Independent Kalman Filters Research Articles

Managing Communication Delays and Model Error in Demand Response for Frequency Regulation

This study develops and compares several networked control and estimation algorithms to manipulate the power consumption of a population of residential thermostatically controlled loads (TCLs) to fulfill PJM frequency regulation requests given an imperfect communication network and modeling error. The algorithms rely on a model of the plant to reduce the effects of communication delays, and include a stochastic, predictive controller and two Kalman filter-based state estimation techniques. The first estimator uses a set of independent Kalman filters that run in parallel, and the second incorporates individual TCL models that rely on identified thermal parameters. We use simulations to examine 1) the algorithms’ ability to adequately provide frequency regulation under a range of delay severities, and 2) the effect of increased modeling error. We find that both estimator–controller combinations provide acceptable frequency regulation with average delays of 20 s and minor modeling error. When we increase the modeling error by using a higher order model to represent the TCLs within the plant, the first estimator provides acceptable frequency regulation, whereas the second estimator provides poor frequency regulation.

Sliding-Mode Control for a Three-Phase Unity Power Factor Rectifier Operating at Fixed Switching Frequency

This paper presents an improved variable hysteresis-band current-control in natural frame for a three-phase unity power rectifier. The proposed control algorithm is based on three decoupled sliding-mode controllers combined with three independent Kalman filters. The use of Kalman filters instead of a nonadaptive state observer improves the quality of the estimated signals in presence of noise, increasing the immunity of the control loop in noisy environments. To reduce drastically the computational load in the Kalman algorithm, a reduced bilinear model is derived which allows to use a Kalman filter algorithm instead of an extended Kalman filter. A fast output-voltage control is also presented which avoids output-voltage variations when a sudden change in the load or a voltage sag appears. Moreover, a fixed switching frequency algorithm is proposed which uses a variable hysteresis-band in combination with a switching decision algorithm, ensuring a switching spectrum concentrated around the desired switching frequency. The overall control proposal has been fully integrated into a digital signal processor. Selected experimental results are introduced to validate the theoretical contributions of this paper.

Dual FastSLAM: Dual Factorization of the Particle Filter Based Solution of the Simultaneous Localization and Mapping Problem

The process of building a map with a mobile robot is known as the Simultaneous Localization and Mapping (SLAM) problem, and is considered essential for achieving true autonomy. The best existing solutions to the SLAM problem are based on probabilistic techniques, mainly derived from the basic Bayes Filter. A recent approach is the use of Rao-Blackwellized particle filters. The FastSLAM solution factorizes the Bayes SLAM posterior using a particle filter to estimate over the possible paths of the robot and several independent Kalman Filters attached to each particle to estimate the location of landmarks conditioned to the robot path. Although there are several successful implementations of this idea, there is a lack of applications to indoor environments where the most common feature is the line segment corresponding to straight walls. This paper presents a novel factorization, which is the dual of the existing FastSLAM one, that decouples the SLAM into a map estimation and a localization problem, using a particle filter to estimate over maps and a Kalman Filter attached to each particle to estimate the robot pose conditioned to the given map. We have implemented and tested this approach, analyzing and comparing our solution with the FastSLAM one, and successfully building feature based maps of indoor environments.

Multi-sensor fusion for robust computation of breathing rate

Respiratory information can be obtained from the changes in electrical impedance across the chest, the electrocardiogram or the changes in light absorption across the finger. Running estimates of the breathing rate from each of these can be obtained from independent Kalman filters. A robust estimate of breathing rate is derived by fusing the outputs of these independent Kalman filters using a weighting calculated from the squared differences between each prediction from the filters and the corresponding measurements.

Historical perspective on estimation techniques for position and gravity survey with inertial systems

Introduction I NVESTIGATIONS into the application of inertial technology to geodetic surveys have covered a span of at least 25 years, involving a number of private companies and government and university laboratories. The principal agency sponsoring this work in the United States has been the U.S. Army Engineer Topographic Laboratories (USAETL), whose original efforts began in the early 1960s. In 1965 the Guidance and Control Systems Division of Litton Industries began its initial tradeoff studies on the Position and Azimuth Determining System (PADS) under USAETL sponsorship. The original estimation problem posed for the PADS was to design software to process discrete measurements from an odometer, or a laser-velocimeter, of the error in systemcomputed velocity observable at vehicle stops (subsequently called Zero-Velocity Updates or ZUPTS), so that accuracies of 20 m in level position, 10 m in elevation, and 0.3 mrad in azimuth were preserved over a 6-h open-traverse mission. The solution obtained was that of two independent Kalman filters, one for control of the errors associated with the local-level navigation axes and the other for control of errors associated with the vertical axis. Periodic vehicle stops were required, depending on the mode of system operation, to make system corrections with the Kalman filter(s) of the observed velocity error. The design was verified for all three modes of operation via simulation with extensive system error models and successfully field-tested in 1972. The first paper available to the general public on the basic principles for control of system error in the design of the PADS, along with simulation and test results was published in the same year. As in practice the stopping of the vehicle every 10 min to make ZUPT corrections 4s not excessively inconvenient, the principal operational mode for the PADS has become the unaided, free-inertial mode during vehicle travel. Also contributing to this result was the fact that the laser-velocimeter was an expensive sensor which gave the vehicle a distinctive signature, although it allowed the travel period to be extended to 1 h. Utilization of the odometer has also diminished as it extended the travel period to only 20 min. The success of the PADS prototype hardware-testing program for the free-inertial mode in 1972' stimulated interest by a number of groups on the utilization of this inertial equipment for a higher-accuracy position, elevation, and gravityvector survey capability in helicopters as well as land vehicles. The higher accuracy was easily obtained by relaxing the constraints imposed by the real-time artillery survey mission. The techniques employed were longer pre-mission alignment and calibration time, shorter travel periods between vehicle stops, longer stop periods during which system corrections of the observed error in system-computed velocity were made, and simple adjustment of the real-time position and elevation survey values, using the errors in these quantities available (only) at traverse closure.''