Kalman Filter Development for Real Time Proper Orthogonal Decomposition Disc Temperature Model

Extending disc life through online health monitoring has been a proven method of minimising engine downtime and maintenance costs. To properly monitor the disc requires a robust model of the disc’s non-linear thermal dynamics. A model can be improved by filtering the output using a measurement of the disc in real time. The damage models can then be computed with higher statistical confidence leading to increased safe life prediction. Recently, a model of disc temperature has been developed based on the proper orthogonal decomposition of simulated data. The model produced detailed thermal gradients for use in damage calculation and life assessment. This paper presents the development and implementation of a Kalman filter to augment that model. The location of the measurement has been assessed in order to select the most appropriate target for instrumentation. Points all around the front and back of the disc have been assessed, and the best practice result is found to be near the centre of the disc neck. Matching temperatures at this point represents a compromise between the fast dynamic response of the rim, with the slower response of the cob. The new model has been validated against an independent flight simulation that had previously been excluded from any training process. The addition of the Kalman filter allows the model to match aircraft dynamics outside the regular training trajectories. The accuracy is approximately ±30K, and there is a root-mean-square error of only 2K over the whole model at any one point in time.