黃燕


摘 要: 針對當前復印機故障信號檢測提取方法中存在誤檢率高的問題,提出基于蟻群的復雜復印機故障信號的檢測與提取方法。基于蟻群的復雜復印機故障信號的檢測中,利用檢測某一路徑的最大代價和最小代價得到螞蟻于該路徑上所釋放信息素的濃度,以此計算蟻群對于某條路徑選取的概率。更新該條路徑上信息素濃度,按照路徑上的螞蟻存留的信息素濃度對復印機故障檢測過程中路徑選擇優先順序進行判斷,以檢測出復印機故障信號源。將復印機故障信號源代入小波包分析中,得到復印機總故障信號,計算故障信號中的各個頻帶信號相應能量,利用各頻帶相應能量,構建復印機故障信號特征向量。實驗結果表明,與當前方法相比,所提方法誤檢率最低約為0.3%,故障檢測準確性較高,檢測性能更為優越。
關鍵詞: 復印機; 故障信號; 信號檢測; 信號提取; 蟻群; 小波包
中圖分類號: TN911.23?34; TH165 文獻標識碼: A 文章編號: 1004?373X(2018)22?0103?03
Abstract: In allusion to the high error detection rate of the current fault signal detection and extraction method of the photocopier, a fault signal detection and extraction method based on the ant colony is proposed for the complex photocopier. During the ant colony based fault signal detection of the complex photocopier, the concentration of the pheromone released on the path by the ant is obtained by using the maximum cost and minimum cost of detecting a certain path, so as to calculate the selection probability of a certain path by the ant colony. The pheromone concentration on the path is updated. The path selection priority during the fault detection proces of the photocopier is judged according to the pheromone concentration retained on the path by the ant, so as to detect the fault signal source of the photocopier. The fault signal source of the photocopier is substituted into wavelet packet analysis to obtain the total fault signals of the photocopier. The corresponding energy of each frequency band signal in fault signals is calculated, which is used to construct the feature vector for fault signals of the photocopier. The experimental results show that, in comparison with the current method, the proposed method has a higher fault detection accuracy and better detection performance with a false detection rate of about 0.3% at minimum.
Keywords: photocopier; fault signal; signal detection; signal extraction; ant colony; wavelet packet
當今社會中,各種類型的復印機在各行各業中均有著十分廣泛的應用[1]。因復印機為光、機和電為一體的電子設備,它的集成化程度比較高,且內部結構復雜,在日常的運作中一旦產生故障,通常情況下非專業人員難以將其中的故障信號檢測出來[2?3]。由于復印機在工作中使用較為頻繁,在一定時期內會產生靜電等問題,這樣會導致與故障連接的其他位置也出現故障。綜上可知,復印機故障信號的檢測與提取成為了當前急需解決的問題。
劉洋等人提出基于RBF的設備故障檢測方法[4?5]。檢測過程中,先構建單個傳感器預測模型與任意兩個傳感器預測模型,其次利用上述兩個模型對任意一個傳感器預測值與任意兩個傳感器預測值進行計算,利用預測值和實際值間差值對傳感器的故障個數和位置等信息進行判斷。該方法檢測耗時較少,但誤檢率較高。王迪等人提出基于多信號流的設備故障檢測方法[6]。以多信號為基礎,引入故障先驗知識,得到多信號流故障檢測方案,利用引入故障概率改進多信號流檢測方案。將該方法應用于BEPCⅡ磁鐵電源控制設備故障檢測中,通過TEAMS測試工具箱實現該方法。此方法較為簡單,但也存在誤檢率高的問題。
上述方法不具備較為完善的性能,因此提出基于蟻群的復雜復印機故障信號的檢測與提取方法。
1.1 復印機故障信號檢測
在Matlab 2017上搭建實驗平臺,以圖1所示復印機作為實驗對象進行實驗。實驗過程中,分別使用不同方法對比的形式,驗證基于蟻群的復印機故障信號的檢測與提取方法有效性。實驗指標為設備故障檢測誤檢率。
分析圖2實驗結果:在額定的噪聲信號下,基于RBF的設備故障檢測方法誤檢率最低約為7.2%;基于多信號流的設備故障檢測方法誤檢率最低約為5.7%;基于蟻群的復印機故障信號的檢測方法誤檢率最低約為0.3%。通過數據對比可知,基于蟻群的復印機故障信號的檢測與提取方法誤檢率要低于當前方法。該結果主要是由于所提基于蟻群的復印機故障信號的檢測與提取方法在運行過程中,利用SVD理論對復印機故障中的噪聲信號進行去除,降低了復雜復印機故障信號檢測的誤檢率。
實驗結果如圖2所示。
鑒于當前設備故障信號檢測方法中存在的問題,提出基于蟻群的復印機故障信號的檢測與提取方法。過程中,利用SVD理論對復印機中的噪聲信號進行去除,通過蟻群算法對復印機故障信號進行檢測,采用小波包分析將檢測結果提取出來。實驗表明,該方法具有較強的可實踐性。
參考文獻
[1] 吳魁,王仙勇,孫潔,等.基于深度學習的故障檢測方法[J].計算機測量與控制,2017,25(10):43?47.
