Particle Filter Approach Research Articles

A real-time probabilistic channel flood-forecasting model based on the Bayesian particle filter approach

Reliable real-time probabilistic flood forecasting is critical for effective water management and flood protection all over the world. In this study, we develop a real-time probabilistic channel flood-forecasting model by combining a channel hydraulic model with the Bayesian particle filter approach. The new model is tested in the upstream river reach of Three Gorges Dam (TGD) on the Yangtze River, China. Stage observations at seven hydrological stations are used simultaneously to adjust the Manning's roughness coefficients and to update discharges and stages along the river reach to attain reliable probabilistic flood forecasting. The synthetic experiments are applied to demonstrate the new model's correction and forecasting performances. The real-world experiments show that the new model can make accurate flood forecasting as well as derive reliable intervals for different confidence levels. The new probabilistic flood forecasting model not only outperforms the existing deterministic channel flood-forecasting models in accuracy, but also provides a more robust tool with which to incorporate uncertainty into flood-control efforts. A real-time flood-forecasting model is proposed by assimilating real-time stage observations into a hydraulic model.Particle filter is adopted as the data assimilation method to update/correct stage, discharge, and roughness coefficient.Synthetic experiments are employed to explore model settings and evaluate model performance.Model performance is compared with previous studies using Kalman Filter based methods.Probabilistic predictions provided by the model are more accurate and reliable.

A joint particle filter and expectation maximization approach to machine condition prognosis

This paper presents a probabilistic model based approach for machinery condition prognosis based on particle filter by integrating physical knowledge with in-process measurements into a state space framework to account for uncertainty and nonlinearity in machinery degradation process. One limitation of conventional particle filter is that condition prognosis is performed based on the model with predetermined parameters obtained from simulation studies or lab-controlled tests. Due to the stochastic nature of machinery defect propagation under varying operating conditions, model parameters may vary in practice which causes prediction errors. To address it, an integrated state prediction and parameter estimation framework based on particle filter and expectation-maximization algorithm is formulated and investigated. The model parameters are adaptively estimated based on expectation-maximization algorithm utilizing hidden degradation state and available in-process measurements. Particle filter is then performed on the identified model with estimated parameters following Bayesian inference scheme to improve the robustness and accuracy of machinery condition prognosis. The effectiveness of the developed method is demonstrated through a simulation study and an experimental run-to-failure bearing test in a wind turbine.

A Fusion Method Based on Unscented Particle Filter and Minimum Sampling Variance Resampling for Lithium-ion Battery Remaining Useful Life Prediction

It is important to predict the capacity of lithium-ion battery for future cycles to assess its health condition and to estimate remaining useful life (RUL). Particle filter approaches are widely applied into the estimation of battery capacity. However, after several iterations, the degeneracy and impoverishment of particles can cause unreliable and inaccurate prediction results in particle filter (PF). In this paper, a fusion method is proposed by integrating unscented Kalman filter (UKF) and minimum sampling variance resampling (MSVR) into the standard PF for RUL prediction of batteries. The UKF is employed to generate the proposal distribution of particles, which is used by PF to calculate the weights of particles. Next, the MSVR algorithm is introduced for performing resampling procedure to improve the performance. Finally, the performance of the proposed method is validated and compared to other predictors with four different battery datasets from NASA. According to the results, the integrated method has high reliability and prediction accuracy.

State-of-charge estimation in lithium-ion batteries: A particle filter approach

Abstract The dynamics of lithium-ion batteries are complex and are often approximated by models consisting of partial differential equations (PDEs) relating the internal ionic concentrations and potentials. The Pseudo two-dimensional model (P2D) is one model that performs sufficiently accurately under various operating conditions and battery chemistries. Despite its widespread use for prediction, this model is too complex for standard estimation and control applications. This article presents an original algorithm for state-of-charge estimation using the P2D model. Partial differential equations are discretized using implicit stable algorithms and reformulated into a nonlinear state-space model. This discrete, high-dimensional model (consisting of tens to hundreds of states) contains implicit, nonlinear algebraic equations. The uncertainty in the model is characterized by additive Gaussian noise. By exploiting the special structure of the pseudo two-dimensional model, a novel particle filter algorithm that sweeps in time and spatial coordinates independently is developed. This algorithm circumvents the degeneracy problems associated with high-dimensional state estimation and avoids the repetitive solution of implicit equations by defining a ‘tether’ particle. The approach is illustrated through extensive simulations.

