摘要
构造Vogl快速算法误差反向传播(EBP)神经网络,应用该神经网络对若干固体在超临界流体中的溶解度进行预测,对21体系共612个数据点进行训练和预测,预测的总平均相对误差为4 02%,优于状态方程法所计算的结果。
An error back propagation (EBP) artificial neural network with Vogl algorithm was constructed to predict the solubilities of different solids in supercritical fluids.612 experimental point for 21 systems had been trained and predicted,the predicting total average relative error is 4.02%.This method is superior to equation of state method.
出处
《化学工业与工程》
CAS
2004年第3期189-192,共4页
Chemical Industry and Engineering
关键词
人工神经网络
超临界流体
溶解度
artificial neural networks
supercritical fluids
solubility