Multiple Model Adaptive Estimation Research Articles

Active Packet Dropout Kalman Filter Based Diagnosis Scheme For Liquid Rocket Engines With Measurement Fault

Various faults of the liquid rocket engine (LRE) threaten the safety of space launches. The LRE system faults are challenging to detect and isolate owing to sensor faults. In this paper, an active packet dropout Kalman filter (APD-KF) based method is proposed for solving the problem of LRE system fault detection and isolation (FDI) under measuring fault. Sensor faults are isolated firstly based on distributed extended multiple model adaptive estimation (EMMAE) scheme, and the sensor fault detection isolation (FI) result are then transformed to the package drop matrix and inserted into the system fault factor estimation process to discard fault measurement information. Experimental results based on data from simulation are used to decorate the efficiency of the method. This study may help to improve LRE health management and reduce launch risk by solving the problem of LRE system FDI with measurement fault.

Cable failure tolerant control and planning in a planar reconfigurable cable driven parallel robot.

The addition of geometric reconfigurability in a cable driven parallel robot (CDPR) introduces kinematic redundancies which can be exploited for manipulating structural and mechanical properties of the robot through redundancy resolution. In the event of a cable failure, a reconfigurable CDPR (rCDPR) can also realign its geometric arrangement to overcome the effects of cable failure and recover the original expected trajectory and complete the trajectory tracking task. In this paper we discuss a fault tolerant control (FTC) framework that relies on an Interactive Multiple Model (IMM) adaptive estimation filter for simultaneous fault detection and diagnosis (FDD) and task recovery. The redundancy resolution scheme for the kinematically redundant CDPR takes into account singularity avoidance, manipulability and wrench quality maximization during trajectory tracking. We further introduce a trajectory tracking methodology that enables the automatic task recovery algorithm to consistently return to the point of failure. This is particularly useful for applications where the planned trajectory is of greater importance than the goal positions, such as painting, welding or 3D printing applications. The proposed control framework is validated in simulation on a planar rCDPR with elastic cables and parameter uncertainties to introduce modeled and unmodeled dynamics in the system as it tracks a complete trajectory despite the occurrence of multiple cable failures. As cables fail one by one, the robot topology changes from an over-constrained to a fully constrained and then an under-constrained CDPR. The framework is applied with a constant-velocity kinematic feedforward controller which has the advantage of generating steady-state inputs despite dynamic oscillations during cable failures, as well as a Linear Quadratic Regulator (LQR) feedback controller to locally dampen these oscillations.

Uncertainty-aware transfer across tasks using hybrid model-based successor feature reinforcement learning☆

Sample efficiency, which refers to the number of samples required for a learning agent to attain a specific level of performance, is central to developing practical reinforcement learning (RL) for complex and large-scale decision-making problems. The ability to transfer and generalize knowledge gained from previous experiences to downstream tasks can significantly improve sample efficiency. Recent research indicates that successor feature (SF) RL algorithms enable knowledge generalization between tasks with different rewards but identical transition dynamics. It has recently been hypothesized that combining model-based (MB) methods with SF algorithms can alleviate the limitation of fixed transition dynamics. Furthermore, uncertainty-aware exploration is widely recognized as another appealing approach for improving sample efficiency. An agent can efficiently explore to better understand an environment by tracking uncertainty about the value of each available action. Putting together two ideas of hybrid model-based successor feature (MB-SF) and uncertainty leads to an approach to the problem of sample efficient uncertainty-aware knowledge transfer across tasks with different transition dynamics or/and reward functions. In this paper, the uncertainty of the value of each action is approximated by a Kalman filter (KF)-based multiple-model adaptive estimation. This KF-based framework treats the parameters of a model as random variables. To the best of our knowledge, this is the first attempt at formulating a hybrid MB-SF algorithm capable of generalizing knowledge across large or continuous state space tasks with various transition dynamics while requiring less computation at decision time than MB methods. We highlight why previous SF-based methods are constrained to knowledge generalization across same transition dynamics, present our novel approach on a firm theoretical foundation, and design a set of demonstration tasks to empirically validate the effectiveness of our proposed approach. The number of samples required to learn the tasks was compared to recent SF and MB baselines. The results show that our algorithm generalizes its knowledge across different transition dynamics, learns downstream tasks with significantly fewer samples than starting from scratch, and outperforms existing approaches. We believe that our proposed framework can account for the computationally efficient behavioural flexibilities observed in the empirical literature and can also serve as a solid theoretical foundation for future experimental work.

