scholarly journals A simple model for learning in volatile environments

2019 ◽  
Author(s):  
Payam Piray ◽  
Nathaniel D. Daw
Keyword(s):  
Kalman Filter ◽  
State Space Model ◽  
Choice Data ◽  
Exact Inference ◽  
Second Order Statistics ◽  
Decision Neuroscience ◽  
Learning Tasks ◽  
Neuroscience Research ◽  
Proposed Model ◽  
Psychological Models

AbstractSound principles of statistical inference dictate that uncertainty shapes learning. In this work, we revisit the question of learning in volatile environments, in which both the first and second-order statistics of observations dynamically evolve over time. We propose a new model, the volatile Kalman filter (VKF), which is based on a tractable state-space model of uncertainty and extends the Kalman filter algorithm to volatile environments. The proposed model is algorithmically simple and encompasses the Kalman filter as a special case. Specifically, in addition to the error-correcting rule of Kalman filter for learning observations, the VKF learns volatility according to a second error-correcting rule. These dual updates echo and contextualize classical psychological models of learning, in particular hybrid accounts of Pearce-Hall and Rescorla-Wagner. At the computational level, compared with existing models, the VKF gives up some flexibility in the generative model to enable a more faithful approximation to exact inference. When fit to empirical data, the VKF is better behaved than alternatives and better captures human choice data in two independent datasets of probabilistic learning tasks. The proposed model provides a coherent account of learning in stable or volatile environments and has implications for decision neuroscience research.

Real-time SOC Estimation of the EKF and UKF Algorithms Under Different Working Conditions

2021 ◽  
pp. 1-29
Author(s):  
Xiongbin Peng ◽  
Yuwu Li ◽  
Wei Yang ◽  
Akhil Garg
Keyword(s):  
Kalman Filter ◽  
Working Conditions ◽  
Unscented Kalman Filter ◽  
Stress Test ◽  
Estimation Method ◽  
Influential Factor ◽  
Recursive Least Squares ◽  
Maximum Deviation ◽  
Soc Estimation ◽  
Proposed Model

Abstract In the battery thermal management system (BMS), the state of charge (SOC) is a very influential factor, which can prevent overcharge and over-discharge of the lithium-ion battery (LIB). This paper proposed a battery modeling and online battery parameter identification method based on the Thevenin equivalent circuit model (ECM) and recursive least squares (RLS) algorithm. The proposed model proved to have high accuracy. The error between the ECM terminal voltage value and the actual value basically fluctuates between ±0.1V. The extended Kalman filter (EKF) algorithm and the unscented Kalman filter (UKF) algorithm were applied to estimate the SOC of the battery based on the proposed model. The SOC experimental results obtained under dynamic stress test (DST), federal urban driving schedule (FUDS), and US06 cycle conditions were analyzed. The maximum deviation of the SOC based on EKF was 1.4112%~2.5988%, and the maximum deviation of the SOC based on UKF was 0.3172%~0.3388%. The SOC estimation method based on UKF and RLS provides a smaller deviation and better adaptability in different working conditions, which makes it more implementable in a real-world automobile application.

scholarly journals Variational Bayesian-Based Improved Maximum Mixture Correntropy Kalman Filter for Non-Gaussian Noise

Entropy ◽  
2022 ◽  
Vol 24 (1) ◽  
pp. 117
Author(s):  
Xuyou Li ◽  
Yanda Guo ◽  
Qingwen Meng
Keyword(s):  
Kalman Filter ◽  
Gaussian Noise ◽  
State Space Model ◽  
Performance Evaluations ◽  
Gaussian State ◽  
Variational Bayesian ◽  
Space Model ◽  
Bandwidth Parameter ◽  
Non Gaussian ◽  
Filtering Problem

