Fuel volume measurement in aircraft using neural networks

Measurement of fuel quantity in aircraft tanks is a multidimensional estimation problem. It involves a nonlinear transformation of a set of noisy sensor signals into a single fuel quantity estimate. In a typical passenger aircraft, this calculation is usually performed by means of a set of one-dimensional, linearly interpolated look-up tables. This paper presents a neural net approach to the problem. A feedforward neural net is trained to estimate fuel quantity directly from sensor readings. A simulation example is given to compare efficiency of the neural net technique to the standard look-up table method. Practical ramifications of the proposed method are discussed.