Search

This paper proposes a joint optimization strategy for nonlinear systems subject to unknown external disturbances. The co-design problem is formulated as a two-player zero-sum game, where the control policy is treated as the first player and the control input error caused by aperiodic feedback is treated as the second player. Besides, a robust term is incorporated to the performance index to suppress the negative effect of external disturbances. Based on the game theory, the optimal performance index and the solution to the associated Hamilton–Jacobi–Isaacs (HJI) equation can be obtained. Then, the sampling intervals are optimized by designing an event-triggered condition according to Lyapunov direct method. Furthermore, a self-triggered strategy is introduced to predict the next triggering instant in advance, avoiding the requirement of continuous state measurements. Through critic-only neural network (NN) implementation, the event-based HJI equation is approximated by using adaptive dynamic programming technique. The closed-loop nonlinear system and the weight estimation error for the critic NN are both guaranteed to be uniformly ultimately bounded under the proposed aperiodic sampling mechanism. Finally, simulation results and comparison studies demonstrate the effectiveness of the proposed co-design approach.

This paper investigates the distributed model predictive control (MPC) problem of linear systems where the network topology is changeable by the way of inserting new subsystems, disconnecting existing subsystems, or merely modifying the couplings between different subsystems. To equip live systems with a quick response ability when modifying network topology, while keeping a satisfactory dynamic performance, a novel reconfiguration control scheme based on the alternating direction method of multipliers (ADMM) is presented. In this scheme, the local controllers directly influenced by the structure realignment are redesigned in the reconfiguration control. Meanwhile, by employing the powerful ADMM algorithm, the iterative formulas for solving the reconfigured optimization problem are obtained, which significantly accelerate the computation speed and ensure a timely output of the reconfigured optimal control response. Ultimately, the presented reconfiguration scheme is applied to the level control of a benchmark four-tank plant to illustrate its effectiveness and main characteristics.

In the refining process, Fluid Catalytic Cracking Unit (FCCU) end-point quality prediction plays an important role in real-time quality monitoring, optimization and control. Due to the process uncertainties (such as changes of raw materials, abrasion of mechanic components, catalyst deactivation, changes in the external environment, etc), data distribution of process data in FCCU is time-varying, which leads to accuracy degradation of the quality prediction model. Therefore, to deal with the time varying characteristic of process data and avoid accuracy degradation of the quality prediction model, real-time processing should be considered. In this paper, an enhancing incremental deep learning approach is proposed for the online quality prediction of the absorption-stabilization system in FCCU. First, the offline model is built by Stacked Auto Encoder-Deep Neural Network (SAE-DNN). To determine whether the data distribution has changed and model modification is needed, a concept drift detector is proposed for the regression problem by defining an error bound in the Statistical Test of Equal Proportions (STEPD). If the model modification is needed, then the top layer of the offline SAE-DNN model is expanded by Random Vector Functional Link (RVFL) structure, and the parameters in the expansion layer is dynamically assigned by the new coming data with the group lasso regularization and the L2 regularization. The proposed approach is validated by predicting the Saturated Vapor Pressure (SVP) of stabilized gasoline in the FCCU. The experimental results show that the proposed approach can deal with the time-varying characteristic of process data and avoid accuracy degradation under process uncertainties.

Precise measurement of tailor-welded blanks is crucial for quality control of laser welding. The difficulty in measuring the seam profile of similar gage tailor–welded blanks lies in lacking of solutions to locate their feature points such as those representing the laser stripe. A laser sensor–based method is proposed to measure tailor-welded blanks based on its laser stripe and texture features, from which the proposed algorithm is employed to extract the feature points for seam profile assessment. The algorithm is evaluated on a laser-based weld seam measurement system, the experiment results show that the proposed method measures the tailor-welded blanks with high accuracy and is suitable for online inspection.

This paper studies the sliding mode control for T-S fuzzy systems in the framework of finite-time boundedness. It is assumed that the control signals are transmitted via vulnerable channels, where injection attacks might happen. A fuzzy sliding mode controller is firstly synthesized to guarantee the finite-time reachability of the prescribed sliding surface and attenuate the effect of the injection attacks. By introducing a partitioning strategy, the finite-time boundedness over both the reaching phase and the sliding motion phase are analyzed. Furthermore, sufficient criteria are derived such that the closed-loop system is finite-time bounded over the whole specified finite-time interval despite of the injection attacks. An optimal algorithm is further provided for searching ideal control gains with fewer energy demands. Finally, a simulation example verifies the proposed sliding mode control approach.

Data-driven methods have been regarded as effective methods for modeling in chemical processes. However, with the increasing complexity of chemical processes in spatial domain and time domain, how to extract meaningful features and build corresponding models are keys for accurate modeling tasks. To retain temporal features of original inputs, a Recurrent Denosing Autoencoder (RDAE) is built to extract meaningful features and reduce input dimension in the spatial domain, and Cumulative Percent Variance (CPV) is introduced to decide the number of extracted features. Considering correction and prior knowledge effects of real history outputs, a Weighted Auto Regressive Long Short Term Memory (WAR-LSTM) structure is proposed as the basic cell, then multiple WAR-LSTMs are stacked as deep WAR-LSTM to extract high level representations from multi variables. Hence, spatial and temporal information are sufficiently used both in extracting features and building model. The benchmarked Tennessee Eastman process data and real process data from a Fluid Catalytic Cracking (FCC) unit verify the effectiveness of our work.