Joint Estimation of CFO and Channel in OFDM Systems with Blind Noise Statistics

ABSTRACTWhile inter-symbol interference is not of concern, inter-carrier interference that is caused by carrier frequency offset (CFO) is more concerned in orthogonal frequency division multiplexing (OFDM) systems. The estimation of channel impulse response (CIR) is another important parameter of interest in the system. Recently, a number of approaches for the joint estimation of CFO and CIR in OFDM systems have been proposed in the literature, and most of them are traditional maximum-likelihood-based approaches. In this paper, we propose a number of mixture cost-reference particle filtering (CRPF) approaches for joint estimation of CIR and CFO, and a couple of them can be applied without the knowledge of both state and measurement noise statistics. Conventional least mean squares algorithm or recursive least squares algorithm is jointly employed with recently developed CRPF for nonlinear and multiple parameters estimation in OFDM systems. The computer simulation results show that mixture CRPFs are more robust against particle impoverishment (PI) problem than standard mixture particle filtering (PF) and mixture regularized PF (RPF) approaches although RPF was particularly developed to tackle the PI problem. In this paper, we investigate a number of mixture PF approaches including ones that do not need the information of noise statistics for the joint estimation of CFO and channel in OFDM systems.

Kalman particle filtering algorithm for symmetric alpha‐stable distribution signals with application to high frequency time difference of arrival geolocation

In this study, a non-linear filtering algorithm for state estimation with symmetric alpha-stable (SαS) noise is presented. The dynamic system model investigated here can be described by a linear state-space equation and a non-linear observation equation. The contribution of this study can be summarised as follows. First, particle filtering approach is employed for coarse estimation of the unknown parameters and then Kalman filter is performed to achieve better estimation. Second, SαS noise is considered as the additive disturbance in the observed signal and Gaussian approximation is used to compute the characteristics. Third, the calculation complexity is analysed according to the proposed algorithm. The proposed method is compared with the standard particle filter, extended Kalman filter and unscented Kalman filter for static parameter estimation of a periodic signal. As a practical application, the proposed method is used in high frequency source localisation based on time difference of arrival measurements.

GbLN-PSO and Model-Based Particle Filter Approach for Tracking Human Movements in Large View Cases

Camera tracking systems have become a common requirement in today’s society. The availability of high-quality and inexpensive video cameras and the increasing need for automated video analysis have generated a great deal of interest in numerous fields.In general, it is not easy to track human behavior in an environment with a large view. This paper aims to address four problems associated with large view in camera tracking system: 1) multiple targets in nonlinear motion; 2) relative size of the targeted object; 3) occlusion; and 4) processing time. This paper presents a new method of tracking human movements using global best local neighborhood oriented particle swarm optimization and model-based particle filter to address the above problems. The proposed method has been tested with an experimental module using several sets of video data provided by the 11th IEEE International Workshop on Performance Evaluation of Tracking and Surveillance and two other video streams of University of British Columbia (UBC) hockey and Malaysian football games. The experiment has shown that the accuracy of tracking performance has increased up to 25% compared with other reported works in the scientific literature.

Stream travel time prediction using particle filtering approach

Travel-time information is an integral part of Advanced Traveler Information Systems and Advanced Traffic Management Systems. Real-time estimation of stream travel time will be helpful in making trip decisions such as route choice and departure time. The present study proposes a method to predict stream travel time using particle filtering approach which considers the predicted stream travel time as the sum of the median of historical travel times, random variations in travel time over time, and a model evolution error. The present model hypothesizes the median of historical travel times obtained from the collected data as a priori estimate and hence predicting the actual travel time will be equivalent to forecasting the variations in travel time according to the current measurements. In order to capture the random variations in travel time, a dynamic mathematical modeling approach with particle filtering technique is used. The results obtained from the implementation of the above method are compared with the measured travel time data and the prediction accuracy, quantified using the Mean Absolute Percentage Error and Mean Absolute Error was observed to be satisfactory. Performance of this method was compared to a basic speed-based approach, where travel time is obtained by time–distance relationship using space-mean speed, and the proposed method showed better performance.

