Adaptive Boundary Research Articles

Ada-boundary: accelerating DNN training via adaptive boundary batch selection

Neural networks converge faster with help from a smart batch selection strategy. In this regard, we propose Ada-Boundary, a novel and simple adaptive batch selection algorithm that constructs an effective mini-batch according to the learning progress of the model. Our key idea is to exploit confusing samples for which the model cannot predict labels with high confidence. Thus, samples near the current decision boundary are considered to be the most effective for expediting convergence. Taking advantage of this design, Ada-Boundary maintained its dominance for various degrees of training difficulty. We demonstrate the advantage of Ada-Boundary by extensive experimentation using CNNs with five benchmark data sets. Ada-Boundary was shown to produce a relative improvement in test errors by up to 31.80% compared with the baseline for a fixed wall-clock training time, thereby achieving a faster convergence speed.

A phase I dose-finding design with incorporation of historical information and adaptive shrinking boundaries.

Although many novel phase I designs have been developed in recent years, few studies have discussed how to incorporate external information into dose-finding designs. In this paper, we first propose a new method for developing a phase I design, Bayesian optimal interval design (BOIN)[Liu S et al. (2015), Yuan Y et al. (2016)], for formally incorporating historical information. An algorithm to automatically generate parameters for prior set-up is introduced. Second, we propose a method to relax the fixed boundaries of the BOIN design to be adaptive, such that the accumulative information can be used more appropriately. This modified design is called adaptive BOIN (aBOIN). Simulation studies to examine performances of the aBOIN design in small and large sample sizes revealed comparable performances for the aBOIN and original BOIN designs for small sample sizes. However, aBOIN outperformed BOIN in moderate sample sizes. Simulation results also showed that when historical trials are conducted in settings similar to those for the current trial, their performance can be significantly improved. This approach can be applied directly to pediatric cancer trials, since all phase I trials in children are followed by similar efficient adult trials in the current drug development paradigm. However, when information is weak, operating characteristics are compromised.

An Adaptive Trust Boundary Protection for IIoT Networks Using Deep-Learning Feature-Extraction-Based Semisupervised Model

The rapid development of Internet of Things (IoT) platforms provides the industrial domain with many critical solutions, such as joint venture virtual production systems. However, the extensive interconnection of industrial systems with corporate systems in industrial Internet of Things (IIoT) networks exposes the industrial domain to severe cyber risks. Because of many proprietary multilevel protocols, limited upgrade opportunities, heterogeneous communication infrastructures, and a very large trust boundary, conventional IT security fails to prevent cyberattacks against IIoT networks. Recent secure protocols, such as secure distributed network protocol (DNP 3.0), are limited to weak hash functions for critical response time requirements. As a complementary, we propose an adaptive trust boundary protection for IIoT networks using a deep-learning, feature-extraction-based semisupervised model. Our proposed approach is novel in that it is compatible with multilevel protocols of IIoT. The proposed approach does not require any manual effort to update the attack databases and can learn the rapidly changing natures of unknown attack models using unsupervised learnings and unlabeled data from the wild. Therefore, the proposed approach is resilient to emerging cyberattacks and their dynamic nature. The proposed approach has been verified using a real IIoT testbed. Extensive experimental analysis of the attack models and results shows that the proposed approach significantly improves the identification of attacks over conventional security control techniques.

Disturbance observer‐based adaptive boundary iterative learning control for a rigid‐flexible manipulator with input backlash and endpoint constraint

SummaryIn this article, an observer‐based adaptive boundary iterative learning control law is developed for a class of two‐link rigid‐flexible manipulator with input backlash, the unknown external disturbance, and the endpoint constraint. To tackle the backlash nonlinearities and ensure the vibration suppression, the disturbance observers based upon the iterative learning conception are considered in the adaptive boundary control design. A barrier Lyapunov function is incorporated with boundary control law to restrict the endpoint state. Based on the defined barrier composite energy function, the tracking angle error convergence of the rigid part is guaranteed, and the vibrations of the flexible part are suppressed through the rigorous analysis. Finally, a numerical simulation is provided to illustrate the effectiveness of the proposed control.

Stabilisation of a Flexible Spacecraft Subject to External Disturbance and Uncertainties

This paper addresses the problems of vibration reduction and attitude tracking for a flexible spacecraft subject to external disturbances and uncertainties. Based on Hamilton’s principle, flexible spacecraft is modelled by a coupled nonlinear partial differential equation with ordinary differential equations. Adaptive boundary control scheme is adopted to stabilize the vibration displacement of flexible appendage into a small neighbourhood of original position and simultaneously maintain attitude angle within the desired angle region. Two disturbance adaptive laws are constructed to attenuate the effect of unknown external disturbances. The well posedness of the controlled system is proven by using the semigroup theory. The proposed adaptive boundary control scheme can guarantee the uniform boundedness of the closed-loop system. Numerical simulation results illustrate the effectiveness of the proposed control scheme.

