This work aims at developing a novel direct design and analysis method of learning control protocol toward consensus performance of multiagent systems (MASs) without using any model. A nonlinear autoregressive moving average (NARMA) function is designed at first to formulate the inherent consensus dynamics with respect to the consensus error and the control protocols. Then, a consensus performance-related iterative linear data model (CPiLDM) is constructed for equivalently reformulating the NARMA consensus system’s iterative dynamics in a data-driven framework. The CPiLDM does not rely on a model no matter through first-principle modeling or system identification methods. Next, a direct distributed iterative learning control (DirDILC) method is developed through an optimization technique subject to the CPiLDM. The convergence is proved directly for the virtual NARMA consensus system, without relying on the dynamics of the agent itself, and thus simplifies the analysis consequently. Since the presented DirDILC is purely data-driven without relying on an explicit model, it constitutes a significant step forward from the existing consensus control theory.

Aiming at the difficulties in establishing mechanism models, significant influence of load disturbances on control accuracy, and high requirements for operational smoothness of propulsion systems for chemical tankers under complex sea conditions and liquid cargo sloshing conditions, a Model-Free Adaptive Predictive Control (MFAPC) method incorporating variable-gain error feedback correction is proposed. First, the mathematical model of the chemical tanker’s propulsion motor and propeller load is established. Then, based on the compact-form dynamic linearization technique, the pseudo partial derivative is identified online using system I/O data streams, reconstructing the nonlinear data-driven model and multi-step ahead prediction equations of the propulsion system. A performance index function incorporating rotational speed tracking error and control increment constraints is constructed, and the model-free adaptive predictive control law is designed through rolling optimization. Finally, to address potential control deviations caused by persistent load disturbances, a variable-gain error feedback correction term is introduced into the controller, which adjusts the feedback intensity in real time according to the rotational speed error. Simulation results demonstrate that the proposed control method effectively improves the dynamic response speed and steady-state accuracy of the chemical tanker’s propulsion system, achieves smooth rotational speed regulation, and exhibits excellent control performance.

This article studies an iterative learning-based adaptive containment control for a kind of repetitive multiagent systems (MASs) with a nonstrict-feedback structure. The hyperbolic tangent function and the mean value theory are combined to handle the input saturation problem. Then, we establish a relationship between the original system state and compromised system state to resolve the impact of the false data injection (FDI) attack on the system performance. Under the framework of backstepping, the nonstrict-feedback issue is addressed by using the property of the fuzzy logic system (FLS). However, it is inevitable to generate two unknown control gains during the process of addressing the input saturation and FDI attack. As a result, an estimated scheme is proposed to approximate their bound. Moreover, two auxiliary functions are introduced to the virtual controller to reduce the influence of the sign function and achieve the asymptotic convergence. The proposed scheme guarantees that all the followers converge to a convex hull spanned by the leaders as the iteration approaches infinity. Finally, two simulation examples are considered to verify the effectiveness of the designed method.

Iterative learning control (ILC) is an effective tool for repetitive systems to achieve perfect tracking tasks on a finite time interval. However, most of ILC results focus on iterative asymptotic convergence, that is the tracking performance can only be achieved when the iterative operation tends to the infinity, which is undesired in practice. Therefore, a backstepping-based finite-iteration tracking control (FITC) method is designed to ensure the finite-iteration convergence (FIC) of nonlinear discrete-time systems (NDTSs). In this work, the definition of FIC of NDTSs is given for the first time. On this basis, a backstepping-based FITC approach for repetitive NDTSs is proposed by employing variable replacement, and the finite-iteration number is derived via difference inequalities. Moreover, the proposed control strategy avoids the causality contradiction from the traditional backstepping technique of NDTSs. Simulation results are used to illustrate the effectiveness of the presented scheme.

