基于神经网络的海水淡化系统产水模拟及预测

海水淡化是一个涉及传热传质等诸多因素的复杂的非线性过程。为了克服传统数学建模方法在模拟仿真精度和实时性方面的不足,对大样本空间的神经网络构造及网络学习加速方法进行了大量探索及尝试,成功地将神经网络技术引入海水淡化产水过程中构造新的产水模型,建立了以空气入口干球温度、预冷器进口冷却水温度、海水喷淋温度和海水喷淋量作为输入参数的海水淡化系统神经网络模型。分析表明,该模型不仅具有较高的仿真精度,而且保持了系统原有的光滑性,能满足系统实时仿真模拟预测的要求。

Simulation and Forecast of Water-Production for Desalination System on the Basis of Artificial Neural Network

Seawater desalination is a complex and non-linear process which relates to many factors such as heat and mass transfer.The traditional mathematic model has shortages in simulative precision and real-time.In order to avoid these shortages,a huge amount of exploratory work concerning the construction of neural networks have been conducted with a large sample space as well as an intensive study on the network learning acceleration method.As a result,artificial neural network(ANN) technology has been successfully used to simulate water production of the desalination system.Then,a new simulation model has been constructed,in which the dry temperature of the air,the inlet cooling water temperature,the mass and the sprinkler temperature of seawater are input parameters of ANN.Result shows the ANN model for the desalination system has not only higher simulation precision but also maintenance of the system original smoothness.Furthermore,it has also satisfied the system real-time function.