基于NN和MRA的轧机故障诊断信号数据压缩研究

轧机故障诊断信号数据具有有限分辨率、周期性、概率分布冗余和非概率分布冗余等特点。通过构造一新的基于NN的无损数据压缩方案用于压缩轧机故障诊断信号数据的非概率分布冗余空间 ,然后选择一合适的基于MRA的编码系统用于消除轧机故障诊断信号数据的概率分布冗余 ,最后把基于NN的无损数据压缩方案嵌入基于MRA的编码系统 ,获“基于NN和MRA的轧机故障诊断信号数据压缩方法” ,达到较全面消除轧机故障诊断信号数据冗余的目的。实验证明基于NN和MRA的数据压缩方法能有效压缩轧机故障诊断信号数据 ,且恢复的信号数据有较高质量

NN and MRA Based Data Compression for Rolling Mill Fault Diagnosis Signal

Rolling mill fault diagnosis signal have bright characteristics such as finite resolution, periodicity, probability distributing redundancy , non-probability distributing redundancy and etc. NN based new lossless data compression is constructed to eliminate non-probability distributing redundancy of rolling mill fault diagnosis signal. Then, MRA based appropriate coding system is chosen to compress the probability distributing redundancy space of rolling mill fault diagnosis signal.Finally, NN based new lossless compression is embeded in MRA based coding system to obtain a NN and MRA based data compression for rolling mill fault diagnosis signal, thus redundancy of rolling mill fault diagnosis signal is fully eliminated. Experiments show that NN and MRA based data compression is effective, with little distortion of recovery signal and application to rolling mill fault diagnosis signal is successful.