Unmeasurable states, random initial conditions, and high computation complexity are three main challenging problems in the adaptive iterative learning control (AILC) of nonlinear repetitive systems with the non-strict-feedback form. This paper develops an event-based AILC tracking control method using the backstepping technique to address the three problems simultaneously. A neural network (NN)-based learning observer is established to solve unmeasurable states. Furthermore, only one NN is required by transforming all the system nonlinearities to the last state, which further decreases the calculation burden of the observer. As for the random initial conditions, a state tracking approach is developed by designing an attenuated trajectory to remove the limitation of identical initial conditions. To reduce the computation complexity, a tracking differentiator is considered to handle the complexity explosion problem, and the corresponding compensation approach is given to reduce the error influence caused by the tracking differentiator. In addition, a saturated-threshold event-driven mechanism is designed to enhance the actuator service time and to save storage space. The convergence analysis is executed, and the effectiveness of the presented approach is demonstrated through an illustrative simulation.

This work studies the consensus learning control for a nonlinear nonaffine MAS with limited digital channel capacity under quantization parameters mismatch. A finite-level uniform quantizer based on encoding-decoding mechanism is employed to alleviate the communication burden. Zoom-out and zoomin strategies are utilized to achieve adaptive adjustment of the uniform quantizer, thereby ensuring effective quantization. The nonlinear nonaffine dynamics of the MAS are compressed into an unknown parameter of an equivalent linearized data model through iterative dynamic linearization method. A parameter estimation algorithm is developed to estimate the unknown parameter in real-time. Then, a model-free adaptive quantized iterative learning control (MFAQILC) method is proposed by using a mismatch compensation mechanism, where the compensation information is utilized into the learning control updating law. Simulation results demonstrate that the proposed method is effective in dealing with quantization parameters mismatch in encoding-decoding mechanism.

This work proposes three minimum operator based data-driven finite-iteration learning control (MODFILC) methods under a universal lifted design and analysis framework. They deal with different tasks with abilities of tracking a holonomic trajectory, multi-intermediate passing points, and a single endpoint, respectively. A lifted iterative recursive linear data model (LDM) with the extensions of point-to-point LDM and terminal LDM is constructed. They formulate the iterative dynamics of the outputs at the designated time instants with respect to the control inputs. Then, three MODFILC methods are proposed by applying the optimization-based approach to the LDMs where the minimum operators are incorporated into the performance functions. Not only do they accelerate the convergence rate but also avoid large input changes. Three iterative adaptation laws are designed to estimate the non-zero elements of the lifted Jacobian matrices of LDMs not only to enhance the robust performance but also to decrease the computation complexity. The finite-iteration convergence is proved under a universal framework where the upper bound of the settling iteration number can be obtained explicitly and also tuned artificially. The three proposed methods are data-driven without use of a model. Simulation experiments verify the results.

In this paper, a direct data-driven cooperative (DirDDC) control approach is investigated for intelligent connected vehicles (ICVs) platoons against denial of service (DoS) attacks. First, a speed compensation mechanism is designed to counteract the impact of DoS attacks. Based on the inherent relationship between speed and displacement, and combining the communication topology of vehicles, the compensated coordinated output of the ICVs platoon is established. Essentially, there should be a certain dynamic relationship between the compensated coordinated output and the control input among vehicles, which is further formulated as an unknown nonlinear function. On this basis, such relationship is equivalently transformed into a dynamic linearization data model containing an unknown vector parameter which is estimated through the optimization of a designed performance index. Finally, a novel DirDDC method is proposed, which includes a parameter estimation algorithm along with its reset mechanism and control update law that utilizes more historical input information to enhance the control performance. The proposed DirDDC control does not rely on any model information except for the communication topology information and input-output data among the ICVs. Simulation results validate the effectiveness of the proposed DirDDC control.

