摘要
受到设备结构以及工作环境等多方面因素的影响,煤矿设备机电振动状态数据在采集过程中可能包含大量余噪声,这也给异常状态识别增加较大的难度。对此,提出基于孤立森林的煤矿设备机电异常振动状态识别方法。首先,采用移动平均滤波法,对移动窗口进行定义,从而实现振动状态数据的平滑以及去噪处理;然后,将数据方差以及峰值作为时域特征的衡量指标,从原始状态序列中提取出特征参数。并引人特征平衡参数,对特征贡献度进行分配,从而实现特征融合处理;最后,构建孤立森林模型,通过异常分数的计算结果以及设定的异常检测阈值,实现异常波动状态识别。在实验中,对提出的方法进行识别稳定性的检验。最终的测试结果表明,采用提出的方法对异常振动状态进行识别时,识别结果的变异系数较低,具备较为理想的识别稳定性。
Due to the influence of many factors such as equipment structure and working environment,the electromechanical vibration state data of coal mine equipment may contain a lot of redundant noise during the collection process,which also increases the difficulty of abnormal state identification.In this paper,an isolated forest-based method for identifying abnormal vibration state of electromechanical equipment in coal mine is proposed.Firstly,the moving average filtering method is used to define the moving window,so that the vibration state data can be smoothed and denoised.Then,the data variance and peak value are taken as the time domain feature index,and the feature parameters are extracted from the original state sequence.And the feature balance parameter is introduced to distribute the feature contribution degree,so as to realize the feature fusion processing.Finally,the isolated forest model is constructed,and the abnormal vibration state is recognized through the calculation result of the anomaly fraction and the abnormal detection threshold.In the experiment,the recognition stability of the proposed method is tested.The final test results show that when the proposed method is used to identify the abnormal vibration state,the coefficient of variation of the recognition results is low,and the recognition stability is relatively ideal.
作者
靳远志
Jin Yuanzhi(Shandong Energy Zaozhuang Mining Group,Jining 277600,China)
出处
《办公自动化》
2024年第13期7-9,59,共4页
Office Informatization
关键词
孤立森林
煤矿设备
异常状态
振动信号
识别方法
isolated forest
coal mine equipment
abnormal state
vibration signal
identification method
