论文部分内容阅读
神经网络大机组状态预报技术的研究、开发及应用成果于1999年3月27日在岳阳长岭通过了由中国石油化工集团公司组织的专家鉴定。鉴定意见认为,经过3年的现场应用,证明所开发的神经网络大机组状态预报技术可正确反映现场的各种振动变化趋势;在样本选择得当时,其中长期(时间跨度1～2个月)预报精度达90％以上,达到了实用化的要求;如进一步提高样本处理技术水平,完善相关软件,该技术将可实现商品化,走向市场,这是我国设备诊断工程界的一项重要成果。
The research, development and application achievements of neural network large unit state forecasting technology were approved by experts of China Petrochemical Corporation in March, 1999 in Yueyang Changling. According to the appraisal opinion, after 3 years of field application, it is proved that the developed neural network large unit state forecasting technology can correctly reflect various vibration changing trends in the field. When the samples are selected properly, the long-term (time span of 1-2 months) The accuracy of forecasting is over 90%, which meets the requirements of practical application. If we further improve the sample processing technology and improve the related software, the technology will be commercialized and marketed. This is an important achievement in the field of equipment diagnostics in China.