An uncertainty predictive observer-based model-free adaptive disturbance rejection control (UPO-MFADRC) is developed for nonaffine nonlinear systems with uncertain nonlinearities and exogenous disturbances. A linear predictive data model (LPDM) consisting of a linear parametric part and a total residual uncertainty is derived from the original nonaffine nonlinear system. Then, an adaptive law with a prediction algorithm is proposed to address the unknown parameter sequence of the LPDM. An uncertainty predictive observer (UPO) is developed to predict the future behavior of the total residual uncertainty sequence of the LPDM including the unmodeled dynamics and exogenous disturbances. The UPO involves an extended state observer for the uncertainty estimation at the current time instant and a hierarchical prediction algorithm to predict the uncertainty at more than one future time instant. Finally, an entire UPO-MFADRC is constructed by incorporating the adaptive law, UPO, and a control law generated from the moving-window optimization of a designed performance index function. The proposed UPO-MFADRC is almost model-independent except that the control direction needs to be known. The convergence is shown mathematically with the use of contraction mapping-based method. Simulation verifies the ability of the proposed UPO-MFADRC in tolerating the uncertainty.

This paper proposes a model-free finite-time control (MFFTC) scheme for consensus tracking in nonlinear multi-agent systems with unknown dynamics. A linear data model (LDM) is constructed by using only local input-output measurements. Based on this LDM, a fractional-order control law is designed with an adaptive estimation mechanism for pseudo partial derivatives. The proposed MFFTC guarantees the finite-time convergence of tracking errors to a prespecified bound without relying on any prior model knowledge. Simulation studies validate the effectiveness of the MFFTC method in achieving bounded consensus tracking within finite time instants. Simulation illustrates the theoretical results.

This paper proposes a data-driven iterative learning fault tolerant control algorithm for subway train speed tracking with time-iteration-varying actuator and sensor faults and over-speed constraints. First, a novel fault-embodied linear data model is proposed to capture non-repetitive uncertainties caused by actuator and sensor faults alongside external disturbances. Based on this model, a fault detection estimator with an iteration-varying residual threshold is developed to improve sensitivity to fault onset. Upon fault occurrence, radial basis function neural networks, trained under fault-free conditions, are employed to estimate the fault effects using the faulty speed signal. These estimations are then integrated into the proposed framework to achieve timely fault compensation. Subsequently, an over-speed constraint mechanism is designed to ensure safe train operation, and constraints on traction/braking forces are incorporated. Rigorous mathematical analysis proves that the proposed constrained proposed method ensures the convergence of speed tracking errors across iterations. Compared with other controllers, numerical tests are provided to verify the effectiveness and stability of the proposed controller. Tests on the Beijing Yizhuang line further demonstrate that it outperforms the iterative learning fault tolerant control method.

This paper addresses the finite-iteration consensus tracking for an unknown nonlinear multi-agent system under a fixed topology. A data-driven finite-iteration learning control (DDFILC) is proposed on the basis of an iterative linearized data model describing the system dynamics along the iteration axis. The proposed DDFILC consists of an iterative learning control law involving a fractional factor and the error bound to guaranty that the multi-agent system tracks the desired trajectory within finite iterations. The simulations validate the effectiveness of the DDFILC.

This paper proposes an observer-based model-free iterative learning fault tolerant control (ObMFilFTC) algorithm for the nonlinear system with disturbances and non-repetitive time-varying actuator faults. First, an original linearization data model (LDM) considering non-repetitive uncertainties is established. Since it contains fault information, this allows the fault information to be estimated using the parameter estimation law. The external disturbances and the non-repetitive time-varying actuator faults constitute the total non-repetitive uncertainties. Next, to deal with non-repetitive uncertainties, we present a novel iterative output observer (ILO) that considers all historical iteration observation errors to estimate inaccurate outputs ruined by non-repetitive uncertainties. With the introduction of ILO, the tracking accuracy and the ability to suppress non-repetitive uncertainties are improved. Additionally, the inclusion of the tracking error integral term in the ILO enhances the convergence speed. Meanwhile, by utilizing the estimated outputs, an observer-based parameter updating law is proposed. Furthermore, we propose an optimal iterative learning control (ILC) algorithm to ensure precise tracking of the desired trajectory. The convergence of the proposed ObMFilFTC method is proofed strictly. The proposed ObMFilFTC method guarantees that the system can follow the desired trajectory despite non-repetitive actuator faults and disturbances in nonlinear systems, relying solely on input/output(I/O) data. Finally, the simulation results further demonstrate the effectiveness of the proposed algorithm.

