Parallel Extended Kalman Filter Research Articles

Real-time multi-physical system identification and virtual sensing for a lab-scale chemical stirred tank using parallel estimators

In the initial stages of developing a new chemical process, chemists usually use a lab-scale stirred tank to analyze the yield amount and study the mass balance. However, understanding the energy balance is challenging. Monitoring the energy balance requires separating device and process-related effects, which is time-consuming and requires insight into the dynamics of the device and process. As a result, most lab-scale studies often overlook it. A real-time multi-physical virtual sensing and system identification approach is proposed in this study to address this issue by estimating desired unknown electro-thermo-mechanical parameters, states, and disturbances. The estimated unknowns include heat transfer coefficients, motor temperature, stirring torque, and heat generation rate. For this aim, a multi-physical lumped model is developed, which covers the electro-thermal and electro-mechanical responses of the device, as well as the process-related thermo-mechanical effects. The model is decoupled, and three parallel extended Kalman filters are employed. The developed model, estimation procedure, and measurement system are implemented in two experimental examples, including a hydrogenation reaction and the reductive catalytic fractionation of biomass. Both cases use a low-cost data acquisition system designed and manufactured to perform the measurements in this research. Comparing the logged measurements with the estimated values indicates considerable agreement in both cases. Based on the results, decoupling the model and using parallel estimators make the tuning easier, improve observability, and reduce the estimation time.

Fuzzy EKF-Based Intrusion Detection and Accurate State Estimation of Interconnected DC MGs With CPLs

With the advancement of digital technologies in complex and large-scale power systems, the issue of cyber-attack detection has become of paramount importance to maintain system reliability and performance. In this regard, accurate state estimation plays a key role and leads to proper supervisory decision-making. To detect cyber-attacks in direct current microgrids (DC MGs), a new centralized attack detection system based on the Kalman filter and clustering approach is proposed. On the other hand, appearing nonlinear loads such as constant power loads (CPLs) make the overall system behavior nonlinear and conventional linear Kalman filters ineffective. So, in this paper, an extended Kalman filter (EKF) is utilized. By constructing several output vectors and parallel EKFs, local state estimations are achieved. They are then used to detect attacks based on the residuals and localize them based on the fuzzy clustering approach. The local healthy estimations are aggregated to compute global estimations with low noise. The simulation results on a five-area interconnected DC MG subjected to data integrity and denial-of-service (DoS) attacks show that the proposed approach can detect attacks, evaluate how many attacks occur in the case of multiple attacks, localize the attacks, and obtain accurate state estimations.

Developing a cost-effective test system for evaluating dynamic performance of articulated heavy vehicles

In this paper, a three-level test system is proposed to comprehensively and accurately evaluate the dynamic performance of the whole articulated heavy vehicles from the perspective of the third party. At the sensor level, a reduced sensor configuration scheme is adopted to decrease the overall cost. At the fusion level, a group of parallel extended Kalman filters are developed to fuse the information from the sensors to accurately obtain the kinematical parameters of the observed points on the articulated heavy vehicle. At the evaluation level, taking full advantage of the geometry constraint and kinematics relationship, a calculating module is designed to determine the position, velocity, and azimuth of any vehicle unit of the articulated heavy vehicle. Furthermore, an assessing module is developed to evaluate the dynamic performance comprehensively, which includes a set of proposed inferring algorithms to automatically ascertain a series of attributes related to the dynamic performance. Finally, the feasibility and effectiveness of the proposed test system are validated through simulation and real field tests.

Enhancement of SNR in fetal ECG signal extraction using combined SWT and WLSR in parallel EKF

A Fetal Electrocardiogram (FECG) signal will contain some potential information which is precise for assisting the clinicians to make appropriate decisions that are timely at the time of pregnancy and labor. Extraction and detection of such FECG signals from their composite maternal signals of the abdomen using very powerful and advanced methodologies is a very critical need in case of fetal monitoring. The mining of pure FECG signals from the abdominal of the prenatal woman is focused in this paper. This FECG signal is very vulnerable to noise and is very difficult to process it accurately with no significant distortion that can imped its use. So, in order to be able to get some proper information on the status of the fetus or its condition it is important to be able to improve the abdominal signal and its SNR. Since the wavelet transform is well in providing information in time and frequency, the combination of Stationary Wavelet Transform (SWT), Weighted Least Square Regression (WLSR) and parallel Extended Kalman Filter (Par-EKF) are used for FECG extraction. The analysis for different values of Signal to Noise Ratio (SNR) between fetal and maternal ECG’s is implemented using MATLAB. As a Result, the highest SNR obtained is 33.5 which showed that the proposed method is efficient in extracting more information about FECG from Maternal ECG (MECG).

