Simulation optimization and leakage detection of secondary heating networks in complex communities

Abstract To improve the efficiency and accuracy of leakage detection in district heating networks, this study proposes an integrated technical framework of simulation optimization, model construction, and engineering validation. First, in the simulation optimization stage, the resistance characteristic coefficients in the hydraulic model are calibrated using the Broyden–Fletcher–Goldfarb–Shanno algorithm based on actual pressure data from monitoring points. The optimized model reduces the relative error of pressure simulation for all users to within 5%, providing reliable data support for subsequent analysis. Next, in the model construction stage, a leakage detection model is developed by integrating principal component analysis, data standardization, and a back propagation neural network. Meanwhile, a four-level hierarchical hydraulic condition simulation model is established for a residential heating network in Shandong, effectively representing complex topological structures. Finally, in the engineering validation stage, the proposed leakage detection model achieves a prediction accuracy of 93.8% when the detection distance is less than 50 m, and 98% when the distance ranges from 50 m to 200 m. Field test results further demonstrate that the developed system exhibits rapid response and robust performance.