Case 1: the zero data is not zero, but a constant deviation from a certain value;

Case 2: the zero data in a certain area of the random change, and the smooth random process is dominated;

Case 3: the zero data in addition to the random change, but also contains the trend term of the long-term change over time.

The first case, you can zero flow in the case of a one-time correction, you can basically eliminate the so-called "zero drift" effect.

The second case, should contain "zero drift" of the measurement data for a large amount of data averaging, and statistical values as the results of the measurement. Data averaging can greatly weaken the random characteristics caused by the "zero instability", but at the same time will affect the real-time ultrasonic water meter measurement.

The third case, the data volume homogenization process can only eliminate the impact of random variation, but the time drift caused by the trend term can not help. Therefore, there are only two choices for this situation, the first is to choose a better performance of precision timing chips and components, and the second is to compress the measurement range of the ultrasonic water meter, so that the zero-drift trend term on the measurement of the degree of influence to the minimum point.