An adaptive volumetric flux boundary condition for lattice Boltzmann methods

This paper presents a spatially and temporally adaptive boundary condition to specify the volumetric flow rate for lattice Boltzmann methods. The approach differs from standard velocity boundary conditions because it allows the velocity to vary over the boundary region provided that the total flux through the boundary satisfies a prescribed constraint, which is a typical scenario for laboratory experimental studies. This condition allows the boundary pressure to adjust dynamically to yield a specified boundary flow rate as a means to avoid unphysical mismatch between the boundary velocity and the interior flow field that can arise when a standard velocity boundary condition is applied. The method is validated for simulation of one- and two-fluid flow in complex materials, with conditions determined to match typical experiments used to study flow in porous media.

Adaptive boundary control for wave PDEs with unknown in-domain/boundary disturbances and system parameter

In this paper, we develop an adaptive controller tasked with tracking reference velocity for a boundary of a one-dimensional wave PDE. The system parameter associated with that boundary is further assumed to be unknown. The wave equation is subject to unknown, in-domain and boundary, sinusoidal disturbances. Furthermore, boundary states are assumed to be the only measurements available for feedback. The problem is initially cast into a familiar LTI form describing dynamics at the bottom boundary. This uncertain linear system is characterized by an unknown parameter, delayed input, and an unknown harmonic disturbance. From there, we parametrize the disturbance and use backstepping design, via transport PDEs, to construct an adaptive boundary controller. Stability analysis is given for the closed loop system and the convergence of the error state is proven. The paper concludes with a numerical example illustrating the performance of the derived controller.

Adaptive boundary control of a vibrating cantilever nanobeam considering small scale effects

This paper presents vibration control analysis for a cantilever nanobeam system. The dynamics of the system is obtained by the non-local elastic relationship which characterizes the small scale effects. The boundary conditions and governing equation are respectively expressed by several ordinary differential equations (ODE) and a partial differential equation (PDE) with the help of the Hamilton’s principle. Model-based control and adaptive control are both designed at the free end to regulate the vibration in the control section. By employing the Lyapunov stability approach, the system state can be proven to be substantiated to converge to zero’s small neighbourhood with appropriate parameters. Simulation results illustrate that the designed control is feasible for the nanobeam system.

An adaptive failure boundary approximation method for reliability analysis and its applications

In practical engineering problems, accurate reliability assessment often is computationally expensive with time-consuming numerical models or simulation models. How to obtain an accurate reliability index with a fewer number of calls to original performance function in reliability analysis has become an important challenge. For the purpose of reducing the computational cost in reliability analysis, this work develops an adaptive failure boundary approximation method (AFBAM) by combining Kriging and uniform sampling with a new adaptive learning strategy. The proposed AFBAM makes full use of the binary classification feature of reliability analysis in the way that the failure boundary of the original model can be efficiently approximated. The number of experimental design samples is constantly updated by selecting informative samples with the proposed learning strategy. In order to ensure classification accuracy of the constructed Kriging model, a new stopping criterion is designed based on average misclassification probability and misclassification ratio. The proposed AFBAM technically makes reliability evaluation phase independent of adaptive iterative process, which greatly improves the efficiency of model refinement phase. At last, five examples involving nonlinearity problem, small failure probability problem and practical engineering problem are tested to verify the efficiency of the proposed AFBAM.

Solution of Stokes flow in complex nonsmooth 2D geometries via a linear-scaling high-order adaptive integral equation scheme

We present a fast, high-order accurate and adaptive boundary integral scheme for solving the Stokes equations in complex—possibly nonsmooth—geometries in two dimensions. We apply the panel-based quadratures of Helsing and coworkers to evaluate to high accuracy the weakly-singular, hyper-singular, and super-singular integrals arising in the Nyström discretization, and also the near-singular integrals needed for flow and traction evaluation close to boundaries. The resulting linear system is solved iteratively via calls to a Stokes fast multipole method. We include an automatic algorithm to “panelize” a given geometry, and choose a panel order, which will efficiently approximate the density (and hence solution) to a user-prescribed tolerance. We show that this adaptive panel refinement procedure works well in practice even in the case of complex geometries with large number of corners, or close-to-touching smooth curves. In one example, for instance, a model 2D vascular network with 378 corners required less than 200K discretization points to obtain a 9-digit solution accuracy.

