To address the increasing demand for efficient and sustainable port operations amid the complex dynamics of global shipping, this study investigates the coordinated scheduling problem of berth allocation and shipping speed within port groups under the transshipment rationalization strategies (BSCS-TRS). A multi-objective mixed-integer linear programming (MILP) model is proposed for the BSCS-TRS. A multi-objective discrete combinatorial optimization algorithm based on decomposition and adaptive large neighborhood search (MODA/D-ALNS) is proposed to solve the problem, where a shortest path-based ship speeds optimization method (SPSOM) is developed to optimize the shipping speed. The performance of the proposed model and MODA/D-ALNS is systematically through comparative numerical instances and case studies. The results indicate that the proposed model can optimally solve small-scale instances of BSCS-TRS using Gurobi, while MODA/D-ALNS exhibits good performance in benchmark knapsack problems and BSCS-TRS. Sensitivity analysis results indicate that the proposed method can reduce ship waiting time by 79.06% and fuel consumption by 2.25% within port groups. These findings provide a scalable decision-making tool for port operators to balance operational efficiency with environmental sustainability.

While deep learning models have seen significant success across various domains, their black-box learning nature and lack of interpretability affect their reliability in safety-critical applications like medical diagnostics and autonomous vehicles. In an attempt to address these limitations, Bayesian neural networks (BNNs) offer a promising alternative by incorporating uncertainty estimation into model predictions, enhancing transparency and decision-making. However, BNN development has primarily focused on efficient, high-fidelity approximate inference and guaranteed convergence in asymptotic settings. These are unsuitable for modern high-dimensional, multi-modal, and non-asymptotic deep learning applications, undermining their theoretical advantages. To bridge this gap, this paper provides in-depth reviews on how approximate Bayesian inference leverages deep learning optimization to achieve high efficiency and fidelity in high-dimensional spaces and multi-modal loss landscapes. It also reconciles Bayesian consistency with generalization objectives in non-asymptotic settings and investigates the generalization capabilities of BNNs. Additionally, this survey examines the often-overlooked expressiveness of BNNs, emphasizing how weight uncertainty and the absence of in-between uncertainty affect their performance. This survey aims to inspire BNN practitioners to adopt a deep learning perspective and offer valuable insights to propel further advancements in the field.

Abstract Trajectory tracking with disturbance rejection is a challenging problem in robotics, particularly in applications involving selective compliance articulated robot arms (SCARA). In this paper, we address the trajectory tracking problem with the presence of disturbances in applying SCARA, by designing controllers with active force control (AFC)-based control methods. AFC has shown potential in disturbance rejection, and its per efficiency of the designed controllers, we integrated different machine learning techniques into the AFC controller, including iterative learning (IL), adaptive neuro-fuzzy inference system (ANFIS) and reinforcement learning (RL). Two case studies were conducted and compared with two different benchmark controllers to validate intelligent AFC-based controllers: a port-controlled Hamiltonian (PCH) control and a hybrid proportional-integral-derivative (PID) control. The results demonstrate that the AFC-based controllers consistently outperform the benchmark methods. Specifically, in Case 1, the AFC-RL controller achieves a 99.99% improvement in root mean square error for joint 1 compared to the hybrid PID control. In Case 2, the AFC-RL controller outperforms the AFC-IL controller in trajectory tracking accuracy by 98.71%. Also, disturbance rejection ability was tested on the AFC-based controllers with various types of disturbances. Among the three AFC-based controllers, AFC-RL shows the best performance. The findings highlight the potential of integrating machine learning into AFC for more accurate and efficient robotic control.

This article presents reachable set synthesis for a discrete-time Markov jump system (DTMJS) with mode-dependent time-varying delays, subjected to uncertain transition probabilities and actuator faults, based on the ellipsoidal attracting approach. The focus is mainly to reflect more realistic control behaviors for the proposed DTMJS, in which the class of partially asynchronous reliable control (PARC) scheme is designed for the first time under the Markov switching topology. In this regard, the state-feedback and mode-dependent time-varying delayed state-feedback controllers are coupled by employing the Bernoulli variable. Under this framework, the hidden Markov model is formulated, revealing the asynchronism among switching topology, controller, actuators and proposed system in different operational modes. By constructing a double mode-dependent stochastic Lyapunov-Krasovskii functional, the sufficient conditions are derived in terms of linear matrix inequalities, which not only ascertain the stochastic stability of the resultant Markov jump system but also ensure that all reachable states remain within compact ellipsoidal boundaries. Finally, numerical simulations are provided to verify the effectiveness and merits of the presented method.

