Presence Of Unknown Disturbances Research Articles

Filtered Dynamic Inversion for Vibration Control of Structures With Uncertainty

This paper presents a controller for uncertain structures that are minimum phase and potentially subject to unknown-and-unmeasured disturbances. The controller combines dynamic inversion with a low-pass filter to yield a single-parameter high-gain-stabilizing controller. Filtered dynamic inversion requires limited model information, is independent of system order, requires only output feedback, and makes the average power of the command following error arbitrarily small despite the presence of unknown disturbances. The controller is applied to structures modeled by finite-dimensional vector second-order systems with unknown and arbitrarily large order. We also present an adaptive filtered-dynamic-inversion controller, which uses a high-gain adaptive law to increase the controller parameter until the desired performance is achieved. Finally, the controller is extended to vector second-order nonlinear systems, in which case full-state feedback may be required. Examples are given to demonstrate the application and performance of the controller.

A Robust Adaptive Control Law for Satellite Formation Flying

This paper considers the relative control problem of earth-orbiting formations based on the leader-follower scheme. The full nonlinear dynamics describing the relative motion between the leader and the follower are derived. The non-perturbed form of the full dynamics is utilized to formulate the optimal reference trajectory generation problem which is transformed into a nonlinear programming problem for numerical solution by Gauss pseudospectral method. A robust adaptive controller is designed based on the Lyapunov method. Despite the presence of unknown disturbances, unknown reference orbit parameters, unknown control of the leader, and unknown mass of the follower, this controller can guarantee the global uniform ultimate boundedness of the closed-loop system using only relative measurements. The ultimate bound of the tracking error can be made arbitrarily small by proper choice of controller parameters. Simulation results are presented to illustrate the efficiency of the main results in this paper.

A HOSM control algorithm to stabilize the quadrotor vehicle with experimental validation

A High Order Sliding Mode control technique to stabilize the quadrotor vehicle is presented in this paper. The simplified nonlinear equations are evoked to develop the control algorithms. The closed-loop system using this sliding mode methodology is robust even in presence of unknown disturbances and uncertainties parameters in the model. The control scheme is validated in simulations and some experiments are developed to prove the experimental feasibility of the control strategy. Simulation and experimental results demonstrate the well performance of the proposed algorithms.

Force control in a parallel manipulator through virtual foundations

An overwhelming controller provides robustness against uncertain parameters, disturbances and un-modelled dynamics. A simplified inverse dynamics model is used in the present paper to develop an overwhelming controller for a parallel manipulator (a Stewart platform), where the controller accommodates modelling uncertainties such as the simplifications made for development of the inverse model in order to improve the computational efficiency. Such a control strategy leads to good trajectory tracking accuracy in the presence of unknown disturbances. However, in addition to trajectory tracking performance, the controller for a Stewart platform should also be able to control or limit the interaction forces in applications such as robot assisted surgery and low-impact docking. The environmental forces can be accommodated during the interaction period by modulating the impedance at the interface of manipulator and environment through virtual flexible foundations. The positional error induced during the force control phase can be recovered during the free flight or idle phase. While this approach has been used successfully in the past to control serial manipulators, the closed loop kinematic architecture in a parallel manipulator introduces many difficulties. This paper proposes a modified overwhelming control scheme for parallel manipulators with compensations for interaction force control and positional error recovery. Bond graph modelling is used as an integrated model and controller development tool. Force controlled machining on a spherical surface, which is akin to a surgical operation, is considered as an example application of the developed control strategy. The simulation results from the bond graph model of the controlled Stewart platform are presented to demonstrate the performance of the developed hybrid position force controller.

Passivity-based control applied to the dynamic positioning of ships

A family of passivity-based controllers for dynamic positioning of ships is presented. The authors exploit the idea of shaping the energy function of the closed-loop system to obtain different formulations of the passivity-based control law using the interconnection and damping assignment-passivity-based control (IDA-PBC) methodology. A salient feature of this study is that the proposed control laws are output feedback controllers and the relative velocity measurement is not required. First, we design and analyse two static controllers which can be seen as a non-linear version of the conventional proportional-derivative (PD) controllers. In presence of unknown disturbances, these controllers do not provide the desired regulation properties. To remove this discrepancy we propose, also in the context of the IDA-PBC technique, a dynamic extension of the system and obtain two new controllers that have the desired regulation properties. These new control laws can be seen as a non-linear version of the conventional proportional-integral-derivative (PID) controllers. Simulations are included to validate the theoretical results.

