基于卡尔曼滤波的信息融合算法优化研究

通过比较采用联邦卡尔曼滤波的状态向量融合和量测信息融合,得出量测信息融合优于状态向量融合,因为只有当卡尔曼滤波一致时状态向量融合才有效.采用基于最小均方差估计的观测值加权融合法融合了多传感器数据,保持了观测向量的维数.这种方法具有高效性.为了提高该算法的速度和精度,对系统的量测空间进行了等价变换,而等价系统的状态空间却没有改变.给出了等价变换前后的系统误差方差阵和状态估计均一致性的证明.把矩阵分析中的L-D分解算法运用到该算法中以避免计算矩阵的逆,从而改善了算法的稳定性和精度.举例验证了所设计算法的这些优点,给出了采用联邦卡尔曼滤波和所优化滤波算法的状态估计和误差的仿真结果,并依次进行了分析.经过这种优化,算法的精度和速度得到很大提高,已经应用到实际工程中.     

SINS/GPS组合导航序贯滤波算法

针对SINS／GPS的速度、位置组合方式中，GPS接收机的位置测量误差相关的问题，提出了一种序贯形式的Kalman滤波算法。这种算法既具有计算量小的优点，又能克服由于GPS位置测量误差相关性带来的滤波性能变差的问题，实现对SINS／GPS组合系统信息的有效融合。通过仿真证明了该算法的有效性。     

Control of transient flow in irrigation canals using Lyapunov fuzzy filter‐based Gaussian regulator

An optimal fuzzy filter was applied to solve the state estimation problem of the controlled irrigation canals. Using linearized finite‐difference model of the open‐channel flow, a canal operation problem was formulated as an optimal control problem and an algorithm for gate openings in the presence of unknown external disturbances was derived. A fuzzy filter was designed to estimate the state variables at the intermediate nodes based upon measured values of depth at the points in the canal. A Lyapunov function was utilized as a performance index to formulate the fuzzy interference rules of the optimal fuzzy filter. A linear quadratic Gaussian (LQG) optimal controller for a multi‐pool irrigation canal was considered as an example. The state estimation problem in the controller was simulated using two techniques: Kalman estimator and the proposed fuzzy filter. The performance of the fuzzy state estimator designed using the Lyapunov fuzzy technique was compared with the results obtained using the Kalman estimator technique. The obvious advantages of the fuzzy filter were the lower computational costs and ease of implementation. The results of this study demonstrated that proposed Lyapunov‐type fuzzy filter provides both good stability and simplicity in the control of irrigation canals more than a Kalman filter. Copyright © 2005 John Wiley & Sons, Ltd.     

基于FTM算法的GPU加速

为了解决FTM(Front Tracking Method)算法在计算机中计算耗时长的问题,利用CUDA(Compute Unified Device Architecture)来实现FTM算法在GPU中的并行计算。结合GPU并行计算架构的特性以及FTM算法的特点,本文通过共享内存的引入、线程块划分和线程块共享内存边界元素的纳入、迭代方法的改进和迭代过程中存储结构的变换等方法,提出了将FTM算法中的网格计算以及界面标记点处理方法在GPU中的实现方式。最后,通过模拟单气泡在静止液体中的自由上升运动,验证了FTM在GPU中计算的可行性与计算效率的提升。     

A massively parallel GPU‐accelerated model for analysis of fully nonlinear free surface waves

