This study proposes a data-based model-free approach to reference trajectory tracking in two-degree-of-freedom non-linear control system (CS) structures. This model-free control approach tunes both the feedback controller parameters and the reference input sequence accounting for control saturation and control rate constraints. The controller is iteratively tuned in a non-linear framework that employs a gradient descent search approach. The model-free gradient estimates are obtained by a perturbation-based approach. The reference input tuning is carried out in a linear framework using an iterative learning control-based approach, and it also includes a model-free gradient search algorithm where the gradient estimates are obtained by a similar perturbation-based approach. The number of real-world experiments is significantly reduced by the use of simulated models identified as neural networks. A digitally simulated case study concerning the angular position control of a non-linear aerodynamic twin-rotor system shows that the author's approach can effectively improve the CS performance.

Motivated by the successful application for feedback control, this study extends the study of reinforcement learning techniques to the design of two-degree-of-freedom controllers in the data-driven environment. Based on the residual generator based form of Youla parameterisation, all stabilising controllers are first interpreted in the feedback–feedforward situation with a Kalman filter-based residual generator acting as the core part. For the reference tracking problem, further discussions are conducted from the regulatory perspective and using the Q learning, recursive least squares methods and the policy iteration algorithm. The entire design is carried out as a two-stage process that separately achieves the optimal feedback and feedforward controllers. Finally, the effectiveness of the proposed approach is demonstrated with its application in the laboratory continuous stirred tank heater process.

A predictive control using subspace identification is applied for the smart grid integration of refrigeration systems under a direct load control scheme. A realistic demand response scenario based on regulation of the electrical power consumption is considered. A receding horizon optimal control is proposed to fulfil two important objectives: to secure high coefficient of performance and to participate in power consumption management. Moreover, a new method for design of input signals for system identification is put forward. The control method is fully data driven without an explicit use of model in the control implementation. As an important practical consideration, the control design relies on a cheap solution with available measurements than using the expensive mass flow meters. The results show successful implementation of the method on a large-scale non-linear simulation tool which is validated against real data. The performance improvement results in a 22% reduction in the energy consumption. A comparative simulation is accomplished showing the superiority of the method over the existing approaches in terms of the load following performance.

In this study, the authors develop a data and event-driven control scheme that combines data and event-driven control methods, for a class of networked non-linear control systems, which consist of a remote controller and a local non-linear control system. The plant in the local non-linear control system has an unknown dynamic model. They apply a data-based method to design and adjust the feedback gain matrix for the plant, whose desired outputs are given by the remote controller. This method has advantages of both data and event-driven control methods. Moreover, the proposed scheme provides an alternative option to realise remote control, when the dynamic model of the plant is unknown. They then present theoretical analysis on the convergence condition and a computer simulation example to demonstrate the feasibility of the developed control scheme.

The focus of this work is on fatigue estimation and data-based controller design for wind turbines. The main purpose is to include a model of the fatigue damage of the wind turbine components in the controller design and synthesis process. This study addresses an online fatigue estimation method based on hysteresis operators, which can be used in control loops. The authors propose a data-based model predictive control (MPC) strategy that incorporates an online fatigue estimation method through the objective function, where the ultimate goal in mind is to reduce the fatigue damage of the wind turbine components. The outcome is an adaptive or self-tuning MPC strategy for wind turbine fatigue damage reduction, which relies on parameter identification on previous measurement data. The results of the proposed strategy are compared with a baseline model predictive controller.

A new co-state method with a finite number of states was recently expanded to be able to calculate the inflow below the rotor disk. The expanded model is derived in a mathematically rigorous way from the potential flow equations. It offers an efficient mean of doing real-time simulation of rotary wing systems and generating the data for design and control purpose. In this study, as an important application of the generalised dynamic wake theory, the new method is used to compute inflow in the wake of a rotor with a finite number of blades. The three velocity components are calculated in the time domain for axial flow. The velocities are plotted at different distances from the rotor disk and compared with the Glauert prediction for axial flow and wake swirl. In the finite-state model, the angular momentum does not jump instantaneously across the disk (as predicted by Glauert), but it does transit rapidly across the disk to correct Glauert value.

In this study, an adaptive neural backstepping control scheme is proposed for a class of strict-feedback non-linear systems with unmodelled dynamics, dynamic disturbances and input saturation. To solve the difficulties from the unmodelled dynamics and input saturation, a dynamic signal and smooth function in non-affine structure subject to the control input signal are introduced, respectively. Radial basis function (RBF) neural networks are used to approximate the packaged unknown non-linearities, and an adaptive neural control approach is developed via backstepping, which guarantees that all the signals in the closed-loop system are semi-globally uniformly ultimately bounded in mean square. The main contributions of this note lie in that a control strategy is provided for a class of strict-feedback non-linear systems with unmodelled dynamics uncertainties and input saturation, and the proposed control scheme does not require any information of the bound of input saturation non-linearity. Simulation results are used to show the effectiveness of the proposed control scheme.

