【摘要】 大型旋转机械的状态预示技术是实现设备状态维护的关键,针对大型旋转机械的几种典型趋势,提出支持向量机(support vector machines,SVM)进行系统故障趋势预示的模型,采用BP(back propagation)神经网络模型和SVM模型对不同的趋势进行预测,结果表明SVM模型具有预测精度高的特点。在以上研究的基础上,提出一种新的旋转机械系统状态组合预测模型。该模型采用振动烈度和特征频率分量作为预测机械系统状态的敏感因子,采用从时域到频域、频域到时域,构建旋转机械状态预测的组合模型。将基于SVM的组合预测模型应用于旋转注水机组的状态预测,取得较好的预测效果。更多还原

【Abstract】 The condition forecasting technology of large rotating machinery is the key for machinery condition maintenance. A novel system fault trend prediction method based on SVM(support vector machines) for several typical trends is put forward. The SVM model predicting results compared with BP(back propagation) neural network model proves that SVM model is of high predicting accuracy. Based on this, a new condition hybrid forecasting model is presented. This method uses the vibration velocity in time domain and vibration amplitude of the feature frequency in frequency domain as condition trend prediction sensitive factors. The method is used for the experimental system as well as the water injection sets. The good predicting results show that the hybrid method based on SVM has excellent performance of condition trend prediction for mechanical equipment.更多还原