【摘要】 介绍了基于卷积神经网络开发的fCaO软测量软件。操作人员可以根据现场实际情况,选取与f CaO含量相关性较大的变量在软件中进行组态,用历史数据进行软测量模型的训练和测试,获得较好的软测量模型,预测实时的生产数据。在重庆某水泥厂的实际应用中,fCaO软测量软件的总体在线预测误差为4.57%,取得了较好效果。更多还原

【Abstract】 FCaO soft measurement software developed by convolutional neural network is introduced.According to the actual situation on site, operators can select variables that are more relevant to the content of fCaO for configuration in the software, and use historical data for training and testing of softsensing model to obtain a better soft-sensing model and predict real-time production data.According to the practical application of a cement plant in Chongqing, the overall online prediction error is 4.57%, and good results have been obtained.更多还原