The emerging vehicle-to-vehicle communication technique enables vehicle platoon control, which greatly increases road throughput and travel efficiency. For cybernetically connected vehicles (CVs) with dynamics heterogeneities subject to actuator uncertainties, a dynamic event-triggered platoon control protocol is proposed in this paper to significantly reduce communication overheads. By considering the platoon problem of CVs as an output tracking consensus problem of heterogeneous multi-agent systems (MASs), a hierarchical control framework composed of an upper-level interactive event-triggered observer layer and a lower-level local tracking controller layer is established. Within the proposed framework, considering a virtual dynamic leader with non-zero inputs and external disturbances, a distributed observer is first designed for each following vehicle to observe the leader in an event-triggered manner. An internal dynamic variable is introduced to construct the dynamic triggering law, which not only relaxes the requirement on continuous state transmission in triggering detection but also benefits the exclusion of the Zeno behavior. Then, the solutions of a set of regulator equations associated with the proposed observer and the observer itself are integrated into the local tracking controller design to guarantee the tracking ability of each follower agent to its own observer. As thus, an event-triggered platoon control mechanism of heterogeneous CVs is proposed. Simulations and experiments on unmanned ground vehicles (UGVs) are conducted, which validate the effectiveness of the proposed platoon control method.

This paper studies the multi-port berth allocation and time-invariant quay crane assignment problem with speed optimization (MBACAP). In the MBACAP, each vessel visits multiple ports, while the port operators allocate berths and quay cranes (QCs) to the arriving vessels. The objective of the MBACAP is to minimize the sum of vessel fuel consumption cost, waiting cost, delay cost and QC handling cost while meeting the constraints related to vessel sailing, berth allocation and QC assignment. A mixed integer liner programming (MILP) model for the MBACAP is formulated for the first time, and an improved adaptive large neighborhood search (IALNS) algorithm is developed to solve it. In the IALNS, an initial solution generation strategy is proposed, while modified removal and insertion operators combined with the MBACAP features are devised. Moreover, several speed decision operators are proposed to optimize vessel speed during insertion operations. The proposed MILP and IALNS are tested on the MBACAP instances based on real data. The numerical experimental results show that the MILP can be solved by CPLEX optimally for some small scaled instances, while the IALNS can efficiently solve instances of all scales. The importance of considering QC assignment in MBACAP and the influence of vessel fuel consumption in ports are analyzed by comparing the numerical results of different models. In addition, the impact of different fuel prices on the MBACAP is investigated through sensitivity analysis with management insights provided.

Knowledge-based Visual Question Generation aims to generate visual questions with outside knowledge other than the image. Existing approaches are answer-aware, which incorporate answers into the question-generation process. However, these methods just focus on leveraging the semantics of inputs to propose questions, ignoring the logical coherence among generated questions (Q), images (V), answers (A), and corresponding acquired outside knowledge (K). It results in generating many non-expected questions with low quality, lacking insight and diversity, and some of them are even without any corresponding answer. To address this issue, we inject logical verification into the processes of knowledge acquisition and question generation, which is defined as LVˆ2-Net. Through checking the logical structure among V, A, K, ground-truth and generated Q twice in the whole KB-VQG procedure, LVˆ2-Net can propose diverse and insightful knowledge-based visual questions. And experimental results on two commonly used datasets demonstrate the superiority of LVˆ2-Net. Our code will be released to the public soon.

This study focuses on distributed event-triggered consensus control under the scenario where only inaccurate agent model information is available. By designing a novel triggering error, a distributed dynamic event-triggered consensus (DETC) protocol is proposed for multiagent systems (MASs) with general linear dynamics over digraphs, without accurate a priori information of agent models. To improve the efficiency of the dynamic triggering law, a mixed triggering threshold is designed with a resilient function integrated to further enlarge interevent intervals. Within the proposed DETC protocol, the computational cost is significantly reduced especially for MASs with nonsparse and high-dimensional agent system matrices. In addition, for each individual agent, the states of neighboring agents used for triggering detections or controller updates are required in an on-demand (instead of continuous) way, which preserves communication resources and facilitates practical implementation. The feasibility of the designed DETC protocol is corroborated by rigorous theoretical analysis on consensus convergence and Zeno behavior exclusion. Finally, simulations are shown to demonstrate the effectiveness of the studied theory.

