Optimal chilled water temperature calculation of multiple chiller systems using Hopfield neural network for saving energy

The values of chilled water supply temperatures in chillers indicate the load distributions as the chilled water return temperatures in all chillers are the same in a decoupled air-conditioning system. This study employs the Hopfield neural network (HNN) to determine the chilled water supply temperatures in chillers, which are used to solve the optimal chiller loading (OCL) problem. A linear input–output model is utilized as a substitute for the sigmoid function, which eliminates the shortcoming of the conventional HNN method. Notably, HNN overcomes the flaw in the Lagrangian method in that the latter cannot be utilized for solving the OCL problem as its power-consumption models include non-convex functions. The chilled water supply temperatures are used as variables to be solved for a decoupled air-conditioning system and solve the problem using the HNN method to overcome the defect in the Lagrangian method. After analysis of the case study and comparison of results using these two methods, we conclude that the HNN method solves the problem of the Lagrangian method, and produces highly accurate results. The HNN method can be applied to the operation of air-conditioning systems.