Design Of Unmanned Ship Detection System Based On Multi-sensor Information Fusion

It is an important task of urban maintenance to clean up the silt in urban watercourses.The key step for effective sludge removal is to accurately measure the distribution position and depth of the sludge,so as to accurately calculate the maintenance cost.Traditionally,manual measurement is the main method.This method not only requires a lot of manpower,but also has low efficiency,low accuracy and certain safety hazards.In recent years,unmanned ships have been gradually put into use,but unmanned ships are rarely used in mud detection.The main detection solution is to carry a single high-cost hardware sensor,which is relatively expensive.Moreover,the underwater environment is complex,and the anti-interference performance of a single sensor is poor,so the applicability for river silt detection is low.This subject independently designed and implemented a detection unmanned ship system,aiming at a series of problems of traditional detection methods and existing unmanned ship detection schemes,such as high detection cost and low accuracy of detection data.And the data collected by the unmanned ship is processed by the data fusion filtering algorithm.At the same time,the applied algorithm is improved to improve the filtering performance.Finally,based on the independently designed unmanned ship and related algorithms,the river channel detection experiment was carried out.The main work of this paper is as follows:(1)Aiming at the requirements of detecting unmanned ship system,a detailed analysis is carried out,and a variety of hardware modules are selected to independently design an unmanned ship detection device.The unmanned ship software system is designed by writing multiple scripts and programs.The unmanned ship realizes functions such as remote control mode,fixed-point cruising mode,remote control of the backstage of the unmanned ship,realtime transmission of collected data to the cloud,and PTZ video monitoring.(2)In order to improve the reliability and accuracy of the data,a variety of data fusion algorithms are applied and compared,and the track fusion filter algorithm is selected through the analysis of various data fusion algorithms.Secondly,in view of the problem that the motion state changes at any time in the remote control mode and the statistical characteristics of the process noise are difficult to estimate,which leads to the degradation of the filtering performance of the algorithm,the algorithm is improved to overcome the dependence on the statistical characteristics of the process noise and improve the estimation accuracy.Finally,on the established random motion state model of the unmanned ship,the algorithm is verified and compared with the standard Kalman filter algorithm and the traditional track fusion filter algorithm through simulation experiments.The results show that the algorithm has improved accuracy and stability,and it still has good track estimation performance when the process noise is unknown.(3)Aiming at the problem that the tracking filter algorithm using a single model in the fixed-point cruise mode is difficult to adapt to various motion states and will produce large errors or even divergence,an interactive multiple model filter algorithm is applied.And the algorithm is simulated and compared with the single model filtering algorithm.Secondly,in order to further reduce the filtering error and improve the reliability of the sensor,the interactive multiple model filtering algorithm is improved based on the idea of data fusion,and the track fusion filtering algorithm is used in the interactive multiple model algorithm.Through the simulation experiment verification of the unmanned ship’s uniform straight line and constant speed turning models,the experimental results show that the interactive multiple model filtering algorithm based on data fusion has higher accuracy and stability.Finally,the independently designed unmanned ship was used to carry out the field detection experiment in the fixed-point cruise mode,and the collected data was filtered and fused using the corresponding algorithm,and the data with higher accuracy was obtained.And use the processed data combined with the original depth(or ideal depth)of the river to draw a three-dimensional map of the underwater terrain.