Previous Control Input Research Articles

A Model-Free Online Learning Control for Attitude Tracking of Quadrotors

This paper investigates the problem of attitude tracking in quadrotor unmanned aerial vehicles (UAVs) using a model-free online learning control (MFOLC) scheme. The attitude system, which is represented by unit quaternions, is considered in the presence of uncertain and unknown inertia parameters, time-varying external disturbances, and angular velocity measurement noise. A computationally low-cost control scheme consisting of a model-free baseline controller and a module capable of learning from previous control input is designed. The proposed controller does not require precise inertial parameters and does not involve feedforward terms that use these parameters and true system states. This ensures that the approach can protect the control effort from sensor noise as well as parameter uncertainty. We also show that all the signals in the closed-loop system are uniformly ultimately bounded. Comparative simulations and real-world experiments are conducted for validation, which demonstrate the effectiveness and fine performance of the proposed scheme.

A secure event-based quantized networked control system

This paper focuses on network time delay, network congestion, and network security in the networked control system. A secure event-based quantized networked control system (SEQNCS) is proposed to address these problems. The predictive method in the SEQNCS actively compensates time delay both in the forward channel and feedback channel according to the previous states and control inputs. Both the event-driven strategy and quantizer are introduced into the SEQNCS to alleviate network congestion by the event trigger conditions and quantization density. The encrypted encoding quantizer is constructed in the SEQNCS to implement the protection of data security under deception attacks while quantifying the transmitted data. The proposed SEQNCS achieves desired system performance of the networked control system under time delay and deception attacks while alleviating network congestion. The stability of the SEQNCS is proved by the piecewise linear model approach. Practical experiments are implemented to demonstrate the effectiveness of the proposed SEQNCS.

Symposium Title: The Very Young Brain and the Auditory World

This paper proposes a novel proportional-integral observer (PIO) based high-order data-driven iterative learning control (HODDILC) for nonlinear batch processes with non-repetitive disturbances subject to input constraints. First, an equivalent dynamic linearization data model (DLDM) with an uncertainty term arising from the nonrepetitive disturbances is constructed in the batch direction. Based on the established DLDM along with a gradient estimation algorithm, a data-driven PIO is then designed to estimate the uncertainty term along the batch direction, followed by designing a HODDILC law in terms of the tracking errors over more than one previous batches and control inputs in the previous time instants of the current batch. Using the contraction mapping principle and matrix theory, rigorous convergence analysis is conducted in the iteration domain. A notable advantage of the proposed design is that only input and output measurement data are used rather than a process model. Finally, an illustrative example from the literature is given to demonstrate the effectiveness and merit of the proposed method.

Finite-Dimensional Sampled-Data Control of Fractional-Order Systems

In this letter, we address the design of a finite-dimensional control law based on sampled-data for a class of linear continuous-time fractional-order systems. Using the proposed finite-dimensional controller which only requires a finite number of previous control inputs to compute the current control signal, input-to-state stability (ISS) is established for the corresponding closed-loop systems by considering the hereditary and infinite memory properties of the fractional-order systems. A sufficient condition for determining the control signal storage space is also provided to guarantee certain control performance. Numerical simulations illustrate the effectiveness of the proposed design methods and validate the theoretical results.

PIO Based Data-Driven Iterative Learning Control for Nonlinear Batch Processes with Nonrepetitive Disturbances Subject to Input Constraints

Abstract This paper proposes a novel proportional-integral observer (PIO) based high-order data-driven iterative learning control (HODDILC) for nonlinear batch processes with non-repetitive disturbances subject to input constraints. First, an equivalent dynamic linearization data model (DLDM) with an uncertainty term arising from the nonrepetitive disturbances is constructed in the batch direction. Based on the established DLDM along with a gradient estimation algorithm, a data-driven PIO is then designed to estimate the uncertainty term along the batch direction, followed by designing a HODDILC law in terms of the tracking errors over more than one previous batches and control inputs in the previous time instants of the current batch. Using the contraction mapping principle and matrix theory, rigorous convergence analysis is conducted in the iteration domain. A notable advantage of the proposed design is that only input and output measurement data are used rather than a process model. Finally, an illustrative example from the literature is given to demonstrate the effectiveness and merit of the proposed method.

On low-complexity control design to spacecraft attitude stabilization: An online-learning approach

This paper studies the spacecraft attitude stabilization problem with external disturbances. A new control scheme entitled online-learning control is proposed to achieve a robust, accurate, and simple-structure control algorithm. Compared with the conventional control design, an obvious distinction of the online-learning control algorithm is that it together utilizes the previous control input information and the system's current state information, as if learning experience from previous control input. In contrast, the conventional control scheme does not fully use the existing information and chooses to discard the previous control input information when generating control instructions. Due to the learning strategy, the utility of adaptive- or observer-based tools can be avoided when designing a robust control law, making a simple, effective algorithm, moreover saving system resources. The proposed control law can stabilize the attitude system by achieving the uniformly ultimately bounded convergence.