A Reliable Fusion Methodology for Simultaneous Estimation of Vehicle Sideslip and Yaw Angles

Vehicle sideslip and yaw angles are critical for many vehicle safety systems. Although much research has been presented to obtain them individually, simultaneous accurate estimation of them, based on affordable sensors for land vehicle applications, is seldom reported. This paper proposes a fusion methodology for integrating a single-frequency double-antenna Global Positioning System (DA-GPS) with other low-cost in-vehicle sensors to achieve reliable estimation of both vehicle sideslip and yaw angles. The proposed methodology adopts a hybrid decentralized filtering architecture. First, a vehicle state estimator (VSE) is developed as a virtual sensor to estimate vehicle state information, mainly according to vehicle dynamics. It is composed of two parallel extended Kalman filters (EKFs). Through the interaction of two EKFs, the VSE can adapt to the variations in tire-road friction and accurately estimate the vehicle roll angle. Then, according to vehicle kinematics, a global federated estimator (GFE) is designed, based on the federated filtering algorithm, to achieve the global fusion of inertial sensors, DA-GPS, and VSE. In particular, an adaptive inference mechanism is proposed and introduced into the GFE to adapt to complex driving situations, such as GPS failure, various driving maneuvers, etc. Finally, the proposed method is evaluated via intensive simulations and experiments. The overall results show that the proposed methodology can provide reliable estimation of two angles under a wide range of driving situations.

Rao-Blackwellized visual SLAM for small UAVs with vehicle model partition

Purpose – The purpose of this paper is to present a Rao–Blackwellized particle filter (RBPF) approach for the visual simultaneous localization and mapping (SLAM) of small unmanned aerial vehicles (UAVs). Design/methodology/approach – Measurements from inertial measurement unit, barometric altimeter and monocular camera are fused to estimate the state of the vehicle while building a feature map. In this SLAM framework, an extra factorization method is proposed to partition the vehicle model into subspaces as the internal and external states. The internal state is estimated by an extended Kalman filter (EKF). A particle filter is employed for the external state estimation and parallel EKFs are for the map management. Findings – Simulation results indicate that the proposed approach is more stable and accurate than other existing marginalized particle filter-based SLAM algorithms. Experiments are also carried out to verify the effectiveness of this SLAM method by comparing with a referential global positioning system/inertial navigation system. Originality/value – The main contribution of this paper is the theoretical derivation and experimental application of the Rao–Blackwellized visual SLAM algorithm with vehicle model partition for small UAVs.

An Object-Tracking Algorithm Based on Multiple-Model Particle Filtering With State Partitioning

As evidenced by the recent works of many researchers, the particle-filtering (PF) framework has revolutionized probabilistic visual target tracking. In this paper, we present a new particle filter tracking algorithm that incorporates the multiple-model (MM) paradigm and the technique of state partitioning with parallel filters. Traditionally, most tracking algorithms assume that a target operates according to a single dynamic model. However, the single-model assumption can cause the tracker to become unstable, particularly when the target has complex motions and when the camera has abrupt ego-motions. In the new tracking algorithm, a target was assumed to operate according to one dynamic model from a finite set of models. The switching process from one model to another was governed by a <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">jump</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Markov</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">process</i> . Based on the improved MM particle filter framework, we offer a new design strategy that adopts the state-partitioning technique and a bank of parallel extended Kalman filters to construct a better proposal distribution to achieve further estimation accuracy. We have conducted extensive testing for the proposed tracking algorithm, and key outcomes were given in the results section. It has been demonstrated by the experiments that this approach gave significantly improved estimations, enabling the new particle filter to effectively track human subjects.

Robust Robot Monte Carlo Localization

A robot localization algorithm based on particle filter is presented.Firstly,in order to improve the filtering effect and decrease the number of particles needed,one parallel extended Kalman filter is used as the proposal density of particle filter,thus partial observation information can be infused into the filtering process.Secondly,in order to enhance the particles' refining capacity,one improved Markov chain Monte Carlo (MCMC) resampling method with variable boundary of proposal density is put forward.Finally,the robot localization algorithm with the improved MCMC resampling is established,thus the effect of particle impoverishment can be decreased and the localization accuracy can be improved.Experiment results show that this algorithm has the advantages in computational complexity,localization accuracy and robustness.

A study on IMM with NPHMM and an application to speech enhancement

The nonlinear speech enhancement method with interactive parallel-extended Kalman filter is applied to speech contaminated by additive white noise. To represent the nonlinear and nonstationary nature of speech, we assume that speech is the output of a nonlinear prediction hidden Markov models (NPHMM) combining both neural network and HMM. The NPHMM is a nonlinear autoregressive process whose time-varying parameters are controlled by a hidden Markov chain. The simulation results shows that the proposed method offers better performance gains relative to the previous results (Signal Process 65 (1998) 373) with slightly increased complexity.

