To deal with the problem of DHN modelling, this manuscript introduce a refinement approach to develop a novel method for district heat network aggregation based on soft computing techniques (Neural Networks and meta-heuristics). Reducing size and complexity of DHN by preserving the dynamic properties of district heating without physical description (network topology, pipe diameter and insulation, pump characteristics etc.) is a main challenge for simulating various operational strategies. Data from reel DHN system was used to investigate the ability of the proposed approach. This approach embed the time of flow transportation in the inputs of the FFNN model. The time of flow transportation between production plant and consumers substations is determined using the wavelet analysis. The case of study has provided encouraging results about the proposed model. However, this can be more enhanced by considering for instance the quality of the measured data and then finds application for DHN modelling.