印度中央邦马尔瓦地区利用人工神经网络和回归模型预测蒸腾量的比较(摘选)

Comparison of Artificial Neural Network Models and Regression Model for Prediction of Evaporation for Malwa Region of Madhya Pradesh in India(Extracts)

蒸发是水循环的一个重要组成部分,对蒸发量的估算是对水资源和灌溉水量有效利用的一个重要手段。该研究旨在利用多元线性回归模型、多层感知器(MLP)和人工神经网络(ANN)模型模拟印度中央邦马尔瓦地区周蒸发量。利用4种不同天气变量组合训练神经网络模型。多元线性回归模型只将最高温和相对湿度作为输入值,但是模拟结果不令人满意。MLP模型采用的数据集包括最高和最低温度、风速和相对湿度,在训练和验证中都取得了比较好的结果。MLP模型可以用来模拟周开放式蒸发皿蒸发量,估算缺失数据,并可以作为替代模型以验证蒸发量测定值。降雨量数据并不能改善模型性能。

Estimation of evaporation,a major component of the hydrologic cycle,is essential for any efficient management of water resources and irrigation water demand.In this paper,an attempt has been made to study the simulation of weekly evaporation by using multiple linear regression model,multilayer perceptron(MLP)and an artificial neural network(ANN)model for Malwa regions of Madhya Pradesh in India.The four combination of different weather variables were used to train the neural network model.The multiple linear regression model includes only the maximum temperature and relative humidity as input but the simulation result was not satisfactory.The MLP model performed quite well for the data set comprising minimum and maximum temperature,wind speed and relative humidity during training as well as validation periods.The MLP model could be employed to simulate weekly open pan evaporation,estimate missing data and as an alternative model to check the observed evaporation data.Inclusion of rainfall data with meteorological variables has not improved the performance of the model.