Abstract The solar radiation fluctuation occurs at practically anywhere on the earth. When a solar thermal power generation system is designed for the areas with considerable solar radiation fluctuation, the collector tube exit temperature becomes more difficult to predict and requires significant calculation time. This paper presents a fast and accurate transient thermal prediction method to predict the parabolic trough collector tube exit temperature. In this work, an artificial neural network (ANN) is combined with the principle of superposition. ANN is used to predict the exit temperature rise caused by a single heat flux pulse in the first step of the proposed methodology, while superposition is used to predict the from multiple heat flux pulses in the second step. Limited cases of conjugate heat transfer analytical results by the finite element method (FEM) are used to train the ANN. The one-day exit fluid temperature from 7 a.m. to 6 p.m. is predicted within 1 min of computational time with mean absolute deviation less than 2 K. The exit fluid temperature of the collector tube for one year operation can be predicted in less than 6 h. Because fluid velocity is included in the input parameters, the proposed methodology is especially useful for flow control simulations where a constant exit temperature is targeted. Through this, the optimum performance of collector tube under multiple radiation conditions can be assessed during an early design phase of parabolic solar trough systems. The predicted results can be used for initial system planning, heat balance analysis, and system design. Since the method shows good prediction capability under the fluctuating solar radiation as well as the stable solar radiation, it is applicable to be used for designing the parabolic trough technology at any weather conditions in the world.

中文翻译：

---

基于人工神经网络的抛物槽式太阳能集热管瞬态热预测方法

摘要 太阳辐射波动几乎发生在地球上的任何地方。当太阳能热发电系统设计用于太阳辐射波动较大的地区时，集热管出口温度变得更加难以预测并且需要大量的计算时间。本文提出了一种快速准确的瞬态热预测方法来预测抛物线槽集热管出口温度。在这项工作中，人工神经网络（ANN）与叠加原理相结合。在所提出的方法的第一步中，人工神经网络用于预测由单个热通量脉冲引起的出口温度升高，而在第二步中，叠加用于预测由多个热通量脉冲引起的出口温升。有限元法 (FEM) 的共轭传热分析结果的有限情况用于训练 ANN。早上7点到下午6点一天的出口流体温度在计算时间的1分钟内预测，平均绝对偏差小于2 K。 集热管运行一年的出口流体温度可以在不到6小时的时间内预测. 由于流体速度包含在输入参数中，因此所提出的方法对于以恒定出口温度为目标的流量控制模拟特别有用。通过这种方式，可以在抛物面太阳能槽系统的早期设计阶段评估集热管在多种辐射条件下的最佳性能。预测结果可用于初始系统规划、热平衡分析和系统设计。