This article analyzes the stability and stabilization of interval type-2 (IT-2) Takagi–Sugeno (T–S) fuzzy systems by proposing novel time-dependent memory sampled-data control (TDMSDC) scheme based on the sampling-dependent functional approach. Unlike the existing studies, the novel TDMSDC, incorporates both conventional sampled-data control (SDC) and memory SDC schemes, along with the relation of sampling-time variable and time changes within the sampling period, is introduced. The novel TDMSDC overcomes the limitation of conventional SDC schemes, which only stays constant within the sampling period, thus enhancing control performance. Next, a sampling-dependent looped Lyapunov–Krasovskii functional, integrating information on sampling periods with various degrees, fuzzy membership function (MF), sampling-dependent matrices, and signal transmission delay, is constructed. Leveraging this novel functional and constraint condition of MFs, the asymptotic stability conditions are obtained in terms of linear matrix inequalities for IT-2 T–S fuzzy systems under the novel TDMSDC technique. Finally, the numerical examples demonstrate the superiority and efficiency of designed control method.

Precise and reliable measurement of wind power uncertainty plays a significant role in the economic operation and real-time control of the smart grid. In this paper, a novel spiking neural network (SNN) architecture is proposed for solving regression tasks, and a multi-objective gradient descent (MOGD) algorithm is employed to generate high-quality wind power prediction intervals (PIs). SNNs improve upon conventional artificial neural networks (ANNs) by encoding interneuron communication into temporally-distributed spikes, which reduce memory access frequency and data communication, and therefore, the computational power requirements of deep learning workloads. This becomes exceedingly important for continual data analysis in remote geographic regions which often lack reliable cloud access and power supply, where many wind power farms are stationed. Given that neuron spikes are all stereotypically treated to be identical, they are a natural fit for tasks that may conflict in a common network architecture, such as multimodal data or where multiple, potentially competing, objectives are being optimized for. This paper proposes an SNN architecture that achieves comparable performance with its ANN counterpart on a complex regression task, i.e., wind power interval prediction. The resulting multi-objective SNN demonstrates superior performance as compared with those from state-of-art ANNs in wind power interval prediction.

Modern home energy management systems (HEMSs) have great flexibility of energy consumption for customers, but at the same time, bear a range of problems, such as the high system complexity, uncertainty and time-varying nature of load consumptions, and renewable sources generation. This has brought great challenges for the real-time control. To solve these problems, we propose an HEMS that integrates a kernel-based real-time adaptive dynamic programming (K-RT-ADP) with a new preprocessing short-term prediction technique. For the preprocessing short-term prediction, we propose a gated recurrent unit-bidirectional encoder representations from the transformer (GRU-BERT) model to improve the forecasting accuracy of electrical loads and renewable energy generation. In particular, we classify household appliances into the temperature-sensitive loads, human activity sensitive loads, and insensitive/constant loads. The GRU-BERT model can incorporate weather and human activity information to predict load consumption and solar generation. For real-time control, we propose and employ the K-RT-ADP HEMS based on the GRU-BERT prediction algorithm. The objective of the K-RT-ADP HEMS is to minimize the electricity cost and maximize the solar energy utilization. To enhance the nonlinear approximation ability and generalization ability of the adaptive dynamic programming (ADP) algorithm, the K-RT-ADP algorithm leverages kernel mapping instead of neural networks. Hardware-in-the-loop experiments demonstrate the superiority of the proposed K-RT-ADP HEMS over the traditional ADP control through comparison.

