| The development of marine technology is a powerful way to achieve maritime power,and has important practical significance.The Multibeam echo sounding system(MBES)plays an important role in resource exploration and development,maintaining marine ecology,and formulating marine protection and management policies by providing high-precision digital terrain models(DTMs)of the seafloor.As a disruptive new payload,unmanned marine platforms have led to a transformation of the conventional operating mode of MBES,achieving a transition from “manned” to “unmanned”.Under this background,this paper focuses on the development trends and practical needs of MBES based on unmanned marine platforms.Also,focusing on providing a comprehensive and systematic quality control plan for MBES.With the support of the National Key R&D Program Major Scientific Instrument and Equipment Development Special Project "Self-Organizing Networked Marine Environment Multi-Parameter Measuring Instrument",in-depth research is conducted on three aspects: array calibration method of MBES,depth threshold tracking method of MBES,depth data evaluation method and anomaly detection.Firstly,research was conducted on the array calibration method for MBES.In order to ensure the amplitude and phase consistency of the array,and to possess accurate measurement capabilities in theory,a calibration algorithm was developed to ensure that high-precision seabed terrain measurement results could be obtained when the system is deployed on unmanned platforms.The calibration algorithm research was carried out in the following aspects: First,the impact of amplitude and phase consistency on Direction of Arrival(DOA),beamforming,and array gain was analyzed using a near-field focused beamforming model.Secondly,we established a parameter-solving model by utilizing the known precision characteristics of the turntable rotation angle as prior information.This allowed us to fine-tune,evaluate,and determine parameters without the need for iterative solutions.Different sets of data were optimized using the Root Mean Square Error(RMSE)method.At last,the Direction of Arrival(DOA)estimation performance of the proposed method was analyzed using the Cramer-Rao Lower Bound(CRLB),and the results showed that the DOA estimation error of the calibrated array was approximately 0.01° with a signal-to-noise ratio of20 d B.Tank test demonstrated that the DOA estimation accuracy and sidelobe level of the array were significantly improved after calibration,and the overall DOA estimation deviation of the calibrated array was reduced by approximately 0.1°.Secondly,the depth gate tracking method for multibeam echosounders was studied.In the context of the system being deployed on unmanned platforms,gate tracking method is a problem that must be faced and solved to realize autonomous detection.As MBES gradually transitions from manned to unmanned,gate tracking plays a crucial role not only as a precursor to the outlier detection algorithm but also as the foundation for automatic adjustment of other parameters.To investigate the depth gate tracking method that examines multiple historical pings of data,we focused on three aspects.Firstly,we established a similarity distance model based on one-sided Hausdorff distance to extract effective depth data from historical data with similarity.To overcome its sensitivity to outliers,the Order Statistic Constant False Alarm Rate(OS-CFAR)threshold processing scheme was adopted to process the abnormal results in the nearest neighbor sequence and source data,which can suppress potential outlier in historical data while obtaining stable similarity evaluation results.Secondly,to achieve registration of historical data with similarity,the Iterative Closest Point(ICP)registration method was combined with the similarity distance model to solve the problem of high initial position requirements in the ICP registration algorithm,avoiding rough registration steps.Finally,the registration result of historical data was used as prior information to initialize the particle distribution,and the likelihood function was designed to assign reasonable likelihood values to particles,achieving depth gate estimation by tracking effective terrain in depth measurement results.The effectiveness of the proposed algorithm was verified by comparing processing results of depth data with different qualities using measured data.The processing result of continuous multi-ping soudings showed that the performance of the algorithm remained stable even when cluster outlier values were dense.Thirdly,we studied methods for quality assessment and outlier detection of depth measurement data.To ensure effective system measurement and improve the real-time data quality of depth measurement results in unmanned situations,we conducted a study in three aspects.Firstly,we used a Quality Factor(QF)to examine intermediate data,including beam and phase data,in the system’s real-time processing process.This allowed us to use quantifiable indicators to evaluate the quality of each sounding point and reduce the heavy burden of post-processing.This provides data support for the selection of optimal results for seafloor detection while examining measurement effectiveness.Secondly,to address the problem that the quality factor cannot examine the depth consistency between sounding points,a Quality Factor Model based on Forcesting Error(QE)was established.The sliding window method was used to sort the depth results in the window by a series of processing to predict the depth estimation residual and obtain the corresponding quality factor.After analyzing the sliding window and threshold rules,the effectiveness of the proposed method was verified through simulations and experiment.Finally,we focused on the problem of depth estimation of complex seafloor topography under the condition of dense distribution of cluster outliers,and proposed a quadratic detection depth estimation algorithm based on M-estimation.We re-designed the historical window and sample selection rules in the area of M-estimation failure to increase the effective sample ratio,and established a multiple estimation model based on Bayesian factors to identify suspicious targets in depth measurement results.Processing of experiment data shows that this method has more stable performance than M-estimation when estimating the depth of complex seafloor topography,and can effectively identify suspicious seafloor or targets with continuous features.Finally,we developed an unmanned autonomous MBES prototype based on practical application needs.Based on the key technical performance indicators,the system composition was determined.Embedded software design and implementation were carried out around the parallel signal processing platform.In the seafloor detection algorithm part,a DSP design scheme based on TMS320C6657 was proposed and implemented,combined with the proposed quality control and depth threshold tracking techniques.The measurement performance test,data link verification test,sunken ship target measurement test,and two unmanned surface vessel cooperative measurement tests were successfully conducted in the field.The effectiveness of the joint processing scheme of gate tracking method and outlier detection method was validated through offline playback experiments. |
PDF Full Text Request