We implement and evaluate a massively parallel and scalable algorithm based on a multigrid preconditioned Defect Correction method for the simulation of fully nonlinear free surface flows. The simulations are based on a potential model that describes wave propagation over uneven bottoms in three space dimensions and is useful for fast analysis and prediction purposes in coastal and offshore engineering. A dedicated numerical model based on the proposed algorithm is executed in parallel by utilizing affordable modern special purpose graphics processing unit (GPU). The model is based on a low‐storage flexible‐order accurate finite difference method that is known to be efficient and scalable on a CPU core (single thread). To achieve parallel performance of the relatively complex numerical model, we investigate a new trend in high‐performance computing where many‐core GPUs are utilized as high‐throughput co‐processors to the CPU. We describe and demonstrate how this approach makes it possible to do fast desktop computations for large nonlinear wave problems in numerical wave tanks (NWTs) with close to 50/100 million total grid points in double/single precision with 4 GB global device memory available. A new code base has been developed in C++ and compute unified device architecture C and is found to improve the runtime more than an order in magnitude in double precision arithmetic for the same accuracy over an existing CPU (single thread) Fortran 90 code when executed on a single modern GPU. These significant improvements are achieved by carefully implementing the algorithm to minimize data‐transfer and take advantage of the massive multi‐threading capability of the GPU device. Copyright © 2011 John Wiley & Sons, Ltd.     

带模型误差系统的自适应卡尔曼滤波在捷联惯导系统初始对准中的应用研究

对于有模型误差的惯导系统,采用常规卡尔曼滤波会导致较大的状态估计误差,甚至使滤波器发散。可采用自适应卡尔曼滤波算法,通过引入虚拟噪声,利用观测数据带来的信息,在线改进滤波器的设计,并将其运用到捷联惯导系统的初始对准中,由此得到的滤波估计比常规卡尔曼估计具有更高的精度和准确度。试验及计算机仿真结果验证了该方法的有效性。     

Performance comparison among some nonlinear filters for a low cost SINS/GPS integrated solution

The Kalman filter is a familiar minimum mean square estimator for linear systems. In practice, the filter is frequently employed for nonlinear problems. This paper investigates into the application of the Kalman filter’s nonlinear variants, namely the extended Kalman filter (EKF), the unscented Kalman filter (UKF) and the second order central difference filter (CDF2). A low cost strapdown inertial navigation system (SINS) integrated with the global position system (GPS) is the performance evaluation platform for the three nonlinear data synthesis techniques. Here, the discrete-time nonlinear error equations for the SINS are implemented. Test results of a field experiment are presented and performance comparison is made for the aforesaid nonlinear estimation techniques.     

A variational ensemble Kalman filtering method for data assimilation using 2D and 3D version of COHERENS model

Decoupled implementation of data assimilation methods has been rarely studied. The variational ensemble Kalman filter has been implemented such that it needs not to communicate directly with the model, but only through input and output devices. In this work, an open multi‐functional three‐dimensional (3D) model, the coupled hydrodynamical‐ecological model for regional and shelf seas (COHERENS), has been used. Assimilation of the total suspended matter (TSM) is carried out in 154 km² lake Säkylän Pyhäjärvi. Observations of TSM were derived from high‐resolution satellite images of turbidity and chrolophyll‐a. For demonstrating the method, we have used a low‐resolution model grid of 1 km. The model was run for a period from May 16 to September 14. We have run the COHERENS model with two‐dimensional (2D) mode time steps and 3D mode time steps. This allows COHERENS to switch between 2D and 3D modes in a single run for computational efficiency. We have noticed that there is not much difference between these runs. This is because satellite images depict the derived TSM for the surface layer only. The use of additional 3D data might change this conclusion and improve the results. We have found that in this study, the use of a large ensemble size does not guarantee higher performance. The successful implementation of decoupled variational ensemble Kalman filter method opens the way for other methods and evolution models to enjoy the benefits without having to spend substantial effort in merging the model and assimilation codes together, which can be a difficult task. © 2016 The Authors. International Journal for Numerical Methods in Fluids Published by John Wiley & Sons Ltd.     

Estimation of Wave Propagation using a Kalman Filter

An estimation method for gravity water wave propagation combining the finite element method and the Kalman filtering technique is presented. The method is further simplified to reduce the computational effort. According to the results obtained with this method, it is possible to estimate the water surface profile by taking into account the measured data, and to filter out the noise included in the measured data and the finite element approximation.     

