Detection and identification of sensor errors e.g. for aircraft controller

摘要

The sensor produces a measurement vector m = H.x + e , where m is a vector of the measurement parameter, x is a vector of a state parameter being measured, H is the measurement matrix and e is the error vector. The order of m is greater than that of x. Validation vectors are determined as the column vectors of a projection matrix. A characteristic vector is determined as the linear combination of the validation vectors with corresp. elements of the measurement vector.The functional relationship between the characteristic vector and the corresp. classification vector is defined using a neural network whose connection wt. between its input and output layer is adapted during a training phase. The neural network contains only forward combinations between more than two layers of processor elements.