CAJ | 학술논문

针对一维振动信号的压缩特点，提出一种改进的量子BP神经网络（IQBPN）信号压缩方法。本文根据一维振动信号的方差，将信号分为四个部分：平滑区、半平滑区、半边界区和边界区，从而可以选择不同的压缩比来对不同的区域进行压缩，以保持信号的丰富细节，保障压缩的质量。同时，利用量子BP神经网络的计算并行性和算法加速能力，从而提高了神经网络的收敛速度，缩短了压缩时间，为在线实时传输提供了一种新方法。实验证明，与其他方法相比，该算法在相同的压缩比时，可以提高信噪比，缩短运行时间。
침대일유진동신호적압축특점，제출일충개진적양자BP신경망락（IQBPN）신호압축방법。본문근거일유진동신호적방차，장신호분위사개부분：평활구、반평활구、반변계구화변계구，종이가이선택불동적압축비래대불동적구역진행압축，이보지신호적봉부세절，보장압축적질량。동시，이용양자BP신경망락적계산병행성화산법가속능력，종이제고료신경망락적수렴속도，축단료압축시간，위재선실시전수제공료일충신방법。실험증명，여기타방법상비，해산법재상동적압축비시，가이제고신조비，축단운행시간。
Aiming at compression characteristics of one-dimensional vibration signals,a signal compression method based on improved quantum BP neural network was proposed.Based on variance of a vibration signal,it was divided into four parts of smooth one,half-smooth one,half-boundary one and boundary one.For different areas,different compression ratios were chosen to keep many details and compression quality.Meantime,using parallel calculation and algorithm accelerating ability of quantum BP neural network,the convergent speed was improved and the compression time was shortened.The experimental results showed that compared with other methods using the same compression ratio,the proposed method can be used to improve SNR and shorten computing time.