Parameter identification of thermoeletric modules using particle swarm optimization

This paper presents a methodology to estimate thermoelectric module (TEM) internal parameters based on particle swarm optimization (PSO) algorithm. To obtain the correct TEM representation, it is necessary a proper model identification procedure to represent the TEM operation, both in DC and other relevant frequencies. Classical methods for linear parameter estimation are not suitable for the nonlinear TEM characteristics of the proposed model. We devise a model with twenty-one parameters, which represent parts of the two TEMs employed, including top, lower and middle layers and heat-sinks. The TEM is excited using an electrical current signal with power spectral density of a white noise, and the temperature response is adopted as output for the PSO algorithm to make the estimation. For numerical stability and proper estimation, the white noise excitation is filtered before, to obtain a dynamically persistent signal with high and low frequencies components. Simulation results show the effectiveness of the PSO in TEM parameters estimation.