In this study, the thermal performance of solar air heater (SAH) has been predicted using artificial neural network (ANN) with relevant input parameters. To complete this aim, two different types of SAHs were developed using roughened (arc shaped wire rib) and smooth duct. Many researchers have been used system parameters, operating parameters and meteorological parameters to predict the performance of SAH by analytical or conventional approach and ANN technique, but performance prediction by using relevant input parameters has not been done so far. Therefore, neural model that has been developed with relevant input parameters is considered in this study. Total ten parameters are used to find out the relevant parameters for prediction. Seven different neural models have been constructed using these parameters. In each, 10 to 20 neurons have been selected to find out optimal model. It has been found that ANN-II with 8-14-1 is the optimal model as compared to other models. The values of SSE, MRE and R² were found to be 0.02138, 1.82% and 0.99387 respectively, for ANN-II. The effectiveness of neural model has been examined by comparing with Group method of data handling (GMDH) approach and found that the ANN performed better than GMDH model. In addition to this, sensitivity analysis has been performed to find out the most sensitive input parameter and observed that the mass flow rate of air is the most effective input parameter.

在这项研究中，已使用具有相关输入参数的人工神经网络（ANN）预测了太阳能空气加热器（SAH）的热性能。为了实现此目标，使用了粗糙的（弧形金属丝肋）和光滑的导管开发了两种不同类型的SAH。许多研究人员已经使用系统参数，操作参数和气象参数通过分析或常规方法以及ANN技术来预测SAH的性能，但是到目前为止，尚未通过使用相关输入参数来进行性能预测。因此，在这项研究中考虑了已经用相关输入参数开发的神经模型。总共十个参数用于查找相关参数以进行预测。使用这些参数已构建了七个不同的神经模型。每一个，已选择10至20个神经元来找出最佳模型。已经发现，与其他模型相比，具有8-14-1的ANN-II是最佳模型。SSE，MRE和R的值ANN-II的^2个分别为0.02138、1.82％和0.99387。通过与数据处理的分组方法（GMDH）进行比较，检验了神经模型的有效性，发现神经网络的性能优于GMDH模型。除此之外，还进行了敏感性分析以找出最敏感的输入参数，并观察到空气的质量流量是最有效的输入参数。