基于深度信念网络的空压机故障监测研究
|
摘要
在螺杆式空压机的各种故障中,排气温度异常的故障十分常见。针对排气温度异常的故障监测,提出了一种基于深度信念网络的螺杆式空压机排气温度监测方法。首先对空压机运行参数进行相关性分析,然后选取相关性较强的参数,采用深度信念网络建立了空压机温度模型,用于排气温度的故障监测,接着用核密度估计方法确定空压机排气温度异常的故障阈值。最后根据实际数据进行仿真实验,以此来检验方法的有效性,经过性能对比,该方法对空压机排气温度的故障监测具有更高的准确性。
In the various failures of the worm air compressor,abnormalities in the exhaust gas temperature are very common. Aiming at the fault monitoring of abnormal exhaust temperature,a method of monitoring the exhaust temperature of worm air compressor based on deep belief network has been proposed. Firstly,the correlation analysis of the air compressor operating parameters was carried out,and then the parameters with strong correlation were selected to establish the deep belief network temperature model of the normal operation of the air compressor for the prediction of exhaust gas temperature. Finally,the fault threshold of the air compressor exhaust temperature abnormality was determined with kernel density estimation method,and the fault simulation was performed based on the actual data to verify the effectiveness of the method. After comparison with the actual data,the method is accurate and effective in monitoring the air compressor exhaust temperature.
In the various failures of the worm air compressor,abnormalities in the exhaust gas temperature are very common. Aiming at the fault monitoring of abnormal exhaust temperature,a method of monitoring the exhaust temperature of worm air compressor based on deep belief network has been proposed. Firstly,the correlation analysis of the air compressor operating parameters was carried out,and then the parameters with strong correlation were selected to establish the deep belief network temperature model of the normal operation of the air compressor for the prediction of exhaust gas temperature. Finally,the fault threshold of the air compressor exhaust temperature abnormality was determined with kernel density estimation method,and the fault simulation was performed based on the actual data to verify the effectiveness of the method. After comparison with the actual data,the method is accurate and effective in monitoring the air compressor exhaust temperature.
引文
[1]杨磊.螺杆式空气压缩机常见故障分析与维护技术[J].煤炭科学技术,2013,41(S1):133-135.
[2]李成.螺杆式空压机常见故障及消除方法探究[J].科技创新与应用,2015(13):117.
[3]杨信一,孙庆东,郭新,等.螺杆式空压机温度的影响和改进分析[J].天然气技术与经济,2017,11(4):40-42,83.
[4]阴彦磊,何邦华,唐军,等.基于云推理的叶丝干燥筒壁温度预测方法[J].烟草科技,2017,50(3):80-87.
[5]练斯甄,周璇,闫军威.基于改进BP神经网络的冷水机组IPLV预测方法[J].建筑科学,2014,30(6):66-71.
[6]丁美昆,徐昱琳,蒋财军.深度信念网络研究综述[J].工业控制计算机,2016,29(4):80-81.
[7] PAUL S,SINGH L. A review on advances in deep learning[C]//2015 IEEE Workshop on Computational Intelligence:Theories,Applications and Future Directions(WCI).[S. l.],IEEE,2015:1-6.
[8]石鑫,朱永利,萨初日拉,等.基于深度信念网络的电力变压器故障分类建模[J].电力系统保护与控制,2016,44(1):71-76.
[9]孔祥玉,郑锋,鄂志君,等.基于深度信念网络的短期负荷预测方法[J].电力系统自动化,2018,42(5):133-139.
[10] YANG T,HUANG S. Fault Diagnosis Based on Improved Deep Belief Network[C]//2017 5th International Conference on Enterprise Systems(ES).[S. l.],IEEE,2017:305-310.
[11]吉全霖,卿黎,韩宇楠,等.基于随机森林模型的锅炉汽包水位失效分析[J].中国水运(下半月),2018,18(1):99-101.
[12]代杰杰,宋辉,杨祎,等.基于深度信念网络的变压器油中溶解气体浓度预测方法[J].电网技术,2017,41(8):2737-2742.
[2]李成.螺杆式空压机常见故障及消除方法探究[J].科技创新与应用,2015(13):117.
[3]杨信一,孙庆东,郭新,等.螺杆式空压机温度的影响和改进分析[J].天然气技术与经济,2017,11(4):40-42,83.
[4]阴彦磊,何邦华,唐军,等.基于云推理的叶丝干燥筒壁温度预测方法[J].烟草科技,2017,50(3):80-87.
[5]练斯甄,周璇,闫军威.基于改进BP神经网络的冷水机组IPLV预测方法[J].建筑科学,2014,30(6):66-71.
[6]丁美昆,徐昱琳,蒋财军.深度信念网络研究综述[J].工业控制计算机,2016,29(4):80-81.
[7] PAUL S,SINGH L. A review on advances in deep learning[C]//2015 IEEE Workshop on Computational Intelligence:Theories,Applications and Future Directions(WCI).[S. l.],IEEE,2015:1-6.
[8]石鑫,朱永利,萨初日拉,等.基于深度信念网络的电力变压器故障分类建模[J].电力系统保护与控制,2016,44(1):71-76.
[9]孔祥玉,郑锋,鄂志君,等.基于深度信念网络的短期负荷预测方法[J].电力系统自动化,2018,42(5):133-139.
[10] YANG T,HUANG S. Fault Diagnosis Based on Improved Deep Belief Network[C]//2017 5th International Conference on Enterprise Systems(ES).[S. l.],IEEE,2017:305-310.
[11]吉全霖,卿黎,韩宇楠,等.基于随机森林模型的锅炉汽包水位失效分析[J].中国水运(下半月),2018,18(1):99-101.
[12]代杰杰,宋辉,杨祎,等.基于深度信念网络的变压器油中溶解气体浓度预测方法[J].电网技术,2017,41(8):2737-2742.