This work studies Denial-of-Service (DoS) attacks for strongly connected nonaffine nonlinear multi-agent systems (MASs). At first, we develop a network topology-based linear data model (NT-LDM) to reformulate the input/output behavior among the agents. A dynamic consensus protocol is derived using the NT-LDM to enhance the consensus performance of the network topology. Meanwhile, a combined attack compensation mechanism, incorporating both input and output, is introduced for reducing the poor influence of DoS attacks on network communication. Then, an input-behavior-learning-based data-driven control (IBL-DDC) is developed to improve the consensus protocol by designing an attack compensation scheme to the unavailable data due to DoS attacks. The proposed IBL-DDC not only is independent on the system model, but also can reject DoS attacks by learning from the input action of other agents through the proposed compensation scheme. The results are confirmed by the simulation study.

The security tracking control problem is addressed for nonlinear discrete-time networked control systems (NCSs) subject to energy-constrained denial-of-service (EC-DoS) attacks. A probabilistic quantization-based coding-decoding protocol (CDP) is designed to mitigate bandwidth limitations and enhance transmission security. Then, a CDP-based data-driven adaptive sliding mode control (DDASMC) method is presented on the basis of a linear data model (LDM) of the nonlinear NCSs. A detection algorithm is designed to detect EC-DoS attacks so that the proposed CDP-based DDASMC can automatically switch between two cases with and without EC-DoS attacks, respectively. In the duration that the attack does not occur, an integral sliding function and a reaching law are designed to achieve fast convergence while ensuring a good control performance. In the duration that the attack occurs, a prediction mechanism is introduced to construct a predictive LDM for estimating future control information. Further, a sliding mode predictive control is developed to address the EC-DoS attacks and enhance the control performance. All the algorithms of the proposed DDASMC are computed only using the input-output (I/O) data instead of any other information of the physical models. Simulation study validate the results. Note to Practitioners-NCSs have been extensively applied in numerous fields, such as smart grid and aerospace. However, the open network environment provides attackers with many opportunities for cyber-attacks, resulting in the deterioration of control performance. Accordingly, it is urgent to analyze the security problems of NCSs that are vulnerable to cyber-attacks. This study considers the security tracking control problem of NCSs under EC-DoS attacks. A probabilistic quantization-based CDP is introduced for the feedback channel to reduce the bandwidth of communication networks and improve the security of data transmission by compressing the data before transmission. Then, a detection mechanism is proposed to determine whether NCSs are attacked, and a DDASMC with a switching mechanism is developed to address both cases with and without EC-DoS attacks. The presented control method is data-driven and includes an adaptive switching mechanism that not only saves resources but is also more suitable for practical applications.

Considering the four challenges of non-identical initial states, non-repetitive uncertainties, different batch lengths, and unavailable mathematical model of a rubber mixing process (RMP), this article proposes a data-driven iterative learning temperature control (DDILTC) for the RMP. Specifically, an iterative linear data model (iLDM) is developed to formulate the iterative dynamics of RMP and is further used as a one-step iterative linear predictive model to estimate the RMP’s temperature that is unavailable when the current batch length is shorter than the desired one. The unknown parameters of the iLDM are estimated iteratively by designing an iterative adaption law. Further, an iterative learning based observer is designed to estimate the non-repetitive uncertainties and non-identical initial states as an extended state. The proposed DDILTC is a data-driven method and the iLDM is only used to formulate the iterative relationship of the input-output between two batches instead of a mathematical model of the RMP with physical meanings. Simulation study verifies the results. Note to Practitioners —The mixing temperature of a rubber mixing process (RMP) is a critical variable, ensuring the desired plasticity and viscosity of the rubber compounds. Indeed, RMP is a typical batch process performing repetitively over the finite time interval. However, no ILC results about the RMP temperature control have been reported even though ILC can learn the control experience from the past batches to improve control performance. The main reason lies in that the practical environments of RMP make it impossible to satisfy the strictly repetitive conditions, i.e., the initial states, disturbances, and batch lengths are all iteration-varying. Furthermore, it is difficult to establish a mathematical model of the RMP due to its large production scale and complex dynamics along both time and iteration directions. Therefore, the main motivation of this paper is to study the iterative learning temperature control problem of RMP by considering the nonrepetitive uncertainties of initial states, disturbances, and batch lengths, bypassing the use of any model information. An iterative linear data model (iLDM) is established to equivalently reformulate the unavailable two-dimensional dynamic behavior of RMP and to facilitate the controller design and analysis. The gradient uncertainty of RMP is reformulated as the unknown parameters in the iLDM and can be iteratively estimated by designing an iterative adaptation algorithm. The non-repetitive initial states and disturbances can be estimated by designing an iterative observer. Moreover, the unavailable mixing temperatures at the unreachable operation points are estimated by using the iLDM as the iterative predictive model. To summarize, the proposed method is simple in computation and easy in implementation since only the I/O data is used, and thus it is of great practical significance.