Batch-to-batch control of particle size distribution in cobalt oxalate synthesis process based on hybrid model

A hybrid model based batch-to-batch control strategy is proposed for control of particle size distribution (PSD) in cobalt oxalate synthesis process. In order to enhance the model prediction accuracy and generalization capability, a hybrid modeling approach for cobalt oxalate synthesis process in cobalt hydrometallurgy is developed by combining simplified first principle model with PLS model. The simplified first principle model that captures the dominant characteristics of the synthesis process is built to describe PSD evolution. The PLS model is utilized to compensate the unmodeled characteristic of the simplified first principle model and to enhance model generalization capability. Due to the repetitive nature of the process, a batch-to-batch control strategy based on hybrid model is then presented to design the operating policy that drives the process to a target PSD. Applications to a simulated cobalt oxalate synthesis process demonstrate that the proposed approach can improve process performance from batch to batch in the presence of unknown disturbances.

Integral sliding mode compensator for load pressure control of die-cushion cylinder drive

This paper discusses problems of the load pressure control of electro-hydraulic drives in a presence of unknown disturbances and parametrical uncertainties. In many applications, the standard, linear control methods do not assure a satisfactory dynamical behavior and are likely to fail if a working point or the system properties change drastically. In order to guarantee a desired robustness and precision of the closed-loop system, a combination of the input–output linearization technique with the integral sliding mode is proposed. The structure of the presented controller is very simple, and can be easily implemented in standard industrial PLC's. The commissioning and tuning of the controller are uncomplicated and the adjustment procedure is partially automated. Conducted tests confirm a very good and robust performance of the closed loop control. The results are compared with those obtained with conventional linear controllers (P and PI).

Iterative Learning Control of a Reactive Polymer Composite Moulding Process

This paper presents an iterative learning control strategy for a reactive polymer composite moulding process using batch-wise linearised models that are identified from process operational data. In rapid thermal moulding processes, the degree of cure is controlled by the applied mould temperature. Due to the model plant mismatches and the presence of unknown disturbances, a pre-set (off-line optimised) moulding temperature profile would not always lead to the desired degree of cure. The repetitive nature of batch moulding process allow the information of the previous batch (or cycle) being used to enhance the operation of the current batch through iterative learning control. The control policy updating is calculated by solving an optimisation problem using a model linearised around a reference batch. In order to cope with process nonlinearities, process variations, and disturbances, the reference batch is taken as the immediate previous batch and the model is re-identified from the updated historical batch data using principal component regression. The proposed method is tested on a simulated reactive polymer composite moulding process.

Toward Automation of Friction Stir Welding Through Temperature Measurement and Closed-Loop Control

The objectives of this work are to determine an accurate temperature feedback strategy and to develop a closed-loop feedback control system for temperature in friction stir welding (FSW). FSW is a novel joining technology enabling welds with excellent metallurgical and mechanical properties, as well as significant energy consumption and cost savings. However, numerous parameter and condition variations are present in the FSW production environment that can adversely affect weld quality, which has made extensive automation of this process impossible to date. To enable large scale automation while maintaining weld quality, techniques to control the FSW process in the presence of unknown disturbances must be developed. One process variable that must be controlled to maintain uniform weld quality under the inherent workpiece variability (thermal constraints, material properties, geometry, etc.) is the weld zone temperature. Our hypothesis is that the weld zone temperature can be controlled, which can help in controlling the weld quality. A wireless data acquisition system was built to measure temperatures at the tool-workpiece interface. A thermocouple was placed in a through hole right at the interface of tool and workpiece so that the tip is in contact with the workpiece material. This measurement strategy reveals temperature variations within a single rotation of the tool in real time. In order to automate the system, a first order process model with transport delay was experimentally developed that captures the physics between spindle speed and measured interface temperature. The model has a time constant of 110 ms and a delay time of 85 ms. Using this temperature measurement technique, a closed-loop temperature control system with a bandwidth of 0.3 Hz was developed. Interface temperatures in the range from 555 °C to 575 °C were commanded to an integral controller, which regulated the spindle speed between 850 rpm and 1250 rpm to adjust the heat generation and achieve the desired interface temperatures in 6061-T6 aluminum. To simulate changes in thermal boundary conditions, backing plates of different thermal diffusivities were found to effectively alter the heat flow, hence, weld zone temperature. The integral controller that manipulates spindle speed is applied when welding during these intentionally introduced weld disturbances. The measured temperature stayed within ±5 °C after introducing the disturbance, compared to a 50 °C change in temperature when no control was applied.

