Research On Lossless Compression And Feature Extraction Of Multi-Channel Acquisition Data

LZW(Lempel-Ziv-Welch)compression algorithm has an important application value in real-time acquisition and wireless transmission.Generally,it adopts the acquisition-compression-transmission working mode,the higher compression ratio in this mode can greatly reduce the pressure on wireless transmission.But in the case of fast acquisition speed,low data transmission bandwidth and limited hardware resources,it can easily lead to a problem that the compression rate is not high or the speed of acquisition is mismatched with the compression speed if the signal which has a sampling point of uniform probability distribution is compressed.To this end,an improved LZW prefix encoding scheme based on position information is proposed in this paper.Firstly,the improved compression algorithm maps the sampling points based on compression ratio factor,so that it can identify the compression condition of adjacent sampling points.Secondly,through the position information between the sampling points,the code length of the sampling points is shortened,so the data of the sampling points is compressed.Experiments show that,compared with the original LZW compression algorithm,the improved algorithm can increase the compression ratio by 26.25% without incr easing the complexity and hardware storage space.Therefore,the effectiveness of the algorithm in the acquisition system is proved.In real-time acquisition and wireless transmission,the extraction of signal amplitude and frequency characteristics is an important part of acquisition signal analysis.But in the case of fast acquisition speed and limited hardware resources,how to quickly predict the abnormal signal situation is an important research content in the experimental process.To this end,an improved K-means algorithm based on K-cluster is proposed according to the characteristics of rising and falling edges of signal waveform in this paper.Firstly,for amplitude information,based on the characteristics of waveform rising edge and falling edge,the algorithm divides signal waveforms into several groups(called clusters in this paper),and obtains the center value of clusters.According to the cluster center value,whether the signal amplitude is abnormal or not is judged.Secondly,for frequency information,signal frequency characteristics are extracted from the host computer.Through this improved algorithm,signal frequency information is divided into several clusters,and clus ter center values are obtained.Cluster center values are used as prediction parameters,and combined with the prescribed model,signal frequency prediction isrealized.Experiments show that,the amplitude and frequency characteristics extracted by the algorithm are consistent with the original signal characteristics.