A medium-frequency disturbance attenuation strategy based on the virtual-gimbal effect of the active magnetically suspended reaction wheels (MSRWs) is proposed to improve the pointing stability for the high-resolution observation of the spacecraft. The disturbances decreasing the pointing stability include not only the space environment disturbances with low frequency but also the flexible modes and vibrations with medium frequency which are excited by the flexible appendix and the rotating components. The model of the MSRW is constructed firstly and the model of the spacecraft with one MSRW is then developed. Furthermore, a hybrid three-loop control strategy is designed to stabilise the pointing of the spacecraft and the MSRW, where the compensation loop is used to attenuate the medium-frequency disturbances beyond the bandwidth of the attitude control loop. The bandwidth of the MSRW is expanded by tilting the rotor shaft, and the attitude angular velocity is directly feedback to the compensation loop. In addition, the cross feedback control is introduced to attenuate the medium-frequency disturbances. Finally, simulations are constructed and the results indicate the medium-frequency disturbances are effectively attenuated.

In this study, a novel data-driven terminal iterative learning control with high-order learning law is proposed for a class of non-linear non-affine discrete-time multiple-input–multiple output systems, where only the system state or output at the endpoint is measurable and the control input is time-varying. A new data-driven dynamical linearisation is proposed in the iteration domain and the linearisation data model can be updated by a designed parameter updating law iteratively. The high-order learning control law makes it possible to utilise more control knowledge of previous runs to improve control performance. The design and analysis of the proposed approach only depends on the I/O data of the control plant without requiring any explicit model information. Both theoretical analysis and extensive simulations are provided to confirm the effectiveness and applicability of this novel approach.

The conventional approaches for modelling of charge cycle in combustion engines are based on first principles. In this approaches, it is necessary to estimate different parameters in engine, which is very difficult to implement in real-time. In addition, it requires a high parameterisation effort because of a variety of engine characteristics. The model is further used in engine control unit to improve the efficiency of the combustion. Moreover, it can be used for estimation of some physical variables in engine to increase the performance of the controller. It also avoids the installation of extra sensors and reduces the costs of production. In this study, two different data-driven modelling methods for estimation of air mass are investigated. They also offer a flexible modelling approach considering various options for equipment as well as sensors and actuators in the engines. Both data-driven methods, the Gaussian mixture regression (GMR) and the local linear model tree (LOLIMOT) algorithm, allow a flexible modelling with a high input space dimensionality, with sparsity of data, a good interpretability and also offer possibilities of adaptation and local optimisation. While the GMR uses a statistical approaches to train the model, in the LOLIMOT an incremental, heuristic approach is used. Both methods are applied to the data obtained from a spark-ignition engine and the results are compared and discussed.

This study proposes a scheme for fault detection and identification (FDI) of a class of non-linear hybrid systems. For fault detection, the proposed method uses a self-switched unscented Kalman filter (UKF) where a component filter with the appropriate dynamic model is chosen to suit the current mode of the hybrid plant. The mode of the hybrid plant, usually defined by system states, is deduced based on the state estimates. The statistical tests of measured datasets are used to detect and identify the fault parameters. For fault identification, a bank of such switched UKF filters is used. A three-tank system has been used to demonstrate the effectiveness of the scheme. Different types of faults, that is, leakage, clogging and abrupt changes in inflow (actuator fault) have been considered. t-Statistical test has been performed on the residuals for FDI and threshold calculation purposes. It is shown that all the types of faults can be successfully detected and identified by the proposed method. The performance of the proposed system is compared with the same for an extended Kalman filter-based FDI system. It has been shown that the UKF-based scheme has lower latency and higher range coverage.

In this study, a robust approach to global identification of linear parameter varying (LPV) systems in an input–output setting is proposed. In practice, the industrial process data are often contaminated with outliers. In order to handle outliers in process modelling, the robust LPV modelling problem is formulated and solved in the scheme of generalised expectation–maximisation (GEM) algorithm. The measurement noise is taken to follow the Student's t-distribution instead of using the conventional Gaussian distribution, in this algorithm. The extent of robustness of the proposed approach is adaptively adjusted by optimising the degrees of freedom parameter of the Student's t-distribution iteratively through the maximisation step of the GEM algorithm. The numerical example is provided to demonstrate the effectiveness of the proposed approach.

The problem of thruster misalignment angles identification on-line and in real-time is addressed in this study. A data-driven approach is proposed to identify the parameter of satellite's thruster misalignment angles. The satellite considered is represented by Euler angles. As the finite available data is met for on-orbital satellite, the proposed approach is designed by using the techniques of nonlinear observer and data-driven solution simultaneously. As a stepping stone, a terminal sliding-mode observer is developed to estimate the deviation torque introduced by thruster misalignment. It is shown by Lyapunov stability analysis that, precise estimation with zero error is guaranteed with finite-time convergence. Based on such estimated value, a data-driven solution is then synthesised to identify the misalignment angles of the thruster. The performance using the proposed identification methodology is evaluated through a numerical example.

This study presents data-driven super-twisting-like smooth observer design for systems of dimension more than one. The major advantage of the proposed observers is finite-time convergence of the estimation error to zero and smoothness of the observer trajectory. This study concludes with numerical simulations illustrating performance of the designed algorithms.

