A Post-Resampling Based Particle Filter for Online Bayesian Estimation and Tracking

In order to effectively monitor condition and detect fault types of high nonlinear system, and extract the features of system state under strong noise background, this paper proposes a novel fault detection and diagnosis (FDD) method based on modified particle filter (PF). The artificial neural network is incorporated in PF for adaptively adjusting weight of particle. In the modified PF, the large weight particles are split into several small weight particles, the particles with smaller weight is adjusted by using artificial neural network. By which the particles in the low probability density region are adjusted to the high probability density region, and the problem of particle leanness is solved effectively. Moreover, this paper also uses time-varying auto regressive (TVAR) and Akaike information criterion (AIC) methods to establish state space model for state estimation. Finally, the proposed method is implemented for fault diagnosis on a roller bearing. Good results are obtained, and the bearing faults, such as the outer race, the inner race and the roller element defects, have been effectively discriminated.

In this paper, a novel fuzzy quadrature particle filter (FQPF) based on maximum entropy fuzzy clustering for maneuvering target tracking is proposed. The novelties of the fuzzy quadrature particle filter are in the update step in which the predicted and posterior probability density functions are approximated by introducing a set of quadrature point probability densities based on the Gauss–Hermite quadrature rule as a Gaussian. The particle and quadrature point weights can be adaptively estimated based on the weighting exponent and fuzzy membership degrees provided by a modified version of maximum entropy fuzzy clustering algorithm. Unlike the Gaussian particle filter (GPF) using the prior distribution as the proposal distribution, the new FQPF uses a set of modified quadrature point probability densities as the proposal distribution that can effectively enhance the diversity of samples and improve the approximate performance. Finally, simulation results are presented to demonstrate the versatility and improved performance of the fuzzy quadrature particle filter over other nonlinear filtering approaches, namely the unscented Kalman filter, quadrature Kalman filter, particle filter, and GPF, to solve maneuvering target tracking problems.

The interacting multiple model (IMM) algorithm is a widely accepted state estimation scheme for solving maneuvering target tracking problems, which are generally nonlinear. During the IMM filtering process, serious errors can arise when a Gaussian mixture of posterior probability density functions is approximated by a single Gaussian. Particle filters (PFs) are effective in dealing with nonlinearity and non-Gaussianity. This work considers an IMM algorithm that includes a constant velocity model, a constant acceleration model and a 3D turning rate (3DTR) model for tracking three-dimensional (3D) target motion, using various combinations of nonlinear filters. In existing literature on combining IMM and particle filtering techniques to tackle difficult target maneuvers, a PF is usually used in every model In comparison, simulation results show that by using a computationally economical PF in the 3DTR model and Kalman filters in the remaining models, superior performance can be achieved with significant reduction in computational costs.

State and parameter estimation based on a modified particle filter for an in-wheel-motor-drive electric vehicle

Localization is one of many issues in mobile robot study. Localization is an essential ability for mobile robot to determine its location, so that it can plan a movement and go to a desired location. The mutual method for mobile robot localization is using a particle filter. High computation needs in particle filter is one of problem in particle filtering to get accurate location. The paper proposes a low computational mobile robot localization using a particle filter. It uses two methods: local localization using a dead reckoning method and global localization using a landmark-based vision sensor. Simulation results show that the proposed method provides a good estimation on the mobile robot position and orientation.

State and parameter estimation of nonlinear stochastic systems by evolution strategies based particle filter

To improve the filtering precision when dealing with the state estimation problem of nonlinear/non-Gaussian systems, we propose a novel sequential quasi-Monte Carlo (SQMC) filtering algorithm which is analogous to the sequential Monte Carlo (SMC) or particle filtering methods. The central idea of the new algorithm is to apply one of the deterministic sampling methods, i.e., number theoretic sampling method to SQMC. The point set of uniform distribution generated by cyclotomic field can construct more uniform scattered points in unit cube. Therefore, random samples generated by the point set of uniform distribution can adequately describe the posterior probability density function (PDF). Simulation results show that the proposed filtering algorithm provides better performance in nonlinear/non-Gaussian state estimation when compared to classical particle filter, SQMC using Halton sequence in presence of severe nonlinearity.

