论文部分内容阅读
引入基于支持向量机(SVM)的数据挖掘技术,提出了基于SVM的转静碰摩部位诊断知识获取.首先,基于带机匣的航空发动机转子实验器,模拟了4个碰摩部位的碰摩实验,利用机匣4个部位的应变测试,获取了4个碰摩部位和4个测点的大量实验数据;然后提出了一种基于支持向量聚类(SVC)的诊断知识规则提取方法.在该方法中,利用SVC算法得到特征选取后样本的聚类分配矩阵,最后根据聚类分配矩阵构建超矩形规则.为使规则更加简洁,易于解释,采用规则合并、维数约简、区间延伸等方法对超矩形规则进行进一步简化.利用基于SVM的数据挖掘方法,从大量的碰摩部位实验数据中提取出了转静碰摩部位诊断的知识规则,并进行了相应解释和验证,规则识别率达到了99%以上,表明了该方法的正确有效性.
The data mining technology based on Support Vector Machine (SVM) was introduced and the knowledge acquisition of static friction rubbing site based on SVM was proposed.Firstly, based on the aero-engine rotor experimental set with receiver, In the experiment, a large number of experimental data of 4 rubbing sites and 4 measuring points were obtained by using the strain test of four parts of the casing. Then, a new method of extracting knowledge based on support vector clustering (SVC) was proposed. In this method, the SVC algorithm is used to get the cluster distribution matrix of the selected samples, and finally the hyper-rectangle rules are constructed according to the cluster assignment matrix. In order to make the rules more concise and easy to be explained, the rules are merged, the dimension reduction, interval extension and so on Method to further simplify the super-rectangle rule.Using SVM-based data mining method, the knowledge rules for the diagnosis of static rubbing sites are extracted from a large number of rubbing site experimental data, and correspondingly explained and verified, the rule recognition rate Reached more than 99%, indicating the correctness of the method.