Identification of a change in the thermal dynamics of a wall

In this work the problem of the estimation of achange in the dynamics and time-varyingthermal parameters for a wall from experimentaldata is solved in several ways. The problem isfirstly solved by the application of a classicalsteady-state method. This method gives goodresults with time series long enough, but showsno predictive power. Secondly, linear timeinvariant(LTI) dynamic statistical models intransfer function form are applied. Theyestimate the thermal parameters and also offerpredictive power. Finally, linear and timevarying(LTV) continuous time stochasticmodelling is applied. This last technique is ableto estimate the thermal parameters, fits the timevaryingdynamics and shows an accuratepredictive power. The first two methods needsome additional previous treatment of the timeseries by the analyst, while the last one doesn't.The experimental data for the wall correspondto the first case proposed for resolution in theSystem Identification Competition I11 (SIC 111,www.dynastee.info), organized by theDYNASTEE (DYNamic Analysis, Simulationand Testing applied to the Energy andEnvironmental performance of buildings)network. The impleimentation of the EuropeanEnergy Perforniance of Building Directiverequires adequate calculation and modellingtools which is the main reason why this thirdcompetition has been organised. In this context,SIC 111 has been organised with the objective offurther develop knowledge of systemidentification applied to thermal performanceassessment in the built environment.