Evaluation of dissolved oxygen in water by artificial neural network and sample optimization

来源期刊：中南大学学报(英文版)2008年第z2期

论文作者：陈丽华 李丽

文章页码：416 - 420

Key words：dissolved oxygen (DO)； BP neural network； evaluation； concentration

Abstract: Three important factors influencing directly the dissolved oxygen (DO) of river including the outflow, the water temperature and the pH, were used as input parameters to set up a BP neural network based on Levenberg-Marquant algorithm. The neural network model was proposed to evaluate DO in water. The model contains two parts: firstly, the learning sample is unified; secondly, the neural network is used to train the unified samples to ensure the best node number of hidden layer. The proposed model is applied to assessing the DO concentration of the Yellow River in Lanzhou city. The evaluation result is compared with that by the neural network method and the reported result in Lanzhou city. The comparison result indicates that the performance of the neural network model is practically feasible in the assessment of DO. At the same time, the linear interpolation method can add the number of network’s learning sample to improve the prediction precision of the network.