提出一种基于时序预报神经网络的工业过程故障预报方法，同时给出了描述神经网络预报和外推能力的表达方式，并以氯碱电解工艺的现场数据验证了这种故障预报方法的有效性．实验结果表明，该方法可成功地用以实现氯中含氢的２４小时预报． A method of industrial process fault forecasting based on time series forecasting neural network is presented. At the same time, an expression method of neural network forecasting and extrapolation capability is given. The validity of this method is verified by field data of chlor-alkali electrolysis process. The experimental results show that this method can be successfully used to achieve 24-hour forecast of hydrogen in chlorine.