WU Kui, WANG Xianyong, SUN Jie, et al. Fault detection method based on the deep belief network [J]. Computer measurement & control, 2017, 25(10): 43?47.
[2] 蘆竹茂,王天正,俞華,等.基于紅外圖像分析的電力設備熱故障檢測技術研究[J].現代電子技術,2017,40(11):123?126.
LU Zhumao, WANG Tianzheng, YU Hua, et al. Research on electrical equipment thermal fault detection technology based on infrared image analysis [J]. Modern electronics technique, 2017, 40(11): 123?126.
[3] 崔芮華,王紹敏.基于多維特征量的航空串聯故障電弧檢測[J].科學技術與工程,2017,17(13):38?43.
CUI Ruihua, WANG Shaomin. Diagnosis of Arc faults in aviation AC system based on multi?dimensional features [J]. Science technology and engineering, 2017, 17(13): 38?43.
[4] 劉洋,歐文,盧贏,等.基于徑向基神經網絡的稱重設備傳感器故障檢測方法[J].傳感技術學報,2017,30(6):861?866.
LIU Yang, OU Wen, LU Ying, et al. Fault detection method for weighing equipment sensor based on radial basis function neural network [J]. Chinese journal of sensors and actuators, 2017, 30(6): 861?866.
[5] 夏輝麗,郭亞男,余發軍.基于稀疏分類算法的礦物傳送設備故障診斷方法[J].工礦自動化,2016,42(2):43?46.
XIA Huili, GUO Yanan, YU Fajun. Fault diagnosis method of mineral transmission equipment based on sparse classification algorithm [J]. Industry and mine automation, 2016, 42(2): 43?46.
[6] 王迪,劉佳,王巖峰,等.基于多信號流模型的電子設備故障診斷方法[J].強激光與粒子束,2017,29(7):99?103.
WANG Di, LIU Jia, WANG Yanfeng, et al. Modeling method of fault diagnosis of electronic device based on multi?signal flow [J]. High power laser and particle beams, 2017, 29(7): 99?103.
[7] 孫海燕.大型機械設備振動系統故障診斷仿真研究[J].科技通報,2016,32(6):88?92.
SUN Haiyan. Fault diagnosis simulation of large mechanical equipment vibration system [J]. Bulletin of science and technology, 2016, 32(6): 88?92.
[8] 謝世滿.機械設備振動信號采集對故障優化檢測仿真[J].計算機仿真,2017,34(6):419?422.
XIE Shiman. Simulation of fault detection for mechanical equipment vibration signal acquisition [J]. Computer simulation, 2017, 34(6): 419?422.
[9] 高迎平,李洋,常文韜,等.基于模糊動態故障樹的化工設備故障診斷方法研究[J].工業技術經濟,2017,36(4):48?54.
GAO Yingping, LI Yang, CHANG Wentao, et al. Research on failure diagnosis method of chemical equipment based on fuzzy dynamic fault tree [J]. Industrial technology & economy, 2017, 36(4): 48?54.
[10] 段佳雷.基于分段線性非飽和隨機共振的機械早期故障診斷方法研究[J].中國測試,2017,43(8):106?112.
DUAN Jialei. Study on incipient fault diagnosis of machinery based on piecewise linearity and unsaturated stochastic resonance [J]. China measurement & testing technology, 2017, 43(8): 106?112.