Structural Deformation/Vibration Monitoring: A Multipath Mitigation Algorithm Based on Wavelet Denoise

摘要

When using GPS to do structural deformation and vibration monitoring, the baseline is usually short, thus the error related to satellite, receiver and atmosphere can be eliminated by differential positioning method. Therefore, the main factor influencing the positioning is multi-path effect. For carrier phase relative positioning, multi-path effect would bring in 5mm influence. Worst of all, it has much influence on the fixing of ambiguity. In static survey, antenna is commonly located far away from reflectors to reduce multi-path effect. But for structural deformation and vibration monitoring, reflector is usually unavoidable. For example, in bridge structural deformation and vibration monitoring, the water, the suspension rope and high buildings around are the resource of multi-path effect, and they are not avoidable. Therefore, mitigation of multi-path effect is more important for structural deformation and vibration monitoring. As the GPS satellite position is periodic, multi-path effect in static observation is periodic, too. This multi-path effect can be called as systemic multi-path effect, and it can be treated as systemic background noise in structural deformation and vibration monitoring. Taking advantage of the periodicity of static observation multi-path effect, and using multi-day static observations to mitigate multi-path effect in structural deformation and vibration monitoring is effective to eliminate most systemic multi-path error. This paper puts forward a new multi-path mitigation method for structural deformation and vibration monitoring. Based on multi-path periodicity and correlativity of static multi-path and dynamic multi-path, it divides multi-path error into systemic and remained, then uses soft-threshold wavelet denoise method to reduce remained noise and outputs precision-improved result. The experiment suggests that it is effective and feasible for structural deformation and vibration monitoring. First, this paper validates the high correlation of multi-path effect in static observation and dynamic observation. Second, through long time static observation, we get systemic multipath noise. Third, in structural deformation and vibration monitoring, we eliminate systemic noise and discover and verify that the remained noise has normal distribution characteristic. Fourth, soft-threshold wavelet denoise method is utilized to eliminate the remained multi-path error, which improves the precision of structural deformation and vibration monitoring. Experiment shows that this method is feasible and effective. Experiment results shows that multi-path error in dynamic observation and multi-path error in static observation are periodically correlative. The correlativity is up to 84.1% in the experiment. As systemic multi-path error is periodic, it can be eliminated from the multi-path error in dynamic observation. Analysis on remained multi-path error indicates that the remained multi-path error is subjected to normal distribution. Soft-threshold wavelet denoise method presents a satisfying result. Moreover, we find that GPS receiver's measure precision is tightly correlative to vibration frequency and amplitude. When the vibration amplitude is more than 2mm and frequency is less than 1Hz, GPS receiver's measure precision is acceptable using our denoise method.