AM INPUT ESTIMATION APPROACH TO ONLINE 2-DIMENSIONAL INVERSE HEAT-CONDUCTION PROBLEMS

An on-line methodology to solve two-dimensional inverse heat conductio n problems (IHCP) is presented. A new input estimation approach based on the Kalman filtering technique is developed to estimate the two sep arate unknown heat flux inputs on the two boundaries in real time. A r ecursive relation between the observed value of the residual sequence with unknown heat flux and the theoretical residual sequence of the Ka lman filter that assumes known heat flux is formulated. A real-time le ast-squares algorithm is derived that uses the residual innovation seq uence to compute the magnitude of heat flux. This recursive approach f acilitates practical implementation, and its capabilities are demonstr ated in several typical cases with discontinuous and time-varying heat flux inputs.