Traditional control system design to complex networks is generally implemented by integrated structural analysis aiming at a global network. However, such a global method may be inefficient, in particular, when a massive network with a huge number of nodes and associations is considered. In this paper, motivated by the idea of “dividing and dealing”, we propose a block-based approach to the issue of minimum input design for structural controllability (MIDSC) of complex networks that potentially incurs in higher efficiency. Specifically, we consider a large-scale networked system that consists of several local blocks. The main challenge for control configuration design of this class of systems is how to find the minimum inputs of global network according to the local block information while maintaining system’s structural controllability. To this end, two block-based graphical algorithms are developed to meet the conditions required for achieving structural controllability, and meanwhile determine an optimal solution for addressing the MIDSC problem. The complexity of the proposed method is analyzed, which is also compared with existing algorithms designed mainly based on monolithic model. In particular, we show that, under some mild conditions on blocking structure, the complexity of the proposed algorithm is strictly lower than that of existing algorithms to the MIDSC problem.

This paper presents a multi-time hierarchical stochastic predictive control (MHSPC) scheme for an island microgrid, in which electric vehicles (EVs) can be used as mobile energy storages to improve power balance and realise load-frequency control (LFC) with micro-turbines (MTs). At the upper layer, a stochastic model predictive control is proposed to handle the EVs uncertainties on a long time scale, while optimising controllable power adjustment of MTs, the charge/discharge energy of the battery electric storage (BES) unit and guaranteeing EVs to be fully charged at the expected plug-out time. At the lower layer, the coordination between EVs and MTs for LFC is achieved by a standard MPC framework on a short time scale. In this way, the power balance is met, and the frequency fluctuation is inhibited. Finally, simulation results are presented to illustrate the satisfactory operation of the island microgrid.

This paper is concerned with the security control problem for a class of Markov jump systems subject to false data injection attack and incomplete transition rates. An on-line estimation strategy is provided for the time-variant and unknown cyber-attack modes. And then, an adaptive sliding mode controller is synthesized with different robust terms for different modes to guarantee the reachability of the specified sliding surface. Moreover, the sufficient conditions for the stability of the closed-loop systems are derived. Finally, it is shown from simulation results that the effect of both false data injection attack and incomplete TRs can be effectively attenuated by the present adaptive SMC method.

Distributed multivariate statistical process monitoring (MSPM) is an efficient and widely studied monitoring framework for large-scale process. However, fault isolation of Bayesian fusion based distributed MSPM has not been sufficiently discussed. In this paper, enhancing comprehensive contribution plot method is proposed. Firstly, novel Bayesian contribution plot is presented to identify subsystem contributions to global fault. Subsequently, regular T2 and Q contribution plots are applied to calculate variable contributions to local subsystems. Finally, enhancing comprehensive contribution plot is constructed by combining them together. The proposed method is suitable for distributed scheme because it directly identifies responsible variables leading to global fault, providing convenience and improving efficiency in fault isolation. Tennessee Eastman (TE) process is chosen as an application study in this paper. Simulations results verify the efficiency of the proposed method.

This paper studies the consensus problem of multi-agents system with communication management. Communication management means a strategy about how to adjust the sensing range of agents in the consensus process for saving communication energy. We propose a communication management for multi-agents system with second-order dynamics so that with this strategy, different control methods can be applied and the designing work is simplified. And a predictive control method is presented to solve the consensus problem of multi-agents system with the communication management in this paper. Simulation results indicate the effectiveness of the predictive control method and the benefits for solving this problem.

In some practical systems, it often remains difficult to directly measure all state variables. This article investigates the memory output sliding-mode control (SMC) for the finite-time consensus of singularly perturbed multiagent systems (SPMASs). First, the virtual state-feedback sliding surface (SFSS) is constructed to ensure the consensus of all agent states. Then, the unknown output derivatives in SFSS are approximated by a moving finite difference method with error estimation and refinement, which gives rise to a new delay-dependent sliding surface. On this basis, the memory output switching control law is designed to stabilize the consensus errors in finite time, even in the presence of estimation biases, singular perturbations, and input noises. Different from the observer-based SMC, the proposed memory output SMC is of simple static form without introducing extra dynamical structures for state estimation. The effectiveness and superiority of the design method are verified in an SPMAS with double-integrator dynamics.