Over the last decade, different micro-tool navigation and localisation algorithms have been developed. They can be divided into three groups: the eigen-decomposition methods such as principal component analysis (PCA), the transform methods such as Hough transform (HT) and the space random iteration methods such as the random sample consensus (RANSAC) algorithm. To suppress the speckle noise of the ultrasound image, different noise suppression methods are also proposed. In this article, different combinations of preprocessing methods and localisation methods are compared. A tracking system is also developed to adapt these localisation methods to a dynamic situation. So far, the line-filter method has achieved the best contrast ratio, and the threshold method requires the shortest time. These two preprocessing methods are proposed in the localisation algorithm and the tracking system. Simulations and experiments were conducted to verify the combinations of the localisation and tracking results. In static localisation, the line-filter+RANSAC method achieves the highest localisation accuracy. In the dynamic situation, the PCA method achieves the highest tracking accuracy. In the real-time evaluation, the calculation time of the RANSAC algorithm is the shortest. To satisfy the demand for localisation accuracy and calculation time, the line-filter+RANSAC algorithm is the best choice.

In this study, a new statistical monitoring technique based on efficient projection to latent structures (EPLS) is proposed for quality-relevant fault detection in multivariate processes that exhibit collinear measurements. The algorithm decomposes process variables into three subspaces according to singular value decomposition (SVD) and principal component analysis. This method effectively solves the problems of real-time and effectiveness in the quality-relevant fault diagnosis of industrial processes. EPLS takes full advantage of good nature of SVD. As a result, the calculation of the projection process is greatly reduced. It is reasonable that the process data space was projected to three subspaces, a quality-relevant subspace, a quality-irrelevant subspace and a residual subspace. EPLS algorithm not only effectively avoids the tedious iterative calculation, but also makes more explicit spatial resolution. The framework of quality-relevant fault monitoring based on EPLS algorithm was proposed as well in this study. A challenging problem, real industrial hot strip mill process, is used to illustrate the effectiveness of the proposed method.

In this study, parameter faults in a class of non-identifiable closed-loop multiple-input multiple-output systems are considered. A new parameter estimation-based fault diagnosis method is proposed. It is known that in open-loop systems, the system parameters can be identified directly and the on-line identification results can be used for fault detection and isolation. However, as the closed-loop system is non-identifiable because of the correlation introduced by the controller, unique optimal parameter estimation solution cannot be obtained. To address such an issue, a new method to detect and isolate parameter faults of closed-loop systems without persistent excitation condition is proposed. A reduced-order model is firstly constructed, which is the projection of the original model onto the orthogonal direction of the controller. By doing this, the aforementioned correlation can be successfully removed. The parameters of the newly constructed model, called as feature parameters, are then identified. The physical faults are finally detected and isolated based on the on-line identification results of the feature parameters, the projection direction and the known influence matrix. Simulation results are given to show the effectiveness of the proposed method.

This study is concerned with the data-based control of networked non-linear control systems with random packet dropouts in both the sensor-to-controller and controller-to-actuator channels. By taking advantage of the characteristics of networked control systems such as the packet-based transmission, timestamp technique, as well as smart sensor and actuator, a data-based networked predictive control (DBNPC) method is proposed to actively compensate for the two-channel packet dropouts, where only the input and output data of the non-linear plant are required. A sufficient condition for the stability of the closed-loop system is developed. Furthermore, the resulting DBNPC system can achieve a zero steady-state output tracking error for step commands. Finally, extensive simulation results on a networked non-linear system demonstrate the effectiveness of the proposed method.

Without the explicit process identification, the authors propose a model-free adaptive control framework for unknown plant by using the concept of equivalent dynamic linearisation controller. The controller has linear incremental structure and its local dynamics is equivalent to the ideal controller in theory. Hence, the problem of determining the structure of candidate controller is transformed to the problem of finding a sequence of local dynamic controllers to approximate the ideal controller. With the help of gradient information extracted from input and output (I/O) data of the plant, the optimal controller parameter sequence is generated by minimising a user-defined control criterion. This method gives a solution on how to determine the candidate controller structure. The controller design, parameter tuning and controller validation are based on I/O data of the plant. Hence, it could reduce the influence of internal disturbance or unmodelled dynamics. The effectiveness of the proposed method is illustrated by the simulation of a continuous polymerisation reaction process in a jacketed continuous stirred tank reactor system. Meanwhile, a simulation comparison is carried out to show the superiority of neural network data model in model-free adaptive control framework.

In this study, a fault-tolerant control (FTC) system is proposed and applied on a wind turbine benchmark. The significant characteristic of the FTC system is the data-driven fault-tolerant architecture whose core is embedding residual generators into control loop. To begin with, the authors introduce a method for constructing residual generators with parity vectors, which are directly identified from measurements. Then an optimal scheme is proposed for selecting parity vectors that are sensitive to specific variables. Besides, several residual generators are applied in the FTC system for multiple failures, and a switching mechanism is elaborately designed. At last, the effectiveness of the proposed FTC scheme is demonstrated by the simulation results of the wind turbine model.