An improved land cover classification using polarization signatures for PALSAR 2 data

Land cover classification in mixed land cover scenarios is challenging with PolSAR data. Polarimetric decomposition techniques are most popular methods for PolSAR data classification in recent times. These techniques focus on identification of dominant scattering phenomena and hence result in sub-optimal classification in mixed land cover scenarios. Alternatively, polarization signatures (PSs) are good illustrations of SAR target responses as they depict a detailed physical information from target backscatter. Researchers have successfully utilized SAR PSs for land cover (LC) classification. Some reports suggested utilizing correlation between observed PSs and standard target PSs as features for LC classification. This paper presents a study on improved utilization of PSs for optimal LC classification in mixed class scenarios. First, PS based SAR features are derived using fully polarimetric SAR data. The features represent a degree of similarity between observed and standard PSs. The derived features are termed as polarization signatures correlation features or PSCFs. The novel PSCFs are analyzed, evaluated and compared with decomposition based features for the purpose of LC classification. Classification performance indicators highlight potential of PSCFs for mixed LC classification problems. Therefore, further an adaptive and optimal LC class boundary estimation approach for LC classification is proposed and developed. Observed PSs and reference LC class PS statistics are used to build empirical models between classification performance indicators and LC class boundaries. The empirical models are optimized using the evolutionary genetic algorithm to maximize classification performance. A decision tree is constructed based on the optimal class boundaries to prepare LC classification. The proposed classification approach is compared with some recent popular classifiers and comparison suggests that the proposed approach provides satisfactory results for mixed LC classification scenarios.

Adaptive boundary determination for network reduction in long‐term voltage stability analyses

In this study, a simple, fast, and adaptive network reduction procedure is presented for long-term voltage stability analyses, to speed up the simulation time. The proposed procedure consists of two parts including disturbance detection based on data from synchronised measurement devices; and the rest, up to the final estimated values, based on simulations and calculations. In turn, simulations and calculations result in an adaptive dynamic-oriented determination of external sub-network (according to the disturbance occurred) and substituting an appropriate equivalent for that subsystem. The external subsystem is determined without any need to measure post-disturbance data, while the detection criteria are calculated by simulations at 0.8 s after disturbance occurrence. The results have shown considerable reduction in the simulation time while providing acceptable accuracy in delivering the time of voltage instability occurrences, especially when applying the proposed error compensation procedure.

Novel visual tracking approach via ant lion optimiser

Ant lion optimiser (ALO) is a new nature-inspired swarm intelligence optimisation algorithm that mimics the hunting mechanism of antlions in nature. ALO has been proved to have the merits of high exploitation and convergence speed benefiting from adaptive boundary shrinking mechanism and elitism. In this work, visual tracking is expressed as searching for object in whole search space by interaction between antlions and ants. A novel ALO-based visual tracking framework is proposed and the adaptation and sensitivity of the parameters in ALO are discussed to improve tracking performance. In addition, considering that ALO tracker needs a lot of iteration consumption, kernel correlation filter with deep feature is integrated into the ALO tracking framework (ALOKCF) to improve track efficiency. Extensive experimental results prove that the ALO tracker is very competitive compared to other trackers, especially for abrupt motion tracking. At the same time, two visual tracking benchmarks are used to verify ALOKCF tracker achieves state-of-the-art performance.

Microsecond Molecular Dynamics Simulations of Proteins Using a Quasi-Equilibrium Solvation Shell Model.

We describe the development and implementation of a quasi-equilibrium hydration shell model of biomolecular solvation with adaptive boundaries. Applying the model to microsecond-long molecular dynamics simulations of several protein systems of varying complexity, we find that the model simulation results are of comparable quality to those obtained from simulations of fully solvated systems, but at a reduced computational cost. We discuss the dominant sources of error in the model and outline directions for future improvements.

Adaptive isogeometric boundary element methods with local smoothness control

In the frame of isogeometric analysis, we consider a Galerkin boundary element discretization of the hyper-singular integral equation associated with the 2D Laplacian. We propose and analyze an adaptive algorithm which locally refines the boundary partition and, moreover, steers the smoothness of the NURBS ansatz functions across elements. In particular and unlike prior work, the algorithm can increase and decrease the local smoothness properties and hence exploits the full potential of isogeometric analysis. We prove that the new adaptive strategy leads to linear convergence with optimal algebraic rates. Numerical experiments confirm the theoretical results. A short appendix comments on analogous results for the weakly-singular integral equation.

A Robust Adaptive Model Reference Impedance Control of a Robotic Manipulator With Actuator Saturation

This paper develops a robust adaptive model reference impedance controller for an n-link robotic manipulator with parameter uncertainties, actuator saturation, and imprecise force sensor measurement. In order to handle the system uncertainties, the adaptive technique and prediction error are used. To reduce calculation complexity, the adaptation law for robot parameters estimation is designed in the joint space. Actuator saturation is also overcome by an auxiliary system, and imprecise force sensor measurement is compensated with utilization of adaptive bounding technique. Moreover, the joint acceleration vector problem, which is nonmeasurable, is covered by the filtered regressor matrix. Using Lyapunov's direct method, as long as two conditions are satisfied the proposed controller ensures the exponential stability of all the signals of the closed loop system; otherwise, the closed loop system is globally stable. Finally, simulations are performed on prototyped robots to analyze and evaluate the established controller as well as to demonstrate the effectiveness of the proposed scheme.

