基于改進(jìn)PCNN的數(shù)據(jù)降噪方法
中國測試王建國， 閆海鵬， 張文興， 張鑫禮
摘 要：為去除數(shù)據(jù)中存在的噪聲點(diǎn)，提高數(shù)據(jù)質(zhì)量，提出一種基于改進(jìn)PCNN的數(shù)據(jù)降噪方法。該方法在無耦合鏈接的簡化PCNN模型基礎(chǔ)上，改進(jìn)閾值函數(shù)，添加記錄神經(jīng)元是否點(diǎn)火的矩陣以及點(diǎn)火時(shí)間矩陣，根據(jù)神經(jīng)元初次點(diǎn)火時(shí)間辨識并去除噪聲點(diǎn)，從而實(shí)現(xiàn)數(shù)據(jù)降噪。實(shí)驗(yàn)測試結(jié)果表明：該算法能夠有效濾除數(shù)據(jù)中的噪聲點(diǎn)，很好地保持原始數(shù)據(jù)的特征。
關(guān)鍵詞：數(shù)據(jù)降噪；改進(jìn)PCNN模型；閾值函數(shù)；點(diǎn)火時(shí)間矩陣
文獻(xiàn)標(biāo)志碼：A 文章編號：1674-5124（2016）01-0092-04
Data noise reduction method based on modified PCNN
WANG Jianguo， YAN Haipeng， ZHANG Wenxing， ZHANG Xinli
（School of Mechanical Engineering，Inner Mongolia University of Science and Technology，
Baotou 014010，China）
Abstract: To remove the noise points in the data and improve the quality of data， a data noise reduction method based on modified PCNN is presented. In this algorithm， threshold function has been improved and a matrix which can show recorded neurons firing or not and a matrix of ignition time are added， based on the simplified PCNN model of non coupling linking. The noise points are identified and removed by the first ignition time of neurons. Thus the data noise reduction is achieved via the method. The experimental results show that the algorithm can effectively filter out the noise points in the data， and remain the characteristics of the original data.
Keywords: data noise reduction； modified PCNN model； threshold function； ignition time matrix
