In order to effectively improve the identification ability for fault data of wireless sensor network, a new mining algorithm FDMA (Fault Data Mining Algorithm) is proposed by artificial bee colony. In this algorithm, the burst of fault data is reduced to be standardization with wavelet transform, and the distribution range is divided by correlation coefficient. Then, the objective function is built to mining fault data, and it is optimized with artificial bee colony. Finally, a simulation with actual sensors sample data was conducted to study the performance between FDMA and other algorithm, such as throughput, time delay, packet dropping rate and energy consumption. The results show that, FDMA has better adaptability.