Resident Space Object Characterization Using Polarized Light Curves

Light curves, or the time-history of photometric brightness, have previously been demonstrated to allow for estimation of a space object’s attitude, shape, and surface parameters. However, these methods are subject to many limitations, including the need for accurate initial estimates to initialize filters, and the inability to regularly estimate attitude and surface parameters simultaneously due to information dilution, or the increase in uncertainty of each state when more states are simultaneously estimated. Monochromatic, unpolarized light curves can also only allow for surface reflectance parameter determination, and not determination of the materials themselves. This work develops a polarized bidirectional reflectance distribution function, and demonstrates that polarized measurements increase the uniqueness of light-curve attitude solutions to allow larger tolerance to initial error while also providing surface material estimation. Polarization of reflected light can also be measured more accurately than the intensity of reflected light through the atmosphere, thereby increasing the attitude observability from the light curves. Observability of attitude and surface materials is assessed. An unscented multiple-model adaptive estimation filter is implemented to estimate attitude, angular rate, surface material, surface roughness, and size of the space object.

Improved Multiple-Model Adaptive Estimation Method for Integrated Navigation with Time-Varying Noise.

The accurate noise parameter is essential for the Kalman filter to obtain optimal estimates. However, problems such as variations in the noise environment and measurement anomalies can cause degradation of estimation accuracy or even divergence. The adaptive Kalman filter can simultaneously estimate state and noise parameters, while its performance will also be degraded in complex noise. To address the problem of estimation accuracy degradation and result divergence of the integrated navigation system in a complex time-varying noise environment, an improved multiple-model adaptive estimation (MMAE) that combines the Sage–Husa adaptive unscented Kalman filter with the MMAE is proposed in this paper. The forgetting factor is included as an unknown parameter of MMAE so that the algorithm can adjust the value of the forgetting factor according to different system states. In addition, we improve the hypothesis testing algorithm of classical MMAE to deal with the competition problem of undesirable models that severely impacts the performance of variable-parameter MMAE and enhance the algorithm’s parameter identification capability. Simulation results show that this method enhances the system’s robustness to noises of different statistical properties and improves the estimation accuracy of the filter in time-varying noise environments.

A Novel Phasor Power Oscillation Damper With Adaptive Phase Compensation, Achieved Using Multiple Model Adaptive Estimation

In this paper, an adaptive version of the well known Phasor Power Oscillation Damper is presented. The proposed controller is adaptive in the sense that the phase compensation between the measured input and the applied control action is adjusted online during changing operating conditions. This is achieved through Multiple Model Adaptive Estimation, where a bank of Kalman filter estimators are run in parallel. A system model is inferred by assessing the performance of the individual filters, and used to update the tuning of the controller online. The proposed adaptive phase compensation scheme is compared with a similar scheme found in the literature, and is found to exhibit superior performance under the tested conditions. Simulation results from the Single-Machine Infinite Bus system, Kundur's Two-Area System and the IEEE 39-Bus System demonstrate that the proposed controller is able to adjust the phase compensation in response to severe non-linear events, and to eliminate phase lag due to communication latency.

A machine-learning approach to synthesize virtual sensors for parameter-varying systems

This paper introduces a novel model-free approach to synthesize virtual sensors for the estimation of dynamical quantities that are unmeasurable at runtime but are available for design purposes on test benches. After collecting a dataset of measurements of such quantities, together with other variables that are also available during on-line operations, the virtual sensor is obtained using machine learning techniques by training a predictor whose inputs are the measured variables and the features extracted by a bank of linear observers fed with the same measures. The approach is applicable to infer the value of quantities such as physical states and other time-varying parameters that affect the dynamics of the system. The proposed virtual sensor architecture — whose structure can be related to the Multiple Model Adaptive Estimation framework — is conceived to keep computational and memory requirements as low as possible, so that it can be efficiently implemented in embedded hardware platforms.The effectiveness of the approach is shown in different numerical examples, involving the estimation of the scheduling parameter of a nonlinear parameter-varying system, the reconstruction of the mode of a switching linear system, and the estimation of the state of charge (SoC) of a lithium-ion battery.