The berth allocation and quay crane assignment problem (BACAP) is an important port operation planning problem. To obtain an effective and reliable schedule of berth and quay crane (QC), this study addresses the BACAP with stochastic arrival times of vessels. An efficient method combining scenario generation is presented to simulate the stochastic arrival times. After then, a mixed integer linear programming (MILP) model is established, aiming to minimize the expectation of the vessels’ total stay time in port. A multi-objective constraint-handling (MOCH) strategy is adopted to reformulate the developed model, which converts constraint violations into an objective, thus transforming the single-objective optimization model with complex constraints into a dual-objective optimization model with only easy-handling constraints. Then an enhanced non-dominated sorting genetic algorithm II (ENSGA-II) is proposed to solve the dual-objective model, in which a neighborhood search algorithm and a search bias mechanism are incorporated to strengthen the local exploitation capability. Furthermore, a repair method (RM), penalty function (PF) and the superiority of feasible solutions (SF) strategy for constraint handling are designed respectively and incorporated with genetic algorithm to solve the original single-objective optimization model. Finally, numerical experiments on instances in the literature are conducted to validate the effectiveness of the MOCH and the proposed ENSGA-II. The results show that the average total stay time of vessels is reduced when stochastic arrival times are considered. Comparison results with another two multi-objective methods and three single-objective methods combined with different constraint-handling strategies corroborate the superiority of the proposed ENSGA-II and MOCH.

This study employs the Caputo-Fabrizio fractional derivative to determine the model of fractional-order RC circuits with arbitrary voltage input which can be widely used in a variety of electrical systems. Analog circuit implementation of fractional-order RC circuits defined by Caputo-Fabrizio fractional derivative is presented and verified by comparing with the model proposed in this work. For the purpose of judging whether the fractional-order model defined by the Caputo-Fabrizio derivative is practical, the comparison experiments are carried out. By using Laplace transform, the analytical solutions of fractional-order RC circuits based on the Caputo-Fabrizio derivatives with constant and periodic voltage sources are deduced. Fractional-order model of RC circuits with arbitrary input are also calculated using the convolution formula. The correctness of the derivation of the model using the Caputo-Fabrizio derivative is verified. Through discussing the impedance model of capacitor in frequency domain, the analog realization of fractional capacitor based on the Caputo-Fabrizio derivative is derived. The fractional-orders of the RC circuits models defined by the Caputo and Caputo-Fabrizio fractional derivatives are fitted respectively through repeated charging and discharging experiment data. The fractional-order models based on the Caputo and Caputo-Fabrizio derivatives, and the integer-order model are all compared with the experiment data.

In this paper, we attempt to solve the dual-objective short-term hydro-thermal-wind co-scheduling (HTWCS) problem by proposing a novel multi-objective constraint handling (MOCH) based search framework. The two objectives of this problem are minimizing the cost and emissions associated with electric power generation, while satisfying various hydraulic and electric constraints. The violations of constraints of the HTWCS problem is innovatively converted to another objective and forming a triple-objective HTWCS problem, which is further solved by a Borg algorithm, consisting of i) ε-dominance based archive updating process, ii) ε -progress based restart strategy and iii) auto-adaptive multi-operator recombination mechanism. The sophisticatedly designed MOCH based Borg evolution framework guarantees the convergence capability and diversity and can avoid the blindness in selecting the recombination operator and complex constraint repairing strategies, due to the lack of priori knowledge of the HTWCS problem. Comparison among different approaches demonstrates the approach proposed in this paper outperforms its conventional counterparts and the MOCH technique performs better than the conventional constraint repairing strategies when applied to solve the HTWCS problem.

This paper proposes a load-oriented control parameters optimization strategy for Doubly Fed Induction Generator (DFIG) to enhance Low-Frequency Oscillation Damping (LFOD) and improve stability of a power system. Enabled by the smart grid measuring technologies, frequency deviations of generators of interest are obtained and employed as the input signals of the designed Supplementary Damping Controller (SDC) of DFIG. In order to acquire the optimal load-oriented control parameters, an hour-ahead load-forecasting scheme is devised, using Artificial Neural Network (ANN) learning techniques. The ANN is trained by a set of data over a 4-year period, and then the control parameters are optimized using Particle Swarm Optimization (PSO) technique for the purpose of minimizing the Critical Damping Index (CDI) of the power system. Numerical results demonstrate that the low-frequency oscillations (LFOs) of the power system can be effectively mitigated using the proposed controller in smart grids integrated with wind power generators.