Comparison of sequential data assimilation methods for the Kuramoto–Sivashinsky equation

The Kuramoto–Sivashinsky equation plays an important role as a low‐dimensional prototype for complicated fluid dynamics systems having been studied due to its chaotic pattern forming behavior. Up to now, efforts to carry out data assimilation with this 1‐D model were restricted to variational adjoint methods domain and only Chorin and Krause (Proc. Natl. Acad. Sci. 2004; 101 (42):15013–15017) tested it using a sequential Bayesian filter approach. In this work we compare three sequential data assimilation methods namely the Kalman filter approach, the sequential Monte Carlo particle filter approach and the maximum likelihood ensemble filter methods. This comparison is to the best of our knowledge novel. We compare in detail their relative performance for both linear and nonlinear observation operators. The results of these sequential data assimilation tests are discussed and conclusions are drawn as to the suitability of these data assimilation methods in the presence of linear and nonlinear observation operators. Copyright © 2009 John Wiley & Sons, Ltd.     

Application of Kalman Filter Finite Element Method and AIC

This paper presents a numerical simulation for application of the Kalman filter finite element method. The Kalman filter is employed frequently for the solution of time series analysis including observation and system noises. Applying the Kalman filter to the finite element method, the present method is capable of the estimation in time and space directions. In this method, the matrix generated by the finite element method is applied to the state transition matrix. Using the Kalman filter finite element method, the characteristics of both the Kalman filter and the finite element method can be strengthened. In this paper, the state transition matrix is based on the shallow water equations which are approximated by the finite element method. This method can estimate the tidal current not only in time but also in space directions.     

粒子滤波在惯导系统非线性对准中的应用

首先介绍了粒子滤波中自举滤波的原理和算法，说明该算法可用于处理非线性系统的状态估计问题。进而列出了捷联惯导系统速度误差方程和姿态误差方程，并将其用于惯导系统非线性对准。最后，通过对仿真结果的分析，指出通过结合粒子滤波和传统的扩展卡尔曼滤波，可以得到一种精度优于卡尔曼滤波，而计算量小于粒子滤波的非线性滤波方法。     

基于进化人工神经网络的组合系统滤波技术

在组合系统运用Kalman滤波器技术时,准确的系统模型和可靠的观测数据是保证其性能的重要因素,否则将大大降低Kalman滤波器的估计精度,甚至导致滤波器发散.为解决上述Kalman应用中的实际问题,提出了一种新颖的基于进化人工神经网络技术的自适应Kalman滤波器.仿真试验表明该算法可以在系统模型不准确时、甚至外部观测数据短暂中断时,仍能保证Kalman滤波器的性能.     

Portable data‐parallel surface reconstruction on a uniform rectilinear grid

With the increasing heterogeneity and on‐node parallelism of high‐performance computing hardware, a major challenge is to develop portable and efficient algorithms and software. In this work, we present our implementation of a portable code to perform surface reconstruction using NVIDIA's Thrust library. Surface reconstruction is a technique commonly used in volume tracking methods for simulations of multimaterial flow with interfaces. We have designed a 3D mesh data structure that is easily mapped to the 1D vectors used by Thrust and at the same time is simple to use and uses familiar data structure terminology (such as cells, faces, vertices, and edges). With this new data structure in place, we have implemented a piecewise linear interface reconstruction algorithm in 3 dimensions that effectively exploits the symmetry present in a uniform rectilinear computational cell. Finally, we report performance results, which show that a single implementation of these algorithms can be compiled to multiple backends (specifically, multi‐core CPUs, NVIDIA GPUs, and Intel Xeon Phi processors), making efficient use of the available parallelism on each. We also compare performance of our implementation to a legacy FORTRAN implementation in Message Passing Interface (MPI) and show performance parity on single and multi‐core CPU and achieved good parallel speed‐ups on GPU. Our research demonstrates the advantage of performance portability of the underlying data‐parallel programming model.     

