Post-processing Algorithm For Water Current Data Measured By Acoustic Doppler Velocimeter And Its Application

With the increase in global demand for food protein,various types of aquaculture production continue to increase,making a major contribution to global aquatic production.Due to the fast development of aquaculture,the pond culture,a momentous intensive aquaculture production facility,has rapidly expanded and created enormous economic benefits;In the process of freshwater pond culture,hypoxia has become the main reason for affecting the production of pond culture.Under hypoxic conditions,the feeding capacity and disease resistance coefficient of cultured organisms are affected,even lead to suffocation and death.The invention of impeller aerator has successfully improved the anaerobic situation of the pond,increased the feeding rate of fish,and greatly increased production capacity of culture pond.The aeration generated by the aerator has a significant impact on the flow in the pond.A reasonable layout can form a good cycle and facilitate the collection of pond sludge.For the optimal arrangement of aerators in the pond,flow velocity monitoring is a critical part and the Acoustic Doppler Velocimeter(ADV)is commonly used.ADV is susceptible to factors such as bubbles,suspended particles,turbulence intensity and other factors and result in spikes when recording data.Therefore,the post process of despiking doppler data before data analysis is of great significance for improvement of data validity and analysis accuracy of flow velocity distribution and calculation of friction coefficient at the bottom.In the present study,for the despike of outliers caused by bubbles and suspended sludge due to the use of aerator,A new method named the three-dimensional rousseeuw phase-space method(3DRPS)is constructed by optimizing phase space threshold method and robust estimation method through two-step filtering and three-dimensional simultaneous measurement,where the spikes close to the real value can be detected more accurately and the iterative process can be reduced more effectively.The stimulated data is contaminated under different conditions,statistical parameters like number of detected spikes are recorded and detection capability of each algorithm are analyzed,and the following research results were obtained:3DRPS method has a better detection capability.In terms of the correct detection rate(CDR),the CDR of 3DRPS method is significantly higher than the other two algorithms.For the data with less spikes,the CDR is 100%,for the data are highly contaminated,the CDR is higher than 97%.For continuous spikes,the average CDR of 3DRPS is 84.41%.,compared with the robust estimation method(55%)and the phase space threshold method(79.67%),the 3DRPS performs better.In terms of over processed rate(OPR),although some real data are judged as spikes by 3DRPS method,compared with phase space method,OPR is effectively reduced,and the more spikes,the lower the over processed rate.The combination of robust estimation principle and the moving window is used for the first filter,a large number of spikes in sample data are eliminated and no real data are misjudged as spikes so that the mean value and standard deviation of the sample data are not affected.After the first filter,some spikes similar to the real data remains in sample data which accords with the characteristics of phase space threshold method.The phase space threshold method is used for second filter,then the phenomenon of over process is effectively reduced.3DRPS method has the best comprehensive capability and is suitable for highly contaminated data.it is able to improve the detection accuracy and reduce the over process phenomenon caused by iteration.Compared with robust estimation method and phase space method,3DRPS has the characteristics of high detection accuracy,less over treatment phenomenon and wide application range.Comparison of the kurtosis coefficient and skew coefficient of data after detection of three methods shows the 3DRPS method approaches to the largest number of kurtosis and skewness array of the clean data.19 of the 21 groups of data are consistent with the kurtosis coefficient of the clean data,and 20 groups of data are consistent with the skewness coefficient of the clean data,which shows the 3DRPS method has a significant effect on the treatment of pollution data and a higher degree of data reduction.Comprehensive analysis shows that compared with robust estimation and phase space threshold method,3DRPS effectively improved the detection accuracy of the algorithm while avoiding iteration and reducing the over processed rate.It is suitable for highly contaminated flow velocity data,and can be effectively used for the postprocessing of the flow velocity monitoring in the aquaculture pond,providing accurate flow velocity estimation for the subsequent research of the flow velocity distribution and pond water in the bottom friction velocity Data.