Fault and disturbance reconstruction in non-linear systems using a network of interconnected sliding mode observers

A new technique for fault diagnosis and estimation of unknown inputs in a class of non-linear systems is presented in this study. The novelty of the approach is based on utilisation of a network of two interconnected sliding mode observers, the first is used for fault diagnosis and the second is used for estimation of unknown inputs. The two observers exchange information about their respective reconstructed signals online and in real time. Conditions and proofs of conversion are presented. A salient feature of the proposed approach is that the state trajectories do not leave the sliding manifold even in presence of unknown disturbances and faults. This allows for faults and unknown inputs to be reconstructed based on information retrieved from the equivalent output error injection signal.

Input‐output nominalization of linear systems with slow‐varying uncertainties

Purpose – The purpose of this paper is to present a method to compensate slow varying disturbances and plant parameter drifts using a simple yet robust algorithm called input‐output nominalization.Design/methodology/approach – In case of known uncertainties, an analytical expression of pre‐computed feed‐forward compensation command is derived. In presence of unknown disturbances and parameter drifts, the control algorithm uses a proportional‐integrative estimator‐based nominalizer. It creates a nominalizing signal, reflecting the deviation of the system from its nominal form using plant input and output. The signal is subtracted from the nominal controller output to cancel the uncertainty and disturbances effects.Findings – As a result, the uncertain plant and the nominalizer quickly converge to the nominal plant. Therefore, a simple controller tuned according to the nominal plant can be used despite the disturbances and parameter drifts and a nominal response is always obtained. Simulation and experiment...

Unbiased Minimum-Variance Filter for State and Fault Estimation of Linear Time-Varying Systems with Unknown Disturbances

This paper presents a new recursive filter to joint fault and state estimation of a linear time-varying discrete systems in the presence of unknown disturbances. The method is based on the assumption that no prior knowledge about the dynamical evolution of the fault and the disturbance is available. As the fault affects both the state and the output, but the disturbance affects only the state system. Initially, we study the particular case when the direct feedthrough matrix of the fault has full rank. In the second case, we propose an extension of the previous case by considering the direct feedthrough matrix of the fault with an arbitrary rank. The resulting filter is optimal in the sense of the unbiased minimum-variance (UMV) criteria. A numerical example is given in order to illustrate the proposed method.

Output feedback sliding mode control in the presence of unknown disturbances

This paper examines the use of a sliding mode controller and an equivalent output injection sliding mode observer to control a nonlinear plant in the presence of an unknown disturbance. With only the measured output available for feedback, the observer reconstructs the full state in finite time and provides an estimate of the unknown disturbance. The controller ensures asymptotic reference trajectory tracking, or alternatively asymptotic convergence to the origin for the plant. Asymptotic stability of the closed loop system is proved under a set of nonrestrictive assumptions on the plant and the disturbance. These results are verified numerically through simulation.

