Optimisation of a tomography sensor for imaging of temperature in a gas turbine engine

We present a numerical study of tomographic reconstruction of gas temperature distributions from synthetic data of laser absorption measurements through the imaging plane. A mathematical expression which predicts the reconstruction error as a function of the beam configuration is developed, and a global optimization scheme is employed to minimize this predicted error subject to a constraint on the maximum number of permitted beams. This optimization methodology is applied to an annular measurement region and compared to conventional regular fan-beam arrangements in temperature tomography simulations. Synthetic data generated over the optimized beams led to temperature reconstructions with superior accuracy, provided Tikhonov inversion was used as the method of reconstruction.