几何模型分类器具有坚实的几何统计基础和良好的泛化能力,因此在旋转机械故障诊断中取得了较高的分类精度.与仿射包和凸包相比,超圆盘(Hyperdisk,HD)对样本分布区域的估计更加合理.但超圆盘模型属于浅层学习模型,对复杂函数的表示能力有限,存在学习能力和泛化能力差等缺点.针对这个问题提出一种深度超圆盘分类器(Deep Hyperdisk Large Margin Classifier,DHD),该方法通过模块叠加的方式将超圆盘分类器深度化,利用特征提取公式从每层模块的输入样本中自主提取新的特征值,并将其应用在下一层模块的训练学习中.将所提方法应用到旋转机械故障诊断当中,实验结果表明该方法对故障样本的分类准确率高于其他模型算法,且对不均衡样本和强噪声背景下的故障样本均具有良好的分类能力.
Application of Deep Hyperdisk Large Margin Classifier in Rotating Machinery Fault Diagnosis
The geometric model classifier has a solid foundation of geometric statistics and good generalization ability,so it can achieve high classification accuracy in rolling bearing fault diagnosis.Compared with affine hull and convex hull,the estimation of sample distribution region by hyperdisk(HD)is more reasonable.However,the hyperdisk model is a shallow learning model,which has limited ability to express complex functions,and has poor learning ability and generalization ability.To solve this problem,a deep hyperdisk large margin classifier(DHD)is proposed.The method deepens the hyperdisk classifier by module superposition,and automatically extracts new feature values from the input samples of each module by feature extraction formula.And the extracted features are applied to the training of the next module.The proposed method is applied to the fault diagnosis of rotating machinery.The experimental results show that the classification accuracy of the proposed method is higher than that of other model algorithms.It has good classification ability for unbalanced samples and has strong anti-interference ability against noise.
fault diagnosisdeep hyperdisk large margin classifierdeep learningrotary machines
国家科技期刊平台
NSTL
万方数据