Short-Term Multi-Vehicle Trajectory Planning for Collision Avoidance

A trajectory planning is a technique that is widely used to automatically generate an expected geometric path that a vehicle needs to follow within a certain future time horizon. We propose a way to generate trajectories of multiple vehicles for a short-term planning with the formal guarantee of the collision avoidance using a SAT (Satisfiability) solver. We assume that each vehicle sends the local status (e.g., speed or position) and its planned trajectory to the server (e.g., road-side units or cloud servers). The server collects the information, and determines if the group of vehicles would encounter any collision within a fixed amount of time. Then, the server calculates the modified trajectories, and sends them back to the vehicles to avoid collisions. Calculating such trajectories is generally a complex problem, which makes it challenging to find them within a short amount of time. Given a driving scenario where multiple vehicles are engaged, we propose two methods to reduce the computation time to generate the trajectories. Firstly, we introduce the grouping method which systematically identifies vehicles that are not engaged in the collision scenario, and removes the related parameters from the SAT solving process. Our case study of the ramp merging driving scenario shows the grouping method saves the computation time by 40% compared to solving the constraints with the original parameters. Secondly, we introduce the collision check point method that chooses some waypoints to maintain their previous control input to reduce the problem complexity. Our experiment shows up to 98% computation time reduction compared to the approach without this method.

Improved Iterative Learning Control Strategy for Surface-Mounted Permanent Magnet Synchronous Motor Drives

In this article, an improved iterative learning control (ILC), which significantly enhances the speed tracking performance in the transient and steady state for a surface-mounted permanent magnet synchronous motor (SPMSM) drive, is proposed. The proposed ILC encapsulates two control terms: the feedback linearization control terms force the speed error to approach zero, and the iterative learning control terms (ILCTs) improve the performance of the control inputs based on the stored data such as previous speed error and previous control input. Unlike the conventional feedback linearization control (FLC), the proposed ILC does not require the exact information of the SPMSM parameters, significantly rejects the periodic and nonperiodic disturbances, and efficiently provides an enhanced speed tracking performance owing to the included ILCTs. Besides, its stability is proven by showing that the speed tracking error asymptotically goes to zero. In comparative studies, the proposed method offers the better transient performance (e.g., faster transient response and smaller overshoot) and steady-state performance (i.e., smaller steady-state error) than the conventional FLC under load and speed step changes. To prove the practicability of the proposed scheme, the proposed ILC is simulated and implemented on a MATLAB/Simulink software and a prototype SPMSM test-bed using TI TMS320F28335 digital signal processor, respectively.

Distributed Repetitive Learning Control for Cooperative Cadence Tracking in Functional Electrical Stimulation Cycling.

Closed-loop control of functional electrical stimulation coupled with motorized assistance to induce cycling is a rehabilitative strategy that can improve the mobility of people with neurological conditions (NCs). However, robust control methods, which are currently pervasive in the cycling literature, have limited effectiveness due to the use of high stimulation intensity leading to accelerated fatigue during cycling protocols. This paper examines the design of a distributed repetitive learning controller (RLC) that commands an independent learning feedforward term to each of the six stimulated lower-limb muscle groups and an electric motor during the tracking of a periodic cadence trajectory. The switched controller activates lower limb muscles during kinematic efficient regions of the crank cycle and provides motorized assistance only when most needed (i.e., during the portions of the crank cycle where muscles evoke a low torque output). The controller exploits the periodicity of the desired cadence trajectory to learn from previous control inputs for each muscle group and electric motor. A Lyapunov-based stability analysis guarantees asymptotic tracking via an invariance-like corollary for nonsmooth systems. The switched distributed RLC was evaluated in experiments with seven able-bodied individuals and five participants with NCs. A mean root-mean-squared cadence error of 3.58 ± 0.43 revolutions per minute (RPM) (0.07 ± 7.35% average error) and 4.26 ± 0.84 RPM (0.1 ± 8.99% average error) was obtained for the healthy and neurologically impaired populations, respectively.

On New Refining Control System for Dephosphorization Using LD Converter

This paper proposes a new control system for dephosphorization of steel-making process using a LD converter. The control system is composed of two steps and it stabilizes low phosphorous content in the hot metal after treatment. The first step creates target concentration of iron oxide in slag for the present charge, which is optimal to achieve effective dephosphorization. The target is obtained as a time series through component analysis of results of previous charges. The second step decides the control inputs for the present charge, oxygen blowing rate and lance height, by calculating a weighted summation of previous control inputs. The weight is based on the difference between the target concentration and the concentration of previous charge. The proposed system was applied to an actual process, and the result shows the high performance of the control at the end of this paper.