Batch-to-batch optimal control of a batch polymerisation process based on stacked neural network models

A neural network based batch-to-batch optimal control strategy is proposed in this paper. In order to overcome the difficulty in developing mechanistic models for batch processes, stacked neural network models are developed from process operational data. Stacked neural networks have enhanced model generalisation capability and can also provide model prediction confidence bounds. However, the optimal control policy calculated based on a neural network model may not be optimal when applied to the true process due to model plant mismatches and the presence of unknown disturbances. Due to the repetitive nature of batch processes, it is possible to improve the operation of the next batch using the information of the current and previous batch runs. A batch-to-batch optimal control strategy based on the linearisation of stacked neural network model is proposed in this paper. Applications to a simulated batch polymerisation reactor demonstrate that the proposed method can improve process performance from batch to batch in the presence of model plant mismatches and unknown disturbances.

SLIDING MODE CONTROL OF LINEAR SYNCHRONOUS MOTOR SERVODRIVE

This paper proposes a method to design sliding mode control (SMC) for the position tracking control of a permanent magnet linear synchronous motor (PMLSM). The controller is designed to achieve accurate performance in the presence of unknown disturbance and parameter uncertainties. The components of the control law, the equivalent component and the robust component, are designed such that the nominal system exhibits desirable dynamics and the reaching condition is guaranteed. Switching surface is designed based on pole placement and generalized matrix inverse. Simulation results show that the suggested procedure provides high performance dynamic characteristics and robust with regard to external disturbance and parameter variations.

A Dual-Stage Planar Cable Robot: Dynamic Modeling and Design of A Robust Controller with Positive Inputs

Abstract Cable robots have potential usage for loading and unloading of cargo in shipping industries. A novel six-degrees-of-freedom two-stage cable robot has been proposed by NIST for skin-to-skin transfer of cargo. In this paper, we look at a planar version of this two-stage cable robot. The disturbance motion from the sea is considered while modeling the dynamics of robot. The problem of robust control of the end-effector in the presence of unknown disturbances, along with maintaining positive tensions in the cables, is tackled using redundancy of cables in the system. Simulation results show the effectiveness of the control strategy.

AN ASYMPTOTIC SECOND‐ORDER SMOOTH SLIDING MODE CONTROL

ABSTRACTPresented is a method of smooth sliding mode control design to provide for an asymptotic second‐order sliding mode on the selected sliding surface. The control law is a nonlinear dynamic feedback that in absence of unknown disturbances provides for an asymptotic second‐order sliding mode. Application of the second‐order disturbance observer in a combination with the proposed continuous control law practically gives the second‐order sliding accuracy in presence of unknown disturbances and discrete‐time control update. The piecewise constant control feedback is “smooth” in the sense that its derivative numerically taken at sampling rate does not contain high frequency components. A numerical example is presented.

A MULTIPLE-LOOP SLIDING MODE CONTROL SYSTEM WITH SECOND-ORDER BOUNDARY LAYER DYNAMICS

Presented is a method of continuous sliding mode control design to provide for the second-order sliding mode on the selected sliding surfaces in a three-loop control system, where two outer-loop virtual control signals must be smooth enough to be tracked in the inner loops. The control law in the vicinity of the sliding surface is a nonlinear dynamic feedback that in absence of unknown disturbances provides for finite-time convergence of the second-order reaching phase dynamics. The controller with a second-order disturbance observer in a combination with the proposed continuous dynamic feedback attracts the system trajectories to boundary layers around two sliding surfaces of the outer control loops with the second-order sliding accuracy in presence of unknown disturbances and the discrete-time control update.

System and Disturbance Identification for Feedforward and Feedback Control Applications

