中文摘要：

针对雷达发射机的故障状态具有强的随机性和不确定性问题，结合最小二乘支持向量机（LSSVM）能够对信号进行非线性预测和隐马尔可夫模型（HMM）能够进行较为精确的似然度概率计算的特点，提出了基于LSSVM-HMM的故障状态预测模型。通过基于小波包的SSNF算法对采集的磁控管电流信号进行去噪后提取有效的非平稳和非线性特征，用正常时的特征向量来训练HMM，并利用该模型对未知信号的特征向量及用LSSVM对其预测到的特征向量进行状态监测，从而获得故障出现的概率。实验结果表明，该模型用于小样本的发射机故障预测是有效的尝试。

英文摘要：

There is strong randomness and uncertainty lying in the fault status of radar transmitter. Because LSSVM can make nonlinear prediction and HMM has the ability to obtain accurate likelihood probability; a fault prediction model based on LSSVM-HMM is presented. Using wavelet SSNF algorithm, the nonlinear and unsteady feature of the magnetron current signal is extracted and HMM is trained by the normal features. Then the unknown signal feature and its corresponding prediction calculated by LSSVM are both applied to the trained HMM, and their fault likeli-hood is obtained. Experiment results show that the proposed model is an effective trying to fault prediction of trans-mitter with small training samples.