To address the challenges of undetected defective potatoes and real-time processing requirements for surface defect detection of peeled potatoes in intelligent pre-cooked food production lines, this study proposes an enhanced YOLOv5 framework incorporating Adaptive Spatial Feature Fusion (ASFF) and Slimneck lightweight module. The ASFF module facilitates adaptive weighted fusion of multi-scale defect features, significantly improving model robustness against small-scale defects including blackening, sprouting, and mechanical damage. Concurrently, the Slimneck module enhances computational efficiency through parameter redundancy reduction while maintaining detection speed. Experimental results demonstrate that the optimized model achieves 87.1% mAP50 with 12.54 M parameters and 22.9G floating-point operations (FLOPs). Specifically, the model achieves 89.3% precision on the critical bad category, surpassing the baseline YOLOv5s by 2.3 percentage points. In practical deployment at Shandong Yinying's facility, the system achieves over 98% overall accuracy in the binary classification setting. This work provides a reliable and efficient visual inspection solution for agricultural product sorting equipment, demonstrating significant industrial application potential.

This article investigates the control problem for a sort of repetitive discrete-time nonlinear systems subject to random packet dropouts and limited communication bandwidth. In order to compensate the impacts from the constraints on bandwidth, this work designs a communication protocol by designing a two-description coding scheme in combination with the scalar uniform quantization technique. The proposed protocol makes use of two independent channels to transmit data separately, thereby improving the channel utilization efficiency and reducing the probability of packet dropout. Then, with the proposed protocol and the iterative dynamic linearization approach, an adaptive iterative learning controller associated with a parameter estimation strategy is provided for the nonlinear system under investigation. The control law is data-driven, which therefore does not require knowledge of the model. Subsequently, the sufficient condition is derived under which the tracking error is forced to convergent. Finally, with the purpose to show the correctness of our theoretical results, we carry out two numerical simulations to test the effectiveness of the proposed control strategy.

This paper aims to investigate iterative learning control based on the high-order fully actuated (HOFA) system method. Firstly, the discrete-time nonaffine nonlinear HOFA system is transformed into an equivalent representation along the iterative direction, which includes an unknown parameter compressed into all unknown nonlinearities. Then, the parameter is estimated iteratively by a projection algorithm. Based on this, a data-driven HOFA iterative learning control (DD-HOFAILC) method is proposed under the HOFA system framework. Different from the traditional HOFA system methods, the proposed DD-HOFAILC is a data-driven control approach, where the model information of the system is not used. In addition, this work combines the HOFA system method with data-driven ILC, which further enriches the theory of HOFA systems. Finally, rigorous theoretical analysis is provided using mathematical tools such as matrix theory and mathematical induction, and simulation results verify the effectiveness of the proposed DD-HOFAILC method.

A novel data-driven finite-time control (DDFTC) scheme for discrete-time nonlinear non-affine systems is proposed in this paper. Firstly, a linear data model (LDM) is presented for describing the system dynamics satisfying the Lipschitz continuity condition. Then, a fractional order controller involving parameters relating to the error bound is adopted. The tracking error converges to a designated bound within finite time, and the upper bound of the finite-time instant is obtained by using proper Lyapunov function. An example is provided illustrating the result.