论文部分内容阅读
支持向量机具有完备的统计学习理论基础和学习功能。它用核函数建立预测模型,再用已知数据为学习样本训练学习机,用检验样本进行验证、预测系统未来故障。最小二乘支持向量机(LS-SVM)采用最小二乘线性系统作为损失函数,函数估计精度高、收敛速度快。基于支持向量机的多层参数寻优、等维信息一步预测和不等维信息多步预测,可用于飞机状态评估、故障诊断和参数预测以及故障率分析。
Support vector machine has a complete statistical learning theory and learning functions. It uses the kernel function to establish the prediction model, then uses the known data to train the learning machine for learning samples, verifies with the test samples and predicts the future failure of the system. Least square support vector machine (LS-SVM) uses the least-squares linear system as the loss function, with high estimation accuracy and fast convergence speed. Multi-layer parameter optimization based on SVM, one-dimensional prediction of one-dimensional information and multi-step prediction of unequal dimensional information can be used for aircraft condition assessment, fault diagnosis, parameter prediction and failure rate analysis.