%0 Journal Article
%T 基于神经网络的螺杆膨胀机发电变工况预测方法研究<br>Research on Prediction Method of Power Generation Variable Condition of Screw Expander Based on Neural Network
%A 史松伟
%A 王昇
%A 赵诚
%A 孙站望
%A 徐旭
%A 朱万里
%J Dynamical Systems and Control
%P 221-236
%@ 2325-6761
%D 2025
%I Hans Publishing
%R 10.12677/dsc.2025.143023
%X 螺杆膨胀机是余压发电系统的重要动力设备&#65292;为了获得螺杆膨胀机发电系统的较优运行工况&#65292;本文提出了基于BP神经网络模型、广义回归神经网络模型和卷积神经网络模型的螺杆膨胀机变工况的预测方法&#65292;并对三种神经网络模型的预测性能进行了比较分析。预测结果表明&#65306;广义回归神经网络模型的预测效果最为理想&#65292;卷积神经网络模型预测结果的偏差较大&#65292;它们对螺杆膨胀机的功率和发电量预测的平均相对误差分别为4.07%、4.93%、39.32%和8.34%、12.14%、39.39%。将螺杆膨胀机的功率和发电量分为较高、中等和较低三个范围&#65292;采用广义回归神经网络模型预测了相应范围的入口蒸汽流量、蒸汽压力和蒸汽温度&#65292;最终对螺杆膨胀机发电的运行工况进行了优化&#65292;这对于螺杆膨胀机发电系统的节能减排具有重要意义。<br>The screw expander is a critical power equipment in the residual pressure power generation system. To obtain the optimal operating conditions of the screw expander power generation system, this paper proposes a prediction method for the variable operating conditions of the screw expander based on the BP neural network model, the generalized regression neural network model and the convolutional neural network model. It conducts a comparative analysis of the prediction performance of the three neural network models. The prediction results show that the prediction effect of the generalized regression neural network model is the most ideal. In contrast, the deviation of the prediction results of the convolutional neural network model is relatively large. The average relative errors of their predictions for the power and power generation of the screw expander are 4.07%, 4.93%, 39.32% and 8.34%, 12.14%, 39.39% respectively. The power and power generation of the screw expander were divided into three ranges: high, medium and low. The generalized regression neural network model was adopted to predict the corresponding range of inlet steam flow, steam pressure, and steam temperature. Finally, the operating conditions of the screw expander power generation were optimized, which is of great significance for energy conservation and emission reduction of the screw expander power generation system.
%K 螺杆膨胀机&#65292
%K 余压发电&#65292
%K 神经网络&#65292
%K 运行工况<br>Screw Expender
%K Residual Pressure Power Generation
%K Neural Network
%K Operating Conditions
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=118878