high performance computing on graphics processing units: hgpu.org

This paper presents a parallel approach to estimate the parameters in the heat loading of a district heating system by use of the traditional particle swarm optimisation (TPSO) on the Graphic Processing Unit (GPU) using OpenCL. The running time of the algorithm is greatly reduced compared to running on CPU. The heat load is approximated by the sum of time dependent and temperature dependent components. The temperature dependent component is approximated using the sum of two peak functions and temperature dependent component is approximated using a generalised logistic function.