Research On Fault-tolerant Fusion Estimation Method For Multi-sensor Time-delay System In Intelligent Ships

With the intelligent development of ships,the research on the related technologies has attracted the attention of many scholars at home and abroad.The Smart Ship Code,which was promulgated by China Classification Society in 2015,divides intelligent ship technology into six areas: intelligent navigation,intelligent hull,intelligent cabin,intelligent energy efficiency management,intelligent cargo management and intelligent integration platform.Among which the intelligent navigation refers to the analysis and processing of the perceived and acquired information by using computer technology and control technology,and then,designing and optimizing the route and the speed of the ship to realize autonomous navigation.In order to improve the safety of the autonomous navigation,it is necessary to ensure that the ship navigation status information acquired by the ship control system has high precision and credibility.Therefore,multiple sensors of the same or different types are usually used on the smart ship to observe the information.The multi-sensor fusion estimation module can estimate the navigation status information of the ship at the current time from the series of sensor observations,so it is a vital part of the intelligent ship navigation system.This paper focuses on the multi-sensor fusion estimation method in the intelligent ship navigation system.In the intelligent ship multi-sensor system,the types of sensors used and the install locations of the sensors are different,so the observation data of each sensor cannot be directly fused.At the same time,the original measurement of each sensor contains unknown measurement noise and random measurement outliers,which may reduce the accuracy of the final fusion estimation result.Aiming at the above problems,this paper proposes a data preprocessing method.Firstly,the measurement results of each sensor are time and spatially registered,so that the measurement results of each sensor are the navigation state values of the ship's center of gravity at the same time and in the same coordinate system.Then the designed sliding quartile filtering method is used to filter out the interference of measurement noise and outliers.In order to verify the effectiveness of the method,this paper builds a multi-sensor prototype system for intelligent ships,and based on the system,it carries out simulation experiments and model experiments.The results show that the designed data preprocessing method can reduce the measurement noise and the interference caused by the measurement of the wild value to a certain extent,which is beneficial to improve the accuracy of the final fusion estimation result.During the voyage of the intelligent ship,the unknown ocean environment interference,the interference between the sensors and the interference of the other equipment on the ship may cause intermittent fault signals of one or several sensors,thereby reducing the accuracy of the ship's navigation state estimation,even the estimated results will deviate from the true state of navigation of the ship,which ultimately threaten the safety of the navigation of the smart ship.In order to solve the above problems,the paper proposes a passive fault-tolerant fusion estimation method and an active one.The passive fault-tolerant fusion estimation method filters the observation data by adding a quartile filtering method in the sub-estimator,and designs the state prediction to resolve the time-delay problem of the quartile filtering method,and then uses the Kalman filtering method to estimate the local state.Finally,the covariance intersection fusion is performed in the fusion center.The simulation results show that the design method has good performance for intermittent abrupt faults,it also proves that the design method has high real-time performance.The designed active fault-tolerant fusion estimation method introduces the fading memory factor to reduce the influence weight of the old measurement data on the fusion estimation,and adds the fault detection and correction module to detect and correct the fault signal.The simulation results show that the proposed active fault-tolerant fusion estimation method has good fault tolerance and high fusion precision for intermittent abrupt fault and random constant deviation fault.As for the intelligent ship multi-sensor distributed system designed,because the estimated period of each sub-estimator is different,and it also different from that of the fusion center,the data transmitted by each sub-estimator to the fusion center is the ship motion at different times.When the fusion center performs the fusion estimation,the local estimation variables at the current time have not been transmitted to the fusion center,that is,the local estimation results are delayed to the fusion estimation center,which would cause the accuracy of the fusion estimation decrease,even lower than the local estimation accuracy of each sub-estimator.In this paper,an adaptive fusion estimation method is designed by adding the local estimation correction module,the local estimation value of the current fusion time is adaptively estimated based on the latest local estimation result and the one-step prediction result of each sub-estimator,and then the covariance intersection fusion is performed.The simulation results show that the proposed adaptive fusion estimation method has higher fusion estimation accuracy than the covariance intersection fusion estimation without time delay processing.In this paper,a series of researches on fault-tolerant fusion estimation methods for multi-sensor time-delay systems in intelligent ship navigation modules are carried out,which is beneficial to improve the safety of intelligent ship autonomous navigation and provide theoretical reference for the research of related technologies.