Dynamic changepoint detection in count time series: a particle filter approach

ABSTRACTWe study Bayesian dynamic models for detecting changepoints in count time series that present structural breaks. As the inferential approach, we develop a parameter learning version of the algorithm proposed by Chopin [Chopin N. Dynamic detection of changepoints in long time series. Annals of the Institute of Statistical Mathematics 2007;59:349–366.], called the Chopin filter with parameter learning, which allows us to estimate the static parameters in the model. In this extension, the static parameters are addressed by using the kernel smoothing approximations proposed by Liu and West [Liu J, West M. Combined parameters and state estimation in simulation-based filtering. In: Doucet A, de Freitas N, Gordon N, editors. Sequential Monte Carlo methods in practice. New York: Springer-Verlag; 2001]. The proposed methodology is then applied to both simulated and real data sets and the time series models include distributions that allow for overdispersion and/or zero inflation. Since our procedure is general, robust and naturally adaptive because the particle filter approach does not require restrictive specifications to ensure its validity and effectiveness, we believe it is a valuable alternative for dealing with the problem of detecting changepoints in count time series. The proposed methodology is also suitable for count time series with no changepoints and for independent count data.

Absolute Positioning Using the Earth's Magnetic Anomaly Field

Achieving worldwide dependable alternatives to the Global Positioning System is a challenging engineering problem. Current Global Positioning System alternatives often suffer from limitations such as where and when the systems can operate. Navigation using Earth’s magnetic anomaly field, which is globally available at all times, shows promise to overcome many of these limitations. We present a navigation framework that uses Earth’s magnetic anomaly field as a navigation signal to aid an inertial navigation system in an aircraft. The filter utilizes ultra-accurate optically pumped cesium magnetometers to make scalar intensity measurements of Earth’s magnetic field and compare them with a map using a particle filter approach. The accuracy of these measurements allows observability of not only the inertial navigation system errors but also the temporal effects of Earth’s magnetic field, which corrupt the navigation signal. These temporal effects are thoroughly analyzed, and we present a simple model that allows near worldwide use of the navigation filter. We analyze the dependencies on altitude and magnetic storm activity in a realistic simulation using data from test flights and magnetic observatories.

Particle Filters to Estimate Properties of Confined Aquifers

Involving a limited resource, the assessment of groundwater aquifers is of utmost importance. A key component of any such assessment is the determination of key properties that permit water resource managers to estimate aquifer drawdown and safe yield. This paper presents a particle filtering approach to estimate aquifer properties from transient data sets, leveraging recently published analytically-derived models for confined aquifers and using sample-based approximations of underlying probability distributions. The approach is examined experimentally through validation against three common aquifer testing problems: determination of (i) transmissivity and storage coefficient from non-leaky confined aquifer performance tests, (ii) transmissivity, storage coefficient, and vertical hydraulic conductivity from leaky confined aquifer performance tests, and (iii) transmissivity and storage coefficient from non-leaky confined aquifer performance tests with noisy data and boundary effects. On the first two well-addressed problems, the results using the particle filter approach compare favorably to those obtained by other published methods. The results to the third problem, which the particle filter approach can tackle more naturally than the previously-published methods, underscore the flexibility of particle filtering and, in turn, the promise such methods offer for a myriad of other geoscience problems.

동일한 형태의 특징점을 갖는 천장 영상 이용 이동 로봇 위치추정

Abstract This paper reports a localization method of a mobile robot using ceiling image. The ceiling has landmarks which are not distinguishablefrom one another. The location of every landmark in a map is given a priori while correspondence is not given between a detected landmark and a landmark in the map. Only the initial pose of the robot relative to the landmarks is given. The method uses particle filter approach for localization. Along with estimating robot pose, the method also associates a landmark in the map to a landmark detected from the ceiling image. The method is tested in an indoor environment which has circular landmarks on the ceiling. The test verifies the feasibility of the method in an environment where range data to walls or to beacons are not available or severely corrupted with noise. This method is useful for localization in a warehouse where measurement by Laser range finder and range data to beacons of RF or ultrasonic signal have large uncertainty.Key Words : Mobile Robot, Pose Estimation, Ceiling Image, Undistinguishable Landmarks, Particle Filter, Correspondence.이 논문은 2015학년도 조선대학교 학술연구비의 지원을 받아 연구되었음(This study was supported by research fund from Chosun University, 2015)This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

