Wang Zhisheng

Federated filter based multi-sensor fault-tolerant altitude determination system for UAV

An altitude determination system for unmanned aerial vehicle (UAV) is presented in this paper, in which a multi-sensor measurement system is designed by using barometric altitude sensor, GPS, and accelerometer, and the UAV altitude is determined by the federated filter based fusion estimation method. The measurement information fusion problem of the sensors with different output frequency is resolved by designing the subsystems respectively. Calculation method for initial value is presented based on the information fusion theory. The re-initialization time after fault isolation of the system is reduced greatly. The simulation results show that the altitude determination system is characterized by high precision, good reliability, and fast fault restoration.

Brain emotional learning based adaptive identification method for nonlinear dynamic system

A novel online identification by brain emotional learning (BEL) for nonlinear dynamic system is proposed in this paper. BEL is a bionic computational model which is inspired by the emotional learning process in the amygdala-orbitofrontal system of the mammalian brain. Online system identification scheme based on BEL is built up, in which BEL is applied to realize the identification of the system in-output by adjusting the learning weights in A-O system adaptively. Simulation results of the system identification of a certain one-axis simulator show that BEL is characterized by fast learning speed and high identification precision.

Design of Optimal Tracking Controller for Nonlinear Discrete System with Input Delay Using Information Fusion Estimation Method

Based on the theories of information fusion estimation, information fusion optimal tracking control algorithms of nonlinear discrete system with input delay are proposed. All information with respect to control strategy, including ideal control strategy, expected object trajectory, and system dynamical equation are regarded as measuring information of control strategy. Therefore, the question of optimal control can be transferred into one of information fusion estimation, and the algorithms of optimal tracking control of time-delayed nonlinear discrete-system are given in detail. In the end, the mathematical simulation results show the validity of the algorithms presented in this paper.

Information fusion based optimal decentralized control for stochastic largescale system

In allusion to the control problem of the complex stochastic largescale system, a novel decentralized control method based on information fusion estimation is proposed in this paper. Based on the information fusion optimal estimation theorem, by fusing the one-step predictive information of the largescale system state and the sub-system measurement information, the optimal estimation of the largescale system state is derived. Moreover, by fusing the state estimation information, the co-state estimation information and the soft constrained control energy information, the optimal estimation of the control station output is obtained.

Information fusion based filtering for multi-sensor system

In allusion to the state estimation problem of the multi-sensor system, two filtering algorithms based on information fusion estimation theory are presented, which called the measurement fusion filtering and the state fusion filtering. The former is based on the idea of fusion first and then filtering. It fuses the sub-systems measurements information to obtain the system measurement estimation, and then fuses the system state predictive information to obtain the state estimation. The latter is based on the idea of filtering first and then fusion. It fuses the predictive information and the measurement information of the sub-systems states to obtain the sub-systems states estimation, and then fuses all sub-systems states estimation information to obtain the system state estimation. The former filtering is same with the centralized fusion filtering, while the latter filtering is different, because of the different fusion information. The performance of the proposed filtering methods depends on the utilized information weight.

Learning method of RBF network based on FCM and ACO

A new learning method of radial basis function (RBF) network based on fuzzy C-means (FCM) clustering and ant colony optimization (ACO) is presented in the paper. Generally, the crucial problem of learning RBF network is the selection of the unit number, the centers and widths of the Gaussian radial basis function in hidden layer, and the weights between hidden layer and output layer. In this work, the centers are determined by FCM clustering, and the weights are determined by least mean squares (LMS) method. In addition, ACO is introduced to optimize two important parameters of FCM including the weighting exponent which impacts its performance, and the clustering number that is equal to the hidden unit number which impacts the generalization of RBF network, by minimizing an objective function integrated the generalization error with the hidden unit number of RBF network. Simulation results of identifying a nonlinear system illustrate the effectiveness of designing the RBF network with smaller structure but stronger generalization ability, comparing with K-means and orthogonal least squares (OLS) based learning methods.

Multi-sensor Information Fusion for Aircraft Attitude Determination System

A multi-sensor information fusion based aircraft attitude determination system is investigated. Firstly, A scheme of the aircraft attitude determination system including multiple sensors is designed, in which gyroscope measures the angular velocity, the strap-down accelerometer measures the acceleration, the magnetic heading meter measures the yaw angle, and the GPS measures the velocity. Secondly, the measurement models of the multiple sensors are described. Thirdly, an information fusion filtering algorithm which is equal to Kalman filtering is derived based on information fusion estimation theory, and is used to obtain the attitude estimation and the velocity estimation of the aircraft by fusing all the measurement information. Finally, the simulation results show the feasibility and the effectiveness of the proposed aircraft attitude determination system.

Information fusion decoupling compensator for time-delayed MIMO system

In allusion to the strong coupling problem of the time-delayed multivariable linear system, a novel decoupling compensator is designed based on information fusion estimation. For the multi-input and multi-output (MIMO) system, the decoupling compensators of the main channels and their coupling channels are designed separately. Firstly, the controller is disconnected, the input ends of the decoupling compensators are input the set control signals, and then the desired outputs of the coupling channels are zeros. By fusing the soft constraint information of the desired outputs and the control energy of the coupling channels, the optimal decoupling compensation variables under the quadratic performance index can be obtained. Furthermore, the approximate optimal decoupling compensation variables with easier computation can be obtained, when the infinite preview desired outputs are equivalently treated as one-step preview desired outputs. Simulation results show that the decoupling compensator has advantages of small computation and good decoupling quality.

A New Kind of Satellite Integrated Attitude Determination System

An extended federated filtering (EFF) algorithm is developed which can be applied to integrated navigation system and integrated attitude determination system. According to both information fusion theories and information conversation principle, it is proposed that information allocation is the inverse process of information fusion in EFF algorithm. The information allocation theorem is given and proved, which lays a theoretical foundation for EFF. Taking a certain satellite integrated attitude determination system as an example, the realization of EFF is discussed in detail. The theoretical analysis and simulation results show that EFF algorithm is less computation than present federated filtering algorithm, and EFF can guarantee both local optimum and global optimum, while present federated filtering can only guarantee global optimum.