Model-based thermal anomaly detection for lithium-ion batteries using multiple-model residual generation

The continuously increasing energy and power density of lithium-ion batteries will aggravate the safety and reliability concerns of advanced battery management systems (BMSs). To ensure the safety and reliability of lithium-ion batteries, the BMS must implement anomaly detection algorithms that are capable of capturing abnormal behaviors. Thermal anomalies are one of the most critical anomalies that can be potentially catastrophic. Motivated by this, a model-based strategy of anomaly detection of thermal parameters for lithium-ion-batteries is presented in this paper. The algorithm is based on a multiple-model adaptive estimation framework. Firstly, an equivalent-circuit-model-based electrothermal model is proposed to describe battery dynamic behaviors. Then, a combination of the recursive-least-square method and Kalman-filter is employed to generate residual signals for thermal anomaly detection. Furthermore, the probability of the signature anomaly is evaluated through the multiple-model adaptive estimation technique. Distinguished from existing threshold-based methods, the proposed method can determine particular anomalies according to the value of the generated conditional probability, without a manually determined threshold. Simulations are developed to simulate different faults and generate data for algorithm validation. The results show signature thermal anomaly can be detected accurately.

Robust Frequency and Phase Estimation for Three-Phase Power Systems Using a Bank of Kalman Filters

In this paper we propose a powerful frequency, phase angle, and amplitude estimation solution for an unbalanced three-phase power system based on multiple model adaptive estimation. The proposed model utilizes the existence of a conditionally linear and Gaussian substructure in the power system states by marginalizing out the frequency component. This substructure can be effectively tracked by a bank of Kalman filters where each filter employs a different angular frequency value. Compared to other Bayesian filtering schemes for estimation in three-phase power systems, the proposed model reformulation is simpler, more robust, and more accurate as validated with numerical simulations on synthetic data.

Cooperative interception with fast multiple model adaptive estimation

For the case that two pursuers intercept an evasive target, the cooperative strategies and state estimation methods taken by pursuers can seriously affect the guidance accuracy for the target, which performs a bang For the case that two pursuers intercept an evasive target, the cooperative strategies and state estimation methods taken by pursuers can seriously affect the guidance accuracy for the target, which performs a bang-bang evasive maneuver with a random switching time. Combined Fast multiple model adaptive estimation (Fast MMAE) algorithm, the cooperative guidance law takes detection configuration affecting the accuracy of interception into consideration. Introduced the detection error model related to the line-of-sight (LOS) separation angle of two interceptors, an optimal cooperative guidance law solving the optimization problem is designed to modulate the LOS separation angle to reduce the estimation error and improve the interception performance. Due to the uncertainty of the target bang-bang maneuver switching time and the effective fitting of its multi-modal motion, Fast MMAE is introduced to identify its maneuver switching time and estimate the acceleration of the target to track and intercept the target accurately. The designed cooperative optimal guidance law with Fast MMAE has better estimation ability and interception performance than the traditional guidance law and estimation method via Monte Carlo simulation.

Switching TS fuzzy model-based dynamic sliding mode observer design for non-differentiable nonlinear systems

This paper presents four sliding mode observers (SMOs) based on a novel approach in Takagi–Sugeno (TS) fuzzy modeling of multi-input multi-output (MIMO) non-linear systems that have non-differentiable operating points. A comprehensive approach is proposed to using the TS fuzzy model (TSFM) in the field of non-differentiable nonlinear systems, where the TSFM is an approximation with high accuracy and a negligible error (2ε) of the nonlinear model. Furthermore, the considered system can be with measurable or unmeasurable premise variables. The observers are synthesized for the above two cases and dynamic observers for state estimations of MIMO non-linear Lipschitz systems. The dynamic gain of the observer is established from inspiring state-space representation of an LTI system with error as input, internal states, and the gain of the observer as output. The dynamics used in the gain of the observer will increase the degrees of freedom in the design procedure and a generalization to the one used in previous works. The proposed method is applicable for continuous-time, but not necessarily differentiable, nonlinear systems. Considering the inherent strongly nonlinear and coupling performance of the plant, the switching method driven by states is presented. This paper presents a comparison of four SMOs and multiple-model adaptive estimation (MMAE) for benchmark hydraulic wind power transfer (HWPT). Simulation results demonstrate improvement in the state observation convergence rate and simplicity and universality of the proposed approach.