A hybrid framework combining data-driven and model-based methods for system remaining useful life prediction

Graphical abstractDisplay Omitted HighlightsA hybrid/fusion prognostics framework to predict remaining useful life by combining the data-driven methods and model-based methods.Introduce a data-driven method to estimate the measurement model in a model-based particle filter framework.Introduce a data-driven method to predicted future measurement in long term prediction in a model-based particle filter framework.Shown improved prediction accuracy using battery as a case study. Remaining useful life prediction is one of the key requirements in prognostics and health management. While a system or component exhibits degradation during its life cycle, there are various methods to predict its future performance and assess the time frame until it does no longer perform its desired functionality. The proposed data-driven and model-based hybrid/fusion prognostics framework interfaces a classical Bayesian model-based prognostics approach, namely particle filter, with two data-driven methods in purpose of improving the prediction accuracy. The first data-driven method establishes the measurement model (inferring the measurements from the internal system state) to account for situations where the internal system state is not accessible through direct measurements. The second data-driven method extrapolates the measurements beyond the range of actually available measurements to feed them back to the model-based method which further updates the particles and their weights during the long-term prediction phase. By leveraging the strengths of the data-driven and model-based methods, the proposed fusion prognostics framework can bridge the gap between data-driven prognostics and model-based prognostics when both abundant historical data and knowledge of the physical degradation process are available. The proposed framework was successfully applied on lithium-ion battery remaining useful life prediction and achieved a significantly better accuracy compared to the classical particle filter approach.

Theory of segmented particle filters

AbstractWe study the asymptotic behavior of a new particle filter approach for the estimation of hidden Markov models. In particular, we develop an algorithm where the latent-state sequence is segmented into multiple shorter portions, with an estimation technique based upon a separate particle filter in each portion. The partitioning facilitates the use of parallel processing, which reduces the wall-clock computational time. Based upon this approach, we introduce new estimators of the latent states and likelihood which have similar or better variance properties compared to estimators derived from standard particle filters. We show that the likelihood function estimator is unbiased, and show asymptotic normality of the underlying estimators.

Efficient Emulation of Radiative Transfer Codes Using Gaussian Processes and Application to Land Surface Parameter Inferences

There is an increasing need to consistently combine observations from different sensors to monitor the state of the land surface. In order to achieve this, robust methods based on the inversion of radiative transfer (RT) models can be used to interpret the satellite observations. This typically results in an inverse problem, but a major drawback of these methods is the computational complexity. We introduce the concept of Gaussian Process (GP) emulators: surrogate functions that accurately approximate RT models using a small set of input (e.g., leaf area index, leaf chlorophyll, etc.) and output (e.g., top-of-canopy reflectances or at sensor radiances) pairs. The emulators quantify the uncertainty of their approximation, and provide a fast and easy route to estimating the Jacobian of the original model, enabling the use of e.g., efficient gradient descent methods. We demonstrate the emulation of widely used RT models (PROSAIL and SEMIDISCRETE) and the coupling of vegetation and atmospheric (6S) RT models targetting particular sensor bands. A comparison with the full original model outputs shows that the emulators are a viable option to replace the original model, with negligible bias and discrepancies which are much smaller than the typical uncertainty in the observations. We also extend the theory of GP to cope with models with multivariate outputs (e.g., over the full solar reflective domain), and apply this to the emulation of PROSAIL, coupled 6S and PROSAIL and to the emulation of individual spectral components of 6S. In all cases, emulators successfully predict the full model output as well as accurately predict the gradient of the model calculated by finite differences, and produce speed ups between 10,000 and 50,000 times that of the original model. Finally, we use emulators to invert leaf area index ( L A I ), leaf chlorophyll content ( C a b ) and equivalent leaf water thickness ( C w ) from a time series of observations from Sentinel-2/MSI, Sentinel-3/SLSTR and Proba-V observations. We use sophisticated Hamiltonian Markov Chain Monte Carlo (MCMC) methods that exploit the speed of the emulators as well as the gradient estimation, a variational data assimilation (DA) method that extends the problem with temporal regularisation, and a particle filter using a regularisation model. The variational and particle filter approach appear more successful (meaning parameters closer to the truth, and smaller uncertainties) than the MCMC approach as a result of using the temporal regularisation mode. These work therefore suggests that GP emulators are a practical way to implement sophisticated parameter retrieval schemes in an era of increasing data volumes.