We propose a sequential Monte Carlo (SMC) method to efficiently and accurately compute cut-Bayesian posterior quantities of interest, variations of standard Bayesian approaches constructed primarily to account for model misspecification. We prove finite sample concentration bounds for estimators derived from the proposed method along with a linear tempering extension and apply these results to a realistic setting where a computer model is misspecified. We then illustrate the SMC method for inference in a modular chemical reactor example that includes submodels for reaction kinetics, turbulence, mass transfer, and diffusion. The samples obtained are commensurate with a direct-sampling approach that consists of running multiple Markov chains, with computational efficiency gains using the SMC method. Overall, the SMC method presented yields a novel, rigorous approach to computing with cut-Bayesian posterior distributions.

To localize mobile target timely and improve the accuracy of the target state estimation in wireless sensor networks, this paper presents a novel mobile target localization algorithm. By using the mean shift algorithm to construct the proposal distribution to make use of the current measurements, and weighting particles according to the virtual hamming distances in the particle filter, this proposed algorithm improves the accuracy of the target state estimation and reduces the necessary number of samples. It has lower computation and communication cost than the conventional particle filter because of more accurate state estimate of samples. Extensive simulation results confirm that this localization scheme outperforms conventional particle filter and its localization accuracy is comparable to the unscented particle filter, but its computation cost is 50% less than that of UPF.

Parallel Kalman filter group integrated particle filter method for the train nonlinear operational status high-precision estimation under non-Gaussian environment

This paper deals with an efficient tracking of a ballistic target by using certain measurements from radar. An efficient non-linear model for the target along with observed error is developed. Since different targets need different models, a specific target with known properties is chosen. Here the target chosen is 9000 mm air launched ballistic missile. This generally weigh more than 5000 kg and its velocity is 2000 m/s. Since these missiles are highly accurate, a 2-D space is chosen as its path. The radar gives the range and the angle of elevation of the missile. The input data processed by state approximation is called as state estimation. Particle filter is used for this non-linear model. Here the observed noise, the processed noise and the radar noise are taken into account. The performance of particle filter is tested and verified with the simulation. By using this particle filter, the range and altitude of this ballistic target can be predicted in advance. The main reason of particle filter’s popularity is that it is very flexible and adaptive. In practical, all non-linear systems has accurate filters.

This paper develops a stochastic gradient-based modified particle filter algorithm for an auto regressivee xogenous (ARX) model with nonlinear communication output submodel. The outputs of the ARX model are transmitted over a nonlinear communication network, while the outputs of the communication network are available. Based on the modified particle filter and the available outputs, the outputs of the ARX model can be computed, and then the unknown parameters can be estimated by the stochastic gradient algorithm. The simulation results demonstrate that the stochastic gradient-based particle filter algorithm is effective.

This paper is concerned with the problem of recursively estimating the kinematics of a target moving in underwater environments. The measurements are noise-corrupted time differences of arrival (TDOA) of the signal emitted from the source to pairs of spatially separated sensors. The errors associated with these measurements are very small so standard nonlinear filtering techniques, such as particle filter, can not be directly applied. The reason is that the likelihood function may concentrate in a region of the state space that is too small to contain samples of the particle system. We apply the improved approximation based on progressive correction principle to efficiently handle this difficult problem and describe a new approach to finding an optimal sequence of fictitious matrices to be used in the correction substeps. The performance of the proposed filter is tested using Monte Carlo simulations and the resulting mean-square errors in the positition and velocity are compared to their theoretical Cramer-Rao lower bounds.

Sequential importance sampling for structural reliability analysis

Fault diagnostic monitoring in nonlinear hybrid processes requires dedicated techniques being capable of dealing with both nonlinearity and hybrid dynamic characteristics. This paper introduces a modified particle filtering estimation-based methodology for online diagnostic purposes by individual tracking of the most likely faulty modes and the process states in a nonlinear complex continuous stirred tank reactor (CSTR) process plant for start-up operation. A receding horizon nonlinear model predictive controller (NMPC) has been designed and installed on the CSTR process plant using a genetic algorithm (GA) to generate optimal discrete and continuous control inputs so as to drive the process operation from an initial starting state into a final desired state within a target region without violating forbidden regions. A series of simulation experiments have been conducted to demonstrate the effectiveness of the presented online particle filtering diagnostic approach.