The maximum correntropy Kalman filter (MCKF) is an effective algorithm that was proposed to solve the non-Gaussian filtering problem for linear systems. Compared with the original Kalman filter (KF), the MCKF is a sub-optimal filter with Gaussian correntropy objective function, which has been demonstrated to have excellent robustness to non-Gaussian noise. However, the performance of MCKF is affected by its kernel bandwidth parameter, and a constant kernel bandwidth may lead to severe accuracy degradation in non-stationary noises. In order to solve this problem, the mixture correntropy method is further explored in this work, and an improved maximum mixture correntropy KF (IMMCKF) is proposed. By derivation, the random variables that obey Beta-Bernoulli distribution are taken as intermediate parameters, and a new hierarchical Gaussian state-space model was established. Finally, the unknown mixing probability and state estimation vector at each moment are inferred via a variational Bayesian approach, which provides an effective solution to improve the applicability of MCKFs in non-stationary noises. Performance evaluations demonstrate that the proposed filter significantly improves the existing MCKFs in non-stationary noises.

scholarly journals Opportunistic In-Flight INS Alignment Using LEO Satellites and a Rotatory IMU Platform

Aerospace ◽  
2021 ◽  
Vol 8 (10) ◽  
pp. 280
Author(s):  
Farzan Farhangian ◽  
Hamza Benzerrouk ◽  
Rene Landry
Keyword(s):  
Kalman Filter ◽  
State Space Model ◽  
Flight Test ◽  
Satellite System ◽  
Low Earth Orbit ◽  
Measurement Unit ◽  
Starting Point ◽  
Leo Satellites ◽  
Positioning Method ◽  
Global Navigation Satellite

With the emergence of numerous low Earth orbit (LEO) satellite constellations such as Iridium-Next, Globalstar, Orbcomm, Starlink, and OneWeb, the idea of considering their downlink signals as a source of pseudorange and pseudorange rate measurements has become incredibly attractive to the community. LEO satellites could be a reliable alternative for environments or situations in which the global navigation satellite system (GNSS) is blocked or inaccessible. In this article, we present a novel in-flight alignment method for a strapdown inertial navigation system (SINS) using Doppler shift measurements obtained from single or multi-constellation LEO satellites and a rotation technique applied on the inertial measurement unit (IMU). Firstly, a regular Doppler positioning algorithm based on the extended Kalman filter (EKF) calculates states of the receiver. This system is considered as a slave block. In parallel, a master INS estimates the position, velocity, and attitude of the system. Secondly, the linearized state space model of the INS errors is formulated. The alignment model accounts for obtaining the errors of the INS by a Kalman filter. The measurements of this system are the difference in the outputs from the master and slave systems. Thirdly, as the observability rank of the system is not sufficient for estimating all the parameters, a discrete dual-axis IMU rotation sequence was simulated. By increasing the observability rank of the system, all the states were estimated. Two experiments were performed with different overhead satellites and numbers of constellations: one for a ground vehicle and another for a small flight vehicle. Finally, the results showed a significant improvement compared to stand-alone INS and the regular Doppler positioning method. The error of the ground test reached around 26 m. This error for the flight test was demonstrated in different time intervals from the starting point of the trajectory. The proposed method showed a 180% accuracy improvement compared to the Doppler positioning method for up to 4.5 min after blocking the GNSS.

Parameter Estimation in a Hydrostatic System Using Extended Kalman Filter

2001 ◽  
Author(s):  
Yuvin A. Chinniah ◽  
Richard Burton ◽  
Saeid Habibi
Keyword(s):  
Kalman Filter ◽  
Bulk Modulus ◽  
Extended Kalman Filter ◽  
High Performance ◽  
State Space Model ◽  
System Response ◽  
Actuation System ◽  
Accuracy And Precision ◽  
Hydrostatic System ◽  
Entrained Air

Abstract In this paper, the Extended Kalman Filter (EKF) estimation technique is applied to a novel hydrostatic actuation system referred to as the Electrohydraulic Actuator (EHA). A state space model of the EHA is developed and the effective bulk modulus is estimated in simulation. The EHA is a high performance actuation system capable of moving large loads with very high accuracy and precision. In a practical situation, this parameter is very difficult to measure directly as it depends on entrained air which cannot be known at a particular point of time. The bulk modulus is critical for system response and a low bulk modulus as a result of air in the system can seriously hinder the performance of EHA and cause safety problems.

