Stochastic Search Methods Used for Parameter Estimation of Thermal Properties

Abstract

There are many estimation techniques, which are used in Thermal Response Test (TRT) data analysis. The commonly used models, Source Model, Cylindrical Source Model, numerical models do not take into account the nonlinear system effects like for example the phase change. The present work suggests the use of the input/output black box identification technique for TRT data analysis. A nonlinear autoregressive exogenous (ARX) model structure and stochastic search algorithms are used to estimate model parameters. Artificial intelligence techniques, Genetic Algorithm and Particle Swarm Optimization Algorithm are employed to avoid local maxima problems. The study is based on data sets obtained during real TRT tests without phase change effects. All analyses are performed in Matlab environment. The purpose of this paper is to verify that the proposed algorithms are suitable for processing of TRT data with the aim of future identification of thermal parameters of boreholes with phase change effects. The given solution is also useful when common techniques fail in search for the global optimum if the search space is not differentiable or linear in the parameters.