Bayesian Train Localization with Particle Filter, Loosely Coupled GNSS, IMU, and a Track Map

Train localization is safety-critical and therefore the approach requires a continuous availability and a track-selective accuracy. A probabilistic approach is followed up in order to cope with multiple sensors, measurement errors, imprecise information, and hidden variables as the topological position within the track network. The nonlinear estimation of the train localization posterior is addressed with a novel Rao-Blackwellized particle filter (RBPF) approach. There, embedded Kalman filters estimate certain linear state variables while the particle distribution can cope with the nonlinear cases of parallel tracks and switch scenarios. The train localization algorithm is further based on a track map and measurements from a Global Navigation Satellite System (GNSS) receiver and an inertial measurement unit (IMU). The GNSS integration is loosely coupled and the IMU integration is achieved without the common strapdown approach and suitable for low-cost IMUs. The implementation is evaluated with real measurements from a regional train at regular passenger service over 230 km of tracks with 107 split switches and parallel track scenarios of 58.5 km. The approach is analyzed with labeled data by means of ground truth of the traveled switch way. Track selectivity results reach 99.3% over parallel track scenarios and 97.2% of correctly resolved switch ways.

Particle filter and camshift approach for motion detection: a comparative study

Particle filter and camshift approach for motion detection: a comparative study

Lebesgue approximation model of continuous-time nonlinear dynamic systems

Traditional model-based approaches are based on periodic iterations, where the continuous-time model is discretized with a fixed period. Despite the easiness in analysis and design, such periodic approximation model may be undesirable from the computation-efficiency point of view. This paper presents the Lebesgue-approximation model (LAM) of continuous-time nonlinear systems , where the iteration is activated on an “as-needed” basis, but not periodically. We show that the proposed LAM behaves exactly the same as a specific event-triggered feedback system, through which the properties of the LAM can be studied. We provide a sufficient condition to ensure asymptotic stability of the LAM and derive theoretical bounds on the difference between the states of the LAM and the original continuous-time system. The LAM is then integrated in the particle-filtering approach for fault prognosis. Simulation results show that the LAM can dramatically reduce the number of iterations in prognosis without sacrificing accuracy and precision.

A Noise-Robust Convex-Optimized Positioning System Based on Code-Aided RSS Estimation and Virtual Base Station Transform

This paper presents a noise-robust mobile positioning system based on received signal strength (RSS) estimators, particle filters and a convex optimization processor. The positioning system uses a code-aided noise cancellation technique to refine the (RSS) signals and improve the positioning accuracy. This study also presents a virtual base-station transform (VBST) and convex optimization algorithm to deal with the none-light-of-sight (NLOS) travelling problem of the microwave transmission. The proposed positioning processor consists of four code-aided SNR/RSS estimators and four latency-reduced particle filters to refine the RSS signals from four base stations and estimate their distances to the mobile station. Then, the estimated distances are delivered to a convex optimization processor with the VBST algorithm to locate the mobile station. This work implemented the positioning algorithm on Xilinx Virtex-4 FPGA and RF modules to verify the positioning performance. The measurement and analysis results show that the proposed convex-optimized positioning system reduces 20 % RMSE in the mixed NLOS/LOS environment compared to the sole particle filtering approach. The RSS estimator can further improve the positioning performance in the noisy environments.

Incipient Fault Diagnostics and Remaining Useful Life Prediction of Analog Filters

Remaining useful life (RUL) prediction can effectively prevent failure in analog filters. Few researchers consider incipient fault diagnosis before implementing RUL prediction of analog filters. To deal with this problem, a prognostic framework with incipient fault diagnosis included is proposed for analog filters. Wavelet features and Mahalanobis distance (MD) are used to diagnose incipient faults in analog filters. After the incipient fault is detected and isolated, a fault indicator (FI) which can monitor the degradation trend of analog filters is developed based on traditional frequency features. Moreover, a power function model is used to track the degradation of analog filters exhibited by FI. In order to manage the uncertainties, a particle filtering approach is applied to model adaption and RUL prediction. Two case studies involving a band-pass filter and a low-pass filter have been used to validate the performance of the approaches. The simulation results show that the proposed approaches can diagnose incipient faults effectively and predict RUL of analog filters with small errors and high precision.