中文摘要：

提出了一种传感信号采集中的误差受控压缩算法．为适应传感信号特征多变的情况，根据各段信号的自相关系数动态调整梯度预测器的系数；通过改进最大步长均匀量化器降低量化噪声；采用Golomb-Rice编码算法对量化后的预测误差序列进行编码，根据数据采集系统前端噪声水平确定压缩误差参数的上限，进而获得压缩比的上限．算法在供水管道泄漏信号采集中的应用表明，压缩比达2．63时，压缩后重构信号漏点定位误差增加量小于0．2m.

英文摘要：

An error constrained data compression algorithm for sensing signal acquisition is proposed. To accord with the varying characteristics of sensing signal, the coefficients of the gradient predictor are dynamically adjusted relative to the correlation coefficients of data segments; the max-step uniform quantizer is improved to reduce quantization noise; and the Golomb-Rice coding algorithm is adopted to encode the quantized prediction residual sequence. The upper hounds of compression error and compression rate are determined in terms of the noise level of the system front end. The application of the designed algorithm in water supply pipeline leak signal acquisition shows that, at a compression rate of 2. 63, the additional leak location error of the reconstructed signals is less than 0. 2 m.