Submarine floating antenna model for LORAN-C signal processing

An electromagnetic model of the floating antenna used by submarines for LORAN-C radionavigation and very low frequency (VLF) communications is proposed. We present a description of the floating antenna and analyze its electric characteristics. The current in the horizontal wire induced by the lateral wave is then computed and we suggest a model that yields to a nonrational transfer function. A parameter estimation algorithm based on parallel-extended Kalman filters is then proposed and validated on real data.

Mini-UAV altitude estimation using an inertially stabilized payload

A novel method is introduced for autonomous attitude estimation of a mini unmanned aerial vehicle (UAV) carrying an inertially stabilized payload. The method is based on utilizing the outputs of rate gyros normally used to inertially stabilize the payload, and other data that is normally available from conventional aircraft-mounted sensors. A decentralized estimation algorithm is developed, which uses the aircraft/payload mathematical models to bound the estimation errors. Exploiting modern multiprocessor computer technology, the new estimation algorithm comprises two parallel extended Kalman filters (EKFs) and a data fusion algorithm. Real-time experimental tests, incorporating a payload model with real rate gyros mounted on a three-axis flight table, have validated the feasibility of the concept. The theoretical and experimental investigation demonstrates that the estimation algorithm is capable of estimating the attitude angles with an estimation error not exceeding 1 deg, at output rates of 13 Hz, thus constituting a viable substitute for the conventional vertical gyroscope.

Identification of dynamically positioned ships

Today's model-based dynamic positioning (DP) systems require that the ship and thruster dynamics are known with some accuracy in order to use linear quadratic optimal control theory. However, it is difficult to identify the mathematical model of a dynamically positioned (DP) ship, since the ship is not persistently excited under DP. In addition, the ship parameter-estimation problem is nonlinear and multivariable, with only position and thruster state measurements available for parameter estimation. The process and measurement noise must also be modeled in order to avoid parameter drift due to environmental disturbances and sensor failure. This article discusses an off-line parallel extended Kalman filter (EKF) algorithm utilizing two measurement series in parallel to estimate the parameters in the DP ship model. Full-scale experiments with a supply vessel are used to demonstrate the convergence and robustness of the proposed parameter estimator.

Identification of Dynamically Positioned Ships

Todays model-based dynamic positioning (DP) systems require that the ship and thruster dynamics are known with some accuracy in order to use linear quadratic optical control theory. However, it is difficult to identify the mathematical model of a dynamically positioned (DP) ship since the ship is not persistently excited under DP. In addition the ship parameter estimation problem is nonlinear and multivariable with only position and thruster state measurements available for parameter estimation. The process and measurement noise must also be modeled in order to avoid parameter drift due to environmental disturbances and sensor failure. This article discusses an off-line parallel extended Kalman filter (EKF) algorithm utilizing two measurement series in parallel to estimate the parameters in the DP ship model. Full-scale experiments with a supply vessel are used to demonstrate the convergence and robustness of the proposed parameter estimator.

Identification of Dynamically Positioned Ships

Todays model-based dynamic positioning (DP) systems require that the ship and thruster dynamics are known with some accuracy in order to use linear quadratic optimal control theory. However, it is difficult to identify the mathematical model of a dynamically positioned (DP) ship since the ship is not persistently excited under DP. In addition the ship parameter estimation problem is nonlinear and multivariable with only position and thruster state measurements available for parameter estimation. The process and measurement noise most also be modelled in order to avoid parameter drift due to environmental disturbances and sensor failure. This article discusses an offline parallel extended Kalman filter (EKF) algorithm utilizing two measurement series in parallel to estimate the parameters in the DP ship model. Full-scale experiments with a supply vessel are used to demonstrate the convergence and robustness of the proposed parameter estimator.

A reduced order extended Kalman filter for sequential images containing a moving object

The extended Kalman filter (EKF) is applied to the reduction of noise in sequential images containing a moving object and to the estimation of the object's velocity. A computationally tractable approximation of the EKF, called the parallel extended Kalman filter (PEKF), is generated. The PEKF consists of a parallel bank of third-order EKFs, operating on the Fourier coefficients of the image, followed by a finite impulse response filter. The PEKF is shown to converge to the optimal (in the mean square sense) algorithm in the limit as the velocity estimation errors approach zero. The performance of the PEKF is demonstrated for very low signal-to-noise ratio (SNR) images. The PEKF also provides a natural setting for tracking slow changes in the object (real or apparent) and its velocity, since these variations are included in the model. The relation of the PEKF to another frequency domain algorithm for velocity estimation is discussed. The algorithm is illustrated by application to an example and its performance is demonstrated in the presence of velocity estimation errors.