Prediction Model for Superheat of Organic Rankine Cycle Power Generation System Based on NARX Neural Network

In order to accurately establish a prediction model for the superheat at the outlet of the evaporator in the Organic Rankine Cycle low-temperature waste heat power generation system, a method utilizing an improved particle swarm optimization algorithm to create a neural network model for the Organic Rankine Cycle power generation system is proposed. First, the modeling data is obtained from the operational data of a certain Organic Rankine Cycle low-temperature waste heat power generation system in Jiangsu Province over one day. The energy input, output, and loss serve as the indicators for establishing the superheat prediction model of the lowtemperature waste heat power generation system. Additionally, an improved particle swarm optimization algorithm is proposed, and its superiority over traditional particle swarm optimization algorithms and other improved versions is verified using various functions. The improved particle swarm optimization algorithm is employed to create the PSO-NARX neural network model for predicting the superheat of the low-temperature waste heat power generation system, optimizing the initial weights and thresholds of the NARX neural network. Experimental results demonstrate that the PSO-NARX neural network model outperforms the traditional NARX neural network model in establishing the superheat prediction model.