An algorithm for measurement data eliminating gross error and smoothing

For signals of engineering testing, various reasons lead to grocess errors in measurement data, these errors must be eliminated. Based on PaйTa Ctiterion, a new algorithm, self-adapting PaйTa Ctiterion algorithm using shifting windows is proposed. The algorithm can both eliminating gross errors and smoothing data statistically. Besides, the concept of step length, window width and order are proposed. This algorithm is simple, quick, takes less time when processing data, and is suitable to be used under high speed, real time condition. This method has yielded desirable results when it is applied to process the penetration overload signal in the tests of projectile penetrating concrete targets of semi-infinite thickness and projectile penetrating three-layer concrete targets of finite thickness. The theory and the practice result prove that the proposed algorithm is feasible when eliminates gross error and smoothes data, and effective as a algorithm for short-time and fast real-time processing. It meet the engineering needs and can be generalized as a algorithm for fast and real time processing.