Research of UAV fault and state forecast technology based on particle filter

This paper presents an UAV fault and state prediction approach which is based on particle filter. In the UAV system, on account of its dynamic environment, mechanical complexity and other factors, it is difficult to avoid all potential faults. So, in order to early detect the potential fault, fault forecast is necessary so as to avoid enormous losses. As the input and output response model of UAV system is nonlinear and multi-parameters, it is need to find an appropriate way to of fault prediction for system maintenance and real-time command. Particle filters are sequential Monte Carlo methods based on point mass (or `particle´) representations of probability densities, which can be applied to any state-space model. Their ability to deal with nonlinear and non-Gaussian statistics makes them suitable for application to the UAV fault prediction. UAV is an extremely complex system, two important aspects of monitoring are focused on this paper: 1) Engine condition monitoring and fault prediction; 2) UAV flight track forecast. The experimental result indicates the effectiveness of this approach.