机组状态参数的趋势分析及计算机表达是影响在线故障诊断的重要因素。文中运用奇异谱理论对状态参数的趋势进行了分析 ,运用人工神经网络完成对机组状态参数典型趋势的在线识别。该方法可以对机组状态参数 (如振动、温度、压力等 )进行有效的识别 ,为水轮机组的状态分析、状态评估和预测提供有效的辅助分析手段 ,从而为水电厂状态维修提供了参考。 The trend analysis and computer expression of unit status parameters are important factors that affect on-line fault diagnosis. In this paper, we analyze the trend of state parameters by using singular spectrum theory and use artificial neural network to identify the typical trend of unit state parameters online. The method can effectively identify unit state parameters (such as vibration, temperature, pressure, etc.) and provide effective auxiliary analysis methods for state analysis, status assessment and prediction of hydro-turbine units, thus providing a reference for state maintenance of hydropower plants.