An ISPH with k–ε closure for simulating turbulence under solitary waves

In this paper an Incompressible Smoothed Particle Hydrodynamics (ISPH) method solving the 2D RANS (Reynolds Averaged Navier-Stokes) equations with the k–ε turbulence closure is constructed. In the present model, the concept of “massless ISPH” utilizing the definition of “particle density” (number of computational particles within unit volume) is stressed. The skills of this numerical model are tested by applying to two laboratory experiments: (1) A non-breaking solitary wave propagating over a bottom-mounted barrier and (2) a solitary wave breaking on a 1 on 50 slope. In the former case flow separation occurs behind the barrier as the wave crest passes by and a vortex is generated, which later interacts with free surface causing breaking. In the latter case wave breaking and bottom friction both generate significant turbulence. For both cases, the effects of initial seeding of turbulent kinetic energy, required in the k–ε model, are studied and it is concluded that initial values of O (10−10) to O (10−8) m2/s2 should be used. An adaptive wall boundary condition for k–ε turbulence model is employed to avoid the unrealistic production of turbulence near the wall boundary. The numerical results, in terms of free surface profile, mean velocity field, vorticity field, turbulent kinetic energy and turbulent shear stress, are compared with experimental data. Very reasonable agreement is observed. This paper presents the first comprehensively validated 2D ISPH model with the k–ε turbulence closure, which can be applied to transient free surface wave problems.

A dual adaptive fault-tolerant control for a quadrotor helicopter against actuator faults and model uncertainties without overestimation

This paper presents a dual adaptive fault-tolerant control strategy for a quadrotor helicopter based on adaptive sliding mode control and adaptive boundary layer. Within the proposed adaptive control strategy, both model uncertainties and actuator faults can be compensated without the knowledge of the uncertainty bounds and fault information. By virtue of the proposed adaptive control scheme, the minimum discontinuous control gain is adopted, which significantly reduces the control chattering effect. As compared to the existing adaptive sliding mode control schemes in the literature, larger actuator faults can be tolerated by employing the proposed control scheme while suppressing control chattering. Moreover, boundary layer is used to smoothen control discontinuity and further eliminate control chattering. Nevertheless, the choice of boundary layer thickness is a trade-off between system stability and tracking accuracy. By explicitly considering this fact, an adaptive boundary layer is developed and synthesized with the proposed adaptive control framework to ensure stability and tracking accuracy of the considered system. When the control parameter tends to be overestimated, the thickness of boundary layer can be appropriately adjusted to avoid control parameter overestimation. Simulation and experimental tests of a quadrotor helicopter are both conducted to validate the effectiveness of the proposed control scheme. Its advantages are demonstrated in comparison with a conventional adaptive sliding mode control scheme.

Adaptive boundary observer design for linear hyperbolic systems; Application to estimation in heat exchangers

In this work, we consider the estimation of temperature along the pipes of concentric heat exchanger tubes in which CO2 is the working fluid. The transport phenomenon is modeled using (2x2) linear hyperbolic partial differential equations, with one rightward equation for the hot flow and one leftward equation for the cold flow. Both flows exchange energy through the wall interface, which physically induces a coupling between the two dynamics. Our objective is to estimate the temperature distribution along the flows from measurements at the tubes boundaries only. In this framework, we propose an adaptive boundary observer that can estimate not only the full state of the system, but also unknown in-domain parameters. The design is based on transforming the error system via a finite-dimensional backstepping-like transformation into a desired filter-based system, for which standard backstepping observer techniques and adaptation laws can be used. The theoretical results are evaluated against the temperature measurements taken from a CO2 refrigeration apparatus built at CERN, Switzerland.

Adaptive output regulator for wave PDEs with unknown harmonic disturbance

This paper focuses on developing boundary controller for a wave PDE with two disturbances; (i) spatially varying and time-invariant in-domain disturbance, (ii) time-varying sinusoidal boundary disturbance. We consider the case that the sensor is collocated with the actuator and harmonic disturbance is anti-collocated with the sensor/actuator. We aim to regulate the rate of change of the opposite boundary, relative to the actuator, around a given constant reference. Towards that aim, we reformulate the opposite boundary equation as a linear system with known parameters, the sinusoidal disturbances and simultaneous input/output delays. In this way, the control input and sensor become collocated with the disturbance. Then, we parametrize the disturbance and represent the delays as a transport PDE. We develop an observer-based adaptive boundary controller for the ODE–PDE cascade system. We prove that all signals are bounded and the rate of change of the opposite boundary tracks a given constant reference. The performance of the controller is demonstrated in a numerical simulation.