A new robust Kalman filter with measurement loss based on mixing distribution

2021 ◽  
pp. 014233122110549
Author(s):  
Chenghao Shan ◽  
Weidong Zhou ◽  
Yefeng Yang ◽  
Hanyu Shan
Keyword(s):  
Kalman Filter ◽  
Likelihood Function ◽  
State Space Model ◽  
Random Variable ◽  
Superior Performance ◽  
Gaussian State ◽  
Bernoulli Random Variable ◽  
Mixing Distribution ◽  
Robust Kalman Filter ◽  
Heavy Tailed

A new robust Kalman filter (KF) based on mixing distribution is presented to address the filtering issue for a linear system with measurement loss (ML) and heavy-tailed measurement noise (HTMN) in this paper. A new Student’s t-inverse-Wishart-Gamma mixing distribution is derived to more rationally model the HTMN. By employing a discrete Bernoulli random variable (DBRV), the form of measurement likelihood function of double mixing distributions is converted from a weighted sum to an exponential product, and a hierarchical Gaussian state-space model (HGSSM) is therefore established. Finally, the system state, the intermediate random variables (IRVs) of the new STIWG distribution, and the DBRV are simultaneously estimated by utilizing the variational Bayesian (VB) method. Numerical example simulation experiment indicates that the proposed filter in this paper has superior performance than current algorithms in processing ML and HTMN.

A Novel Neural Model with Lateral Interaction for Learning Tasks

2020 ◽  
pp. 1-24
Author(s):  
Dequan Jin ◽  
Ziyan Qin ◽  
Murong Yang ◽  
Penghe Chen
Keyword(s):  
Numerical Experiments ◽  
State Of The Art ◽  
Neural Model ◽  
Parameter Tuning ◽  
Small Sample ◽  
Lateral Interaction ◽  
Learning Tasks ◽  
Proposed Model ◽  
High Level ◽  
Elementary Field

We propose a novel neural model with lateral interaction for learning tasks. The model consists of two functional fields: an elementary field to extract features and a high-level field to store and recognize patterns. Each field is composed of some neurons with lateral interaction, and the neurons in different fields are connected by the rules of synaptic plasticity. The model is established on the current research of cognition and neuroscience, making it more transparent and biologically explainable. Our proposed model is applied to data classification and clustering. The corresponding algorithms share similar processes without requiring any parameter tuning and optimization processes. Numerical experiments validate that the proposed model is feasible in different learning tasks and superior to some state-of-the-art methods, especially in small sample learning, one-shot learning, and clustering.

scholarly journals Graph-Driven Generative Models for Heterogeneous Multi-Task Learning

2020 ◽  
Vol 34 (01) ◽  
pp. 979-988
Author(s):  
Wenlin Wang ◽  
Hongteng Xu ◽  
Zhe Gan ◽  
Bai Li ◽  
Guoyin Wang ◽  
...  
Keyword(s):  
Generative Models ◽  
Healthcare Applications ◽  
Convolutional Network ◽  
Learning Tasks ◽  
Proposed Model ◽  
Heterogeneous Learning ◽  
Clinical Topic ◽  
Generative Processes ◽  
Admission Type ◽  
Uniform Manner

We propose a novel graph-driven generative model, that unifies multiple heterogeneous learning tasks into the same framework. The proposed model is based on the fact that heterogeneous learning tasks, which correspond to different generative processes, often rely on data with a shared graph structure. Accordingly, our model combines a graph convolutional network (GCN) with multiple variational autoencoders, thus embedding the nodes of the graph (i.e., samples for the tasks) in a uniform manner, while specializing their organization and usage to different tasks. With a focus on healthcare applications (tasks), including clinical topic modeling, procedure recommendation and admission-type prediction, we demonstrate that our method successfully leverages information across different tasks, boosting performance in all tasks and outperforming existing state-of-the-art approaches.

Sign in / Sign up

Export Citation Format

Share Document