基于渐消记忆滤波的1点RANSAC单目视觉姿态估计算法

针对1点RANSAC(Random Sample Consensus)单目视觉EKF(Extended Kalman Filter)算法中的滤波发散问题,分析了滤波发散的产生原因,提出了一种基于渐消记忆滤波的1点RANSAC单目视觉姿态估计算法。该算法通过在EKF滤波方程中引入加权因子,逐渐加大当前数据的权重,相应地减少旧数据的权重,有效地扼制了算法中的滤波发散问题。最后通过两组验证性实验验证说明了算法的有效性。实验结果表明:该算法能够有效地解决1点RANSAC单目视觉EKF算法中的滤波发散问题,具有更高的精度。第一组双轴联动实验,航向角的平均误差减小2.4158?,俯仰角平均误差减小0.1782?;第二组偏航轴大角度转动实验,摄像机航向角的估计误差一直保持在1.5?以内。     

基于新息自适应卡尔曼滤波算法的多类型结构响应重构

为改善传统卡尔曼滤波KF(Kalman filter)算法在过程噪声方差和测量噪声方差未知的情况下响应重构精度降低甚至发散的问题，提出了一种基于新息自适应卡尔曼滤波IAKF(innovation-based adaptive Kalman filter)算法的多类型响应重构方法。首先根据新息统计特性对卡尔曼滤波增益和状态估计误差协方差矩阵进行实时自适应调整；然后利用有限测点的加速度传感器的测量数据，结合模态法对结构各个位置的加速度、速度、位移以及应变进行响应重构；最后对起重机桁架和简支梁分别进行数值模拟和试验分析。结果表明，该方法能够有效地调整过程噪声方差并估计测量噪声方差，未测点的重构响应时程曲线与计算响应或测量响应时程曲线吻合良好。     

基于交互式多模型秩滤波的移动机器人组合导航算法

针对非结构化环境下移动机器人组合导航系统中存在的时变或非高斯噪声,将秩滤波器(rank Kalman filter,RKF)与交互式多模型算法(interactive multiple model filter,IMM)相结合,提出一种交互式多模型秩滤波算法(IMM-RKF)。秩滤波根据秩统计量相关原理确定采样点和权值,可适用于具有非高斯噪声的非线性系统;交互式多模型算法是解决结构和参数易发生变化系统中状态估计问题的重要途径,能够抑制组合导航系统中时变噪声引起的导航参数估计误差。仿真实验表明,相比于交互式多模型扩展卡尔曼滤波(IMM-EKF)和交互式多模型无迹卡尔曼滤波(IMM-UKF),提出的IMM-RKF算法能够提高组合导航系统姿态、速度和位置估计精度。     

基于新型滤波器-HABF的SINS传递对准仿真

传统动基座传递对准主要采用扩展卡尔曼滤波技术。但在动基座传递对准的非线性、非高斯条件下,这种基于模型线性化和高斯假设的滤波方法在估计系统状态及其方差时误差较大且可能发散。混合退火粒子滤波针对非线性、非高斯系统状态的在线估计问题,提出一种新的基于序贯重要性抽样的粒子滤波算法。在滤波算法中,用状态参数分解和退火系数来产生重要性概率密度函数,此概率密度函数综合考虑了转移先验、似然、噪声的统计特性以及最新的观察数据,因此更接近于系统状态的后验概率。实验仿真结果表明,这种基于混合退火粒子滤波器不仅比扩展卡尔曼滤波提高了传递对准的精度,而且又比传统的粒子算法减少了时间。     

基于UKF的INS/GNSS/CNS组合导航最优数据融合方法

为了提高INS/GNSS/CNS组合系统的导航精度,提出了一种基于UKF的多传感器最优数据融合方法。该方法具有两层融合结构,第一层中,GNSS和CNS分别通过两个局部UKF滤波器与INS组合,以并行的方式获得INS/GNSS和INS/CNS子系统的局部最优状态估值;第二层中,根据线性最小方差准则推导出一种矩阵加权数据融合算法,对局部状态估值进行融合,获取系统状态的全局最优估计。提出的方法无需采用方差上界技术对局部状态进行去相关处理,克服了联邦卡尔曼滤波(FKF)及其优化形式存在的缺陷。仿真结果表明,相比于FKF,提出方法的导航精度可至少提高36.4%;相比于UKF-FKF,其导航精度也可至少提高21.0%。