Nowadays, enabled by current smart grid technology, electricity consumers can play an active role in providing ancillary service (AS) as a type of demand response. Participating AS can assist stabilizing the power grid by following the frequency regulation signal, or dynamic regulation signal (RegD) in this study while receiving economic benefits. Industrial loads are an indispensable component as a demand-side regulating resource of ancillary service due to their intensive electricity consumption. In this paper, we use grid-connected solar photovoltaics panels combined with the energy storage system (ESS) to produce continuous electricity consumption signals in order to follow the RegD signal. The participation of solar energy in real-time regulation provision process is emphasized, which is modeled based on a variety of operation modes in accordance to Australian Standard. Through a particular case study, with the integration of solar energy, the proposed method poses cost-effectiveness in industrial plant scheduling and a favorable load following capability, helping ensure the frequency stability of the electric grid. The proposed methodology is more economically advantageous compared to identical industrial loads only equipped with on-site ESS, and requires less switchings on machines compared to industrial plants with passive use of solar energy.

This paper proposes a load-oriented control parameter optimization strategy for static synchronous compensator (STATCOM) to enhance low-frequency oscillation damping (LFOD) and improve stability of overall complex power systems. Frequency deviations of generators of interest are employed as the input signals of the designed supplementary damping controller of STATCOM. In order to obtain the optimal load-oriented control parameters, a day-ahead load-forecasting scheme is devised, using artificial neural network (ANN) learning techniques. The ANN is trained by a set of data over a 4-year period, and then the control parameters are optimized using Particle Swarm Optimization technique by minimizing the critical damping index. The proposed control strategy is implemented in the IEEE standard complex power system, and the numerical results demonstrate that the low-frequency oscillations (LFOs) of the power system can be effectively mitigated using the proposed controller. Compared to conventional robust controller with universal parameters, this novel load-oriented optimal control strategy shows its superiority in alleviating LFOs and enhancing the overall stability of the power system. Since the proposed control scheme aims to adaptively adjust the controller parameters in correspondence to load variations, this study is envisaged to have practical utilizations in industrial applications.

This paper enumerates three commonly used filtering algorithms and shows the detailed steps of their incorporation with general nonlinear systems for dynamic state estimation. The mathematical model of a stand-alone solid oxide fuel cell (SOFC) is briefly discussed and derived, which is then mathematically connected to a multiarea, diverse-generator, interconnected complex test system. The mathematical representation of the entire power system is tailored into a certain compact form to provide suitability for the implementation of filtering algorithms for the design of dynamic state estimators. With the utilization of phasor measurement units, the state estimators are able to work in a decentralized manner with the mere knowledge of local noisy voltage and current measurements. Successfully estimating the internal dynamic states of SOFC connected to complex power systems offers a novel methodology for the acquisition of the internal unmeasurable states of SOFC, which will facilitate future controller designs that may require the otherwise inaccessible states.

With power networks undergoing an unprecedented transition from traditional power systems to modern electric grids integrated with renewable energy sources, maintaining frequency stability of generators in modern power systems has become one of the major concerns. Targeting this issue, in this paper, we propose a novel frequency restoration strategy in photovoltaics (PV)-connected power systems using decentralized dynamic state estimation technique and PV power plant as a contingency power source. When a sudden increase in load demand occurs, the output power of PV panels is increased in order to compensate for the shortage of real power capacity of the generator, in order to restore the frequency of a certain generator bus bar. An unscented Kalman filter-based decentralized dynamic estimation is utilized in this study to estimate the frequency of a selected generator bus bar with local noisy voltage and current measurement data acquired by using phasor measurement units. Solar luminous intensity may vary over a period of time in different seasons, weather conditions, etc., which causes the variations in the output power of PV power plants. This irradiance uncertainty is also considered in this study. The proposed control strategy not only incorporates the frequency deviations of a generator bus-bar, but also takes into account the tie-line power deviations under disturbances. Simulation results demonstrate the capacity of proposed control schemes in restoring the frequency of generator bus-bars and also maintaining the tie-line power flowing between adjoining areas at it scheduled value.

This paper uses three popular stochastic filtering techniques to acquire the unmeasurable internal states of the doubly fed induction generator (DFIG) in order to realize the widely adopted control scheme, which involves the inaccessible state variable-stator flux. Filtering methods to be discussed in this paper include particle filter, unscented Kalman filter, and extended Kalman filter, where their mathematical algorithms are presented, their implementations in the DFIG wind farm connected to complex power systems are studied, and their performances are compared. The whole power system network topology is taken into consideration for the state estimation, but only local phasor measurement unit measurement data are required. The purpose of using different stochastic filtering techniques to estimate dynamic states of DFIG in power systems is to resolve the long-lasting issue of the unavailability of DFIG internal states used in the DFIG controller design.