Adaptive Fuzzy Kalman Filtering: Applications to General Aviation (GA) Flight Data

In the engineering field measurement systems that can provide the most accurate results are increasingly required to achive optimal control and monitoring of systems. The most efficient and optimal filter widely used for this purpose is the Kalman filter, a recursive algorithm that provides an optimal estimate of the state vector for a dynamic and stochastic linear system, starting from a series of measurements perturbed by noise and observed over time. It has become a universal tool also in the field of integration of different sensors (Multisensor data fusion). However, the Kalman filter requires that all the plant dynamics (the system model) and statistical values of noise for both process and measurement noises are exactly known and for pratical applications the knowledge of these informations can be problematic.An adaptive estimation formulation of the Kalman filter will result in a better performance or will prevent filter divergence.. This filter adaptation has been relized by many authors through a Fuzzy Inference System (FIS), an inference process based on a method of decision making that greatly simplifies the approach to the adaptation. In this paper the application of an adaptive Fuzzy Kalman filter for data smoothing is explored. This approach avoids the need to know the system model and inputs, obtaining greater versatility and simplicity in pratical applications. Experimental results of our fuzzzy Kalman smoothing technique, implemented in Matlab and applied on datasets acquired from typical GA (General Aviation) avionics sensors (Garmin G1000), show the reliability and efficiency of this approach.