Fuzzy Iterative Learning Control of Servo System

According to the repeatability characteristics of servo system, a fuzzy iterative learning controller is proposed which combining advantages of fuzzy control and iterative learning control. Fuzzy control has better robustness, and does not require accurate model of the system, only need the previous experience, the design method is simple. Fuzzy controller is used in the position-loop, but the static error is difficult to eliminate, iterative learning controller use the control error to adjust the previous control input to reduce the error of next time. The combination of these two intelligent control algorithms can improve the control performance of servo system effectively. Simulation results show that the fuzzy control combined with iterative learning control can achieve better control performance in servo system.

Dynamic simulation of vapor-compression cycle using neural networks

In this paper, the thermodynamic modeling of a vapor-compression cycle, based on a neural network approach is presented. A generalized radial basis function is used for the network, which takes previous control inputs and previous states as the network input and generates the predicted current state as the network output. The trained network is validated by non-trained data and shows all the process characteristics of a vapor-compression cycle for an air-to-water heat pump to a satisfactory degree.

To Zero or to Hold Control Inputs With Lossy Links?

This technical note studies the linear quadratic (LQ) performance of networked control systems where control packets are subject to loss. In particular we explore the two simplest compensation strategies commonly found in the literature: the zero-input strategy, in which the input to the plant is set to zero if a packet is dropped, and the hold-input strategy, in which the previous control input is used if packet is lost. We derive expressions for computing the optimal static gain for both strategies and we compare their performance on some numerical examples. Interestingly, none of the two can be claimed superior to the other, even for simple scalar systems, since there are scenarios where one strategy performs better then the other and scenarios where the converse occurs.

PD-TYPE ON-LINE LEARNING CONTROL FOR SYSTEMS WITH STATE UNCERTAINTIES AND MEASUREMENT DISTURBANCES

In this paper, a PD-type on-line learning control (OLC) is proposed for the tracking problem in a class of nonlinear time-varying systems with state uncertainties and measurement disturbances. In this proposed OLC algorithm, we use a combination of PD feedback control and feedforward control that is based on the previous control input profiles in an iterative manner. The proposed control scheme has a simple and straightforward structure that leads to its easy implementation. The uniform boundary of the tracking error is established based on the convergence analysis. It is demonstrated that the proposed control algorithm is robust in coping with the state uncertainties and measurement disturbances that result from the system modeling inaccuracies and the initial state errors. Further, one unique feature of the proposed PD-type on-line learning control is that the controller allows a wide range of control gains to be selected to improve its convergence rate. Simulation studies demonstrate the effectiveness of the proposed control algorithm.

冗長関節ロボットの作業空間での繰り返し学習制御法

This paper proposes an iterative learning control (ILC) method for robots with redundant joints to acquire the desired control input signals that produce an endpoint trajectory specified in task space. The learning update law of control input signals is constructed only in task space by modifying the previous control input through adding linearly endpoint position and velocity trajectory errors. Although the dimension of the task space is strictly less than the DOF (Degree-of-freedom) of robots, the proposed method need neither consider any inverse kinematics problem nor introduce any cost function to be optimized and determine the inverse kinematics uniquely. Convergence of trajectory trackings to the specified one is shown by numerical simulations in both cases (1) free-endpoint motion and (2) constraint-endpoint motion with specified contact force. A theoretical proof of convergences is also given on the basis of a dynamics linearized around a joint trajectory when the endpoint tracks a desired trajectory.

Credit assigned CMAC and its application to online learning robust controllers

In this paper, a novel learning scheme is proposed to speed up the learning process in cerebellar model articulation controllers (CMAC). In the conventional CMAC learning scheme, the correct numbers of errors are equally distributed into all addressed hypercubes, regardless of the credibility of the hypercubes. The proposed learning approach uses the inverse of learned times of the addressed hypercubes as the credibility (confidence) of the learned values, resulting in learning speed becoming very fast. To further demonstrate online learning capability of the proposed credit assigned CMAC learning scheme, this paper also presents a learning robust controller that can actually learn online. Based on robust controllers presented in the literature, the proposed online learning robust controller uses previous control input, current output acceleration, and current desired output as the state to define the nominal effective moment of the system from the CMAC table. An initial trial mechanism for the early learning stage is also proposed. With our proposed credit-assigned CMAC, the robust learning controller can accurately trace various trajectories online.

Adapthre Control of Servo Mechanism with a Two-Layered Neural Network.

A two-layered neural network was applied to adaptive control of a servo mechanism. The two-layered neural network is simple and can be built in a structure corresponding to inverse dynamics of a controlled plant. A demand signal, previous output signals of a plant, and previous control input signals are fed into the network. The output of the network is control input and is fed into a controlled plant. Initial weights can be set using information on nominal plant parameters. The weights are updated by a back propagation strategy with a normalized learning rate which may specify the rate of convergence of learning. Through experiments using an electrohydraulic servo motor system, the validity of the neural network controller was examined. Adaptation function of the neural network was demonstrated.