Multi-sensor Data Fusion and Estimation with Poor Information Based on Bootstrap-fuzzy Model

Multi-sensor data fusion and estimation with poor information is a common problem in the field of stress measurement. Small and distribution unknown data sample obtained from multi-sensor makes the data fusion and estimation much difficult. To solve this problem, a novel bootstrap-fuzzy model is developed. This model is different from the statistical methods and only needs a little data. At first, the limited stress multi-sensor measurement data is expanded by the bootstrap sampling. Secondly, the data fusion sequence is constructed by the bootstrap distribution. Finally the true value and the interval of the stress multi-sensor measurement data are estimated by the fuzzy subordinate functions. Experimental results show that the data fusion sequence is in a good agreement with the original measurement data. The accuracy of the estimated interval can reach 85%. Therefore, the effect of the proposed bootstrap-fuzzy model is validated.