EKF-Based Actuator Fault Detection and Diagnosis Method for Tilt-Rotor Unmanned Aerial Vehicles

Flight safety is of vital importance for tilt-rotor unmanned aerial vehicles (UAVs), which can take off and land vertically as well as cruise at high speed, especially in different kinds of complex environment. As being the executor of the flight control, the actuator failure will directly affect the controllability of the tilt-rotor UAV, and it has high probability of causing fatal personal injury and financial loss. However, due to the limitation of weight and cost, small UAVs cannot be equipped with redundant actuators. Therefore, there is an urgent need of fault detection and diagnosis method for the actuators. In this paper, an actuator fault detection and diagnosis (FDD) method based on the extended Kalman filter (EKF) and multiple-model adaptive estimation (MMAE) is proposed. The actuator deflections are added to the state vector and estimated using EKF. The fault diagnosis algorithm of MMAE could assign a conditional probability to each faulty actuator according to the residual of EKF and diagnose the fault. This paper is structured as follows: first, the structure and model of tilt-rotor UAV actuator are established. Then, EKF observers are introduced to estimate the state vector and to calculate residual sequences caused by different faulty actuators. The residuals from EKFs are used by fault diagnosis algorithm to assign a conditional probability to each failure condition, and fault type can be diagnosed according to the probabilities. The FDD method is verified by simulations, and the results demonstrate that the FDD algorithm could accurately and efficiently diagnose actuator fault without any additional sensor.

INS/Odometer Land Navigation by Accurate Measurement Modeling and Multiple-Model Adaptive Estimation

Land vehicle navigation based on inertial navigation system (INS) and odometers is a classical autonomous navigation application and has been extensively studied over the past several decades. In this work, we seriously analyze the error characteristics of the odometer (OD) pulses and investigate three types of odometer measurement models in the INS/OD integrated system. Specifically, in the pulse velocity model, a preliminary Kalman filter is designed to obtain accurate vehicle velocity from the accumulated pulses; the pulse increment model is accordingly obtained by integrating the pulse velocity; a new pulse accumulation model is proposed by augmenting the travelled distance into the system state. The three types of measurements, along with the nonhonolomic constraint (NHC), are implemented in the standard extended Kalman filter. In view of the motion-related pulse error characteristics, the multiple model adaptive estimation (MMAE) approach is exploited to further enhance the performance. Simulations and long-distance experiments are conducted to verify the feasibility and effectiveness of the proposed methods. It is shown that the standard pulse velocity measurement achieves the superior performance, whereas the accumulated pulse measurement is most favorable with the MMAE enhancement.

Kalman-based interacting multiple-model wind speed estimator for wind turbines

The use of state estimation technique offers a means of inferring the rotor-effective wind speed based upon solely standard measurements of the turbine. For the ease of design and computational concerns, such estimators are typically built based upon simplified turbine models that characterise the turbine with rigid blades. Large model mismatch, particularly in the power coefficient, could lead to degradation in estimation performance. Therefore, in order to effectively reduce the adverse impact of parameter uncertainties in the estimator model, this paper develops a wind sped estimator based on the concept of interacting multiple-model adaptive estimation. The proposed estimator is composed of a bank of extended Kalman filters and each filter model is developed based on different power coefficient mapping to match the operating turbine parameter. Subsequently, the algorithm combines the wind speed estimates provided by each filter based on their statistical properties. In addition, the proposed estimator not only can infer the rotor-effective wind speed, but also the uncertain system parameters, namely, the power coefficient. Simulation results demonstrate the proposed estimator achieved better improvement in estimating the rotor-effective wind speed and power coefficient compared to the standard Kalman filter approach.

Kernel Adaptive Filtering Multiple-model Actuator Fault Diagnostic For Multi-effectors Aircraft

In the traditional Multiple Model Adaptive Estimation (MMAE) algorithm, the extended Kalman filter has theoretical limitations, and the establishment of accurate aircraft mathematical model is almost impossible. In this paper, the Kernel Adaptive Filter (KAF) is introduced to replace the Kalman filter, a new multi-model adaptive estimation fault diagnosis method is proposed. Based on the kernel methods, the adaptive filter is designed in the high-dimensional feature space without the need to know the system model in advance. After training of KAF using the offline input control signal and output flight states measurement with noise, the estimation of real flight states values and actuator fault detection and isolation can be realized online. The simulation results show good performance of new fault diagnosis method in actuator fault diagnosis.

