刀具破损状态的特征提取及自动识别

Characteristic Extraction and Automatic Recognition for Tool Breakage

文章采用机床功率法和声发射法对车削过程中的刀具破损进行监控。在试验中发现了刀具破损时机床功率信号的四种表现形式 ,说明了刀具破损形式的随机性。针对这种情况 ,首次提出了功率信号处理的延时方差法 ;对切削过程中发出的各种声发射 (AE)信号采用时频分析进行处理并提取出反映刀具破损的特征量 ,最后利用神经网络ART2实现了刀具破损状态的自动识别。

Machine power and AE sensor are used to monitor the tool breakage during turning. In the tests, four presenting forms for machine power signals when tool breaks are found, which means the tool breakage form is stochastic. In respect to this, for the first time, a delayed variance way is put forward for processing power signal; as for various AE signals sent out during cutting, time_frequency analysis is used to process that and characteristic value reflecting tool breakage is extracted, finally, automatic recognition is realized for tool breakage by using ART2 neural network.