Research Of Forecasting The Reliability Of Information Sources

As the D-S evidence theory has a significant advantage in processing uncertain information, it has been widely used in the field of data fusion. However, D-S evidence theory can not effectively deal with the conflicting evidence, so the accuracy of fusion results is affected. Besides, with the increasing number of information sources in the process of data fusion, the fusion efficiency will be gradually reduced. In order to sovle the problems of evidence conflicts and low efficiency, an improved method based on forecast reliability coefficient is proposed.On the basis of researching the relational theories of D-S evidence theory and the exiting methods, an improved method of D-S evidence theory combination rules which is based on forecast reliability is proposed. This method can solve the evidence conflicts problem when there are large amounts of data need to be process, and can improve the fusion efficiency. First, set initial credibility of each information source and introduce the concept of credibility balance factor to forecast the credibility of information sources during the process of data fusion. On this basis, the calculation formula of forecast reliability coefficient is proposed. Then, get the formulaof balance factor by switching the calculation formula of forecast reliability coefficient; use a large number of sample data to train the balance factor until a stable value is obtained. Finally, put the stable balance factor into the calculation formula of forecast reliability coefficient, and integrate the new formula with the traditional D-S evidence theory combination rules. On the basis of this, a complete improved D-S evidence theory method based on forecast reliability coefficient is proposed.In the paper, Matlab is used to simulate the fusion of multi-source data, and the process that use the method proposed to sovle evidence conflicts and fusion efficiency problems is described in detail. In the simulation experiment, the results are analysed from two aspects including the accuracy and the efficiency of data fusion. And compare the results of the method proposed in this paper to the ones of the original method, the compare result shows the effectiveness of the improved method in solving the evidence conflicts and fusion efficiency problems.