This paper examines the problem of system identie cation in the presence of unknown disturbances. Specie cally, weconsiderthecasewherethedisturbanceeffect on theoutputismodeled explicitlyand variouswaysto separateit from the system disturbance-free dynamics. Available for identie cation are the excitation signals and the resultant responses, which may be corrupted by unknown and possibly dominating disturbances. We assume that no actual measurement of the disturbances themselves is available for identie cation. The case where the disturbance proe le is unknown and may be quite complicated, but its period is known, is considered e rst. Next, we consider the case where the disturbance frequencies are known only approximately, and a procedure is applied to iteratively ree ne the estimation of the disturbance frequencies for successful system identie cation. Finally, we examine the situationwheretheunknowndisturbanceisnonperiodic.Theidentie cationproducesresultsthatcanbeusedforthe synthesis of feedforward and feedback control to cancel the disturbance effect. Both simulation and experimental results will be used for illustration. The simulation is carried out with a model of a communications satellite, and the experimental results are obtained from a e exible truss structure. A companion paper addresses the system identie cation problem where the disturbance effect is modeled implicitly. HE control problem of rejecting unwanted disturbance has many applications in electrical, mechanical, aerospace, and acoustic systems. Many methods have been developed to treat this problem. If both the plant and disturbance frequencies are known, classical feedback control approaches call for the design of e lters with high gain at the disturbance frequencies to produce corre- sponding zeros in the closed-loop transfer function relating the dis- turbances to the system response. The closed-loop system is then capable of rejecting the unwanted disturbances at the designed fre- quencies. One such design is included in the attitude control sys- tems of the INMARSAT III and INTELSAT VIII spacecraft. 1 Re- jection of sinusoidal disturbances can also be achieved using the frequency-shaped cost functional method. 2 This is an adaptation of linear quadratic Gaussian (LQG) design methods that determines the control necessary to minimize a cost function expressed in the frequency domain. The cost function penalizes the response at the known disturbance frequencies, resulting in high gain feedback at these frequencies. Another method for disturbance rejection is known as disturbance accommodation control or, as it is sometimes called, disturbance modeling or disturbance estimation. 3 Itassumes an effective disturbance input at the control input location that has the same effect at the output as the actual disturbance. The same harmonic disturbance can be modeled as marginally stable modes and appended to the state-space model of the plant. An observer is designed to estimate the plant states together with the disturbance from which the control signal is computed. The disturbance ob- server approach also assumes an equivalent effective disturbance at the control input, but it uses an inverse plant model to estimate this disturbance for control. 4 Repetitive control is another approach to reject periodic disturbance where the tracking error observed in the previous periods is used to correct the control signal for the cur- rent period. 5,6 Adaptive control can be used to cancel the effects of unknown sinusoidal or periodic disturbance through the use of a sufe ciently overparameterized model so that the disturbance dy- namics can be entirely absorbed in the identie ed model. 7 With the disturbance embedded in the model, based on the internal model principle the resulting feedback control has ine nite open-loop gain at the disturbance frequencies and achieves complete cancellation. The e ltered-X least mean squares is another well-known distur- bance rejection method. 8 This method requires measurement of a disturbance-correlated signal, and together with an assumed model of the system it adaptively tunes the coefe cients of a e nite impulse response e lter driven by a disturbance-correlated reference signal to achieve disturbance cancellation. Adaptive inverse control com- bines adaptive feedforward techniques for command or model fol- lowing and adaptivefeedback techniques for disturbance rejection. 9 Artie cial neural networks can also be used to provide disturbance rejection control. 10 Neural networks are used to identify the dynam- icsfromthedisturbancesources,andthecontrolinputstothesystem outputs and adapts the control signals for disturbance cancellation. Instead of treating the disturbance-rejection problem from a con- trol system synthesis standpoint, we address the problem from a system identie cation perspective. The research e rst focuses on the system identie cation problem in the presence of unknown distur- bance inputs, and then uses the identie cation results to solve the re- lated disturbance-rejection control problem. We assume no a priori knowledge of the system and no actual measurement of the dis-

A Switching State Feedback Control of Constrained Systems with Exogenous Inputs

This paper considers switching state feedback control of systems with unknown but bounded disturbance inputs subject to state and control constraints, and resolves convergence problems which arise with the effects of unknown disturbances. This difficulties have not been handled by the switching control laws which were proposed in the literatures. The proposed switching control law assures convergence of the state and successive controller switching even in the presence of unknown disturbances. The main idea is to construct an upper bound of the state reachable set ' such that convergence of the state is assured by restricting this set to some subset of the state space. The resulting optimization problems which provide a sequence of feedback gains are more difficult to solve than those with no disturbances. However, this optimization problems may be solved with some heuristic manners. In addition, required on-line computational burden is the same as that in case of no disturbance inputs.