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 Mill fault diagnosis signal data has the characteristics of limited resolution, periodicity, probability distribution redundancy and non-probability distribution redundancy. By constructing a new NN-based lossless data compression scheme for compressing the non-probabilistic distribution of rolling mill fault diagnosis signal data and selecting a suitable MRA-based coding system for eliminating the probability distribution of rolling mill fault diagnosis signal data, Finally, the NN-based lossless data compression scheme is embedded into the MRA-based coding system. The “NN and MRA-based rolling mill fault diagnosis signal data compression method” is achieved, which achieves the goal of more comprehensive elimination of rolling mill fault diagnosis signal data redundancy. Experiments show that the data compression method based on NN and MRA can effectively compress the rolling mill fault diagnosis signal data and the recovered signal data has higher quality