Research On Multi-Sensor Data Fusion Of Dynamic Positioning Ships

Dynamic positioning ships have important applications in marine resources exploration,oil exploitation,marine rescue and so on.It is a powerful tool for people to enter the ocean.In order to enable the dynamic positioning ship to accurately locate during operation,multiple position reference systems are generally adopted.How to calculate accurate and reliable fusion values from multiple measurement values has always been a hot research issue.This paper takes the position reference system of dynamic positioning ship as the research object,realizes the position information fusion by using federated filter,and improves the fault tolerance and fault recovery ability of federated filter structure by combining the fault detection algorithm and federated filter in view of the sensor failure caused by external factors.Firstly,the ship motion model is established to lay a foundation for subsequent research.The sensor data sources are obtained by modeling and simulation of the used position reference system,and the measurement models of tensioning cable,acoustic and DGPS are established according to their respective characteristics.Select a federated filter with reset as the basis of your design.The SRCKF method is used to design and implement the sub-filter for each position reference system.Then,by analyzing the influence of the information allocation algorithm on the overall structure,the adaptive information allocation factor composed of the mixed coefficient matrix is designed by combining the error coefficient matrix and the residual coefficient matrix.The simulation results show that the fusion estimation error is better than that of any single sensor,which improves the accuracy of the measurement results of the position reference system.Secondly,an effective fault detection,isolation and recovery algorithm is developed for possible sensor faults in consideration of the complex and changeable ocean environment that dynamic positioning ships have to face during operation.In order to effectively deal with abrupt and gradual faults,a residual Chi-square detection algorithm based on data fusion was designed by combining the residual Chi-square detection method with data fusion.Based on this,the residual Chi-square detection based on data fusion and RBF neural network are discussed as fault detection,isolation and recovery algorithms of the system,which can ensure the stable and orderly operation of the system when a subsystem fails.In order to accurately reflect the state of each subsystem,part of the results of the above two algorithms are introduced into the federated filter for information distribution and fusion,so as to obtain the optimal fusion value.Finally,based on the established simulation environment,the fault-tolerant federated filters designed by two different fault detection methods are simulated and verified.The results show that the two algorithms can achieve the effect of fault detection,isolation and recovery well.The comparative analysis of the two methods shows that the fusion accuracy of the residual Chi-square detection algorithm based on data fusion is higher,while the RBF neural network algorithm is slightly faster in response to the measurement drift.