Low-Complexity Architecture for Cyber-Physical Systems Model Identification

We propose a low complexity architecture for cyber-physical system (CPS) model identification based on multiple-model adaptive estimation (MMAE) algorithms. The complexity reduction is achieved by reducing the number of multiplications in the filter banks of the MMAE algorithm present in the cyber component of the CPS. The architecture has been implemented using FPGA for 16, 32, 64 filter banks as part of position and velocity estimations of autonomous auto-mobile application. It has been found up to 78% reduction in multiplications is possible, which translates to the reduction of 39% lookup tables, 13% FFs, 27% DSPs, and 43% power reduction when compared with the conventional architecture (without multiplications reduction) at 100MHz operating frequency. Furthermore, the proposed architecture is able to identify accurate model of auto-mobile application just within 510 ns, in the presence of external disturbances and abrupt changes.

Online Flooding and Dehydration Diagnosis for PEM Fuel Cell Stacks via Generalized Residual Multiple Model Adaptive Estimation-Based Methodology

<div class="section abstract"><div class="htmlview paragraph">For proton exchange membrane fuel cell (PEMFC) stack, critical issues such as flooding and dehydration, are caused by improper water management. With respect to the water management failure, PEMFC stack outputs power and efficiency decreased. Therefore, proper water management with diagnosis contributes to the reliability and durability. Existing researches establish Electrochemical Impedance Spectroscopy (EIS) measurement to detect and identify different faults, whereas the sophistication, overwhelmed computational consumption of EIS and unaffordable dedicated instrumentation make it’s unsuitable for commercial application. Therefore, EIS is not considered to be a viable solution to online and real-time diagnostic scheme. In this paper, an innovative method based on EIS, is further developed to identify some critical PEMFC fault conditions online, wherein generalized residual multiple model adaptive estimation (GRMMAE) methodology is applied. The diagnosis process consists of multiple equivalent circuit models that represent signature faults separately, such as flooding and dehydration, causing significant variation of model parameters. Applying a small sinusoidal alternating current (AC) as a perturbation signal to various parallel models and measuring the potential response, whilst the noise from perturbation signal and measurement result in utilizing standard Kalman filters (KF) to generate residual signals. The residual signals are used in the GRMMAE methodology to evaluate probabilities in real-time that determine the type of fault. Simulation result shows that the fault conditions can be detected and identified accurately, hence the effectiveness of the proposed method would be indicated.</div></div>

Adaptive Kalman filtering for sensor fault estimation and isolation of satellite attitude control based on descriptor systems

Sensor fault estimation and isolation is significant for an attitude control systems model of a satellite, as it works in a complex environment. The standard unscented Kalman filter algorithm may lose its accuracy when the noise is considerable. Therefore, an adaptive filtering algorithm is proposed based on the sampled-data descriptor model. The performance of the unscented Kalman filter in sensor fault estimation is improved by the adaptive algorithm depending on innovation and the measurement residual, and its convergence is guaranteed. Combining the adaptive unscented Kalman filter with the multiple-model adaptive estimation, a sensor fault isolation method is proposed. Finally, simulation examples show that this algorithm has better estimating accuracy and isolation results.

Comparison of fault detection and isolation methods for a small unmanned aircraft

This paper compares three model-based methods for detecting and isolating stuck control surface faults on a small unmanned aircraft. The first method is parity-space based and compares a sensor measurement against a model-based prediction of the same quantity. The second method is observer-based and involves robust estimation for linear parameter-varying systems. The third method is also observer-based and involves multiple model adaptive estimation. The performance of the three methods are compared using flight data. The comparison shows that the three methods have different detection performance and runtime. The selection of a particular method depends on the application requirements and implementation constraints.

Multiple Model Kalman and Particle Filters and Applications: A Survey

Abstract Kalman Filters (KF) is a recursive estimation algorithm, a special case of Bayesian estimators under Gaussian, linear and quadratic conditions. For non-linear systems, Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) provide first and higher order linearization approximations. Particle Filters (PF), on the other hand, are sequential Monte Carlo methods to provide estimations for non-linear non-Gaussian problems. For complex systems, Kalman or Particle Filter based single model filters may not be sufficient to model the system behaviour. Multiple Model (MM) Filters achieve more reliable estimates by using more than one filter with different models in parallel and the outputs of each filter are fused by assigning a probability to each filter. The most common methods used in the literature for multiple model estimation are Multiple Model Adaptive Estimation (MMAE) and Interacting Multiple Model (IMM). This paper presents an overview of the recent research on multiple model filters.