摘要
机床加工状态对加工工件质量有很大的影响,因此识别机床加工状态有重要的意义。依据采集的机床加工数据,通过FFT频谱分析,划分出机床加工的3种状态。利用小波包分解,分别求出各种状态在不同频带节点上的能量分布百分比,并把它作为隐马尔科夫模型的输入特征向量。按照隐马尔科夫模型模式识别方法,建立3种标准状态的训练优化模型库,把测试样本代入优化模型库中,依据最大对数似然值对机床的加工状态进行了识别。计算结果表明,状态识别结果正确。
The states of machine tools processing are very important to quality of workpiece, so state recognition of machine tools processing is very significance. The machine tool processing was divided into three states by FFT spectrum analysis according to sample data. The energy distribution of frequency band by wavelet packet decomposition was regarded as the feather vector of HMM. According to the HMM pattern recognition method, training optimization model library of three standard states was set up. A case was studied for the state recognition of machine tool processing after the test samples were substituted into optimization model library. It is shown that the recognition results are correct by the wavelet packet-HMM method.
出处
《机床与液压》
北大核心
2013年第7期202-204,共3页
Machine Tool & Hydraulics
基金
江西省自然科学基金资助项目(20114BAB206003)
载运工具与装备教育部重点实验室资助项目(09JD03)