| This paper introduces the theory of SVM into information fusion, does some research on its application, and studies monitoring data of the daily wastewater treatment plant. Using SVM in information fusion can greatly increase the computing speed of information fusion and accuracy of the fusion results. Accuracy has a direct impact on accuracy of the final decision-making. With SVM's application in information fusion, enormous economic and social benefit will come into being, in the military and civilian fields.This paper introduces the relevant knowledge on information fusion and SVM, which have been focused and studied in recent years, the relevant knowledge on information fusion, the principle and conventional methods of fusion. This paper discusses SVM classification algorithm, deeply studies the sequential minimal optimization algorithm, this paper makes analyses on least squares support vector machines algorithm, and proposes the improvement ideas of sparsity increase, for its lack of sparse of results.For more SVM classification algorithm, this paper firstly introduces the popular "one-against-the-rest" and "One-against-one" algorithms, and then puts forward the binary encoding multi-classification algorithm. In the daily monitoring data test of wastewater treatment plants, firstly, we make use of the method of neural network, which is often uesed in information fusion, and then, use three algorithms such as sequential minimal optimization algorithm, least squares support vector machines algorithm and the least squares support vector machines algorithm improved. In SVM, we use two methods such as "one-against-the-rest" and the binary encoding. Through a large number of experiments, the parameters of SVM are determined. At last, it is successful to bring SVM into information fusion, by the analyses on experimental results. In a word, the method of SVM is higher than neural network method, not only in training speed, but also in the right rate of the type of test.
|
PDF Full Text Request