms and parallel processors in order to reach a definite result. In a DF frame that found data having different features, in order to get accurate inference, hybrid structures in which several algorithms are used together should be created. In this thesis a DF algorithm is developed which is called ?Naïve Bayes Classification Aided Neuro-Fuzzy Algorithm (NBNF)? that will make a powerful inference about environment by combining data collected from multisensor. A sample application has been designed in order to predict forest fires beforehand and detect the fire which has just started and NBNF is used in this application. Moreover, in this thesis in order to try accurance and effectiveness of NBNF, the data combining process can bealso made by using on Artificial Neural Network (ANN) in the application. According to the results it is observed that NBNF is more accurate and reliable than unhybrid ANN algorithm because of its hybrid structure.