基于局部切空间排列和最小二乘支持向量机的气缸压力识别

为了提高缸盖振动信号恢复气缸压力的识别精度,提出一种基于局部切空间排列(LTSA)和最小二乘支持向量机(LSSVM)的气缸压力识别方法。首先提取缸盖振动信号时域、频域及小波包能量域特征,组成高维特征集,利用LTSA算法提取高维特征集的低维本征流形特征,然后把降维后的特征参数集作为LSSVM模型输入,缸压信号作为LSSVM模型输出,通过多个样本对LSSVM模型进行训练,从而获得气缸压力的重构模型。试验结果表明:基于局部切空间排列和最小二乘支持向量机的气缸压力识别方法具有精度高、泛化能力强等优点。

Recognition of cylinder pressure based on LTSA-LSSVM

In order to improve the recognition accuracy of cylinder head vibration signal recovery cylinder pressure, a recognition method based on local tangent space arrangement (LTSA) and least squares support vector machine (LSSVM) is proposed. Firstly, the multi-dimensional features of the cylinder head vibration signals were extracted on time-domain, frequency-domain and wavelet packet energy-domain, and the dimension of feature set was reduced by LTSA method. Feature set reduced was used as the input of LSSVM model, and cylinder pressure was used as the output. Then a model for recognition of cylinder pressure was obtained by training the LSSVM model with plurality of samples. The results show that the cylinder pressure recognition method based on local tangent space arrangement and least squares support vector machine has the advantages of high precision and strong generalization ability.