In this paper, the unmanned aerial velhicle (UAV) path planning problem with obstacles is disscussed for an air-ground collaborative system which is composed of a messenger UAV and multiple unmanned ground vehicles (UGVs). The messenger UAV needs to visit each UGV to collect and share information. The aim of this problem is to achieve a shortest path which makes the UAV fly over all UGVs while avoiding obstacles. By the Dubins model, the path planning problem is formulated as a Dynamic Dubins Traveling Salesman Problem with Neighborhood (DDTSPN) under obstacle constraints. To adapt to the change of UGVs’ locations and reduce the complexity of tuning global path for obstacle avoidance, a moving horizon optimization strategy is porposed. After the UAV visits a UGV, the rest of UGVs’ access sequence and access locations are determined using SOM method according to the real-time locations of UGVs. Further, the DAPF algorithm is presented to replan the path which enters the obstacle regions. Simulation results show the advantages of the proposed path planning algorithm.

In this paper, a multi-agent cooperative control problem with conflicting temporal logic tasks is studied. Each agent is assigned a temporal logic task which contains a motion task and safety requirements. We consider the cases where the satisfaction of both the motion task and safety requirements may be conflicting due to the limited velocity, so that such a task can not be fulfilled. In order to solve this problem, we give priority to the the safety requirements and the degree of satisfaction of the motion task is slacked. This work proposes a two-stage distributed receding horizon optimization strategy consisting of offline stage and online stage where signal temporal logic (STL) is utilized to formally describe the temporal logic tasks and the receding horizon optimization framework is adopted for cooperative collision avoidance tasks. At offline stage, according to the motion task, a reference robustness evolution curve is presented for each agent by the robust semantics of STL formulas. At online stage, based on the short-term goal region determined by the reference robustness evolution curve, together with the known obstacles' information and agents' real-time information, constraints of both the motion task and safety requirements are constructed in the receding horizon optimization problem for each agent. When conflicting situations happen, the constraint of the motion task is relaxed by a robustness slackness to find a least violating solution. In the proposed framework, the offline stage and the online stage are combined to satisfy the motion task as much as possible and to guarantee the safety requirements. The effectiveness of the framework is verified by simulation results.

This paper addresses the problem of sliding mode control for discrete-time interval type-2 fuzzy system subject to fading channels. The multiplicative fading model is established for reflecting the fluctuation of the transmitted signals. Meanwhile, the membership function-dependent sliding function involves the fading coefficient. By using the membership functions depending on the fading signals, the mismatched interval type-2 fuzzy sliding mode controller is designed such that the state trajectories can be driven into the neighborhood of the specified sliding surface. Moreover, the closed-loop fuzzy system is ensured to be mean-square asymptotically stable. Finally, the proposed control scheme is verified by a numerical example.

This paper proposes an observer-based event-triggered adaptive dynamic programming (ADP) approach for the nonlinear system with unknown disturbances to efficiently reduce computational cost. Under the event-triggered mechanism, an observer is constructed to identify unknown disturbances and ensure the estimation error to be ultimately uniformly bounded(UUB). Next, an adaptive triggering condition related to the estimation error of disturbances is derived with the help of a modified performance index function. Then, the event-based control algorithm is implemented based on a single critic neural network(NN) structure to approximate the optimal control law. Moreover, the stability analysis for the closed-loop system is presented with the event-driven control law and the weight estimated error of the critic NN is proved to be UUB. Finally, simulation results illustrate the effectiveness of the proposed control scheme. Compared with the traditional ADP approach, the proposed event-triggered strategy can reduce controller updates with guaranteed performance of the system.

This paper investigates the event-triggered distributed predictive control (DPC) problem for multi-agent systems subject to bounded disturbances. A novel event-triggering mechanism which involves the neighbours’ information is derived for each agent to achieve a trade-off between resource usage and control performance. In such a framework, the DPC optimization problem is solved and information is exchanged only at triggering instants, thus achieving asynchronous coordination. To lower computation and communication consumption more significantly, a dynamic variable considering effects of neighbours is introduced to design a dynamic event-triggering condition and we show that larger inter-execution time can be obtained using the dynamic triggering mechanism. The theoretical conditions on ensuring feasibility and closed-loop stability are developed for these two triggering mechanisms, respectively. Finally, numerical simulations are given to illustrate the effectiveness of the proposed control strategy.

The adaptive sliding mode control (SMC) problem is addressed for singular semi-Markovian jump systems (S-MJSs) against actuator attacks, in which the transition rates rely on the random sojourn time and are not constant, and the system states are unavailable. Moreover, the vulnerability of control signals transmitted via communication network means that the actuators may receive the attacked control signals. For the sake of reducing the effect of actuator attacks, the neural network technique is used to approximate the false information injected by adversaries. Meanwhile, a sliding mode observer is introduced to estimate the unmeasured states. An adaptive SMC law is proposed to guarantee that the estimation states and errors can reach to the sliding surfaces, and the stochastic admissibility of the singular S-MJSs can be ensured. In the end, an example is applied to illustrate the method in this paper.

In this work, the problem of asynchronous sliding mode control (SMC) is investigated for a class of uncertain Markovian jump systems (MJSs) with time-varying delays and stochastic perturbation. It is assumed that the system modes cannot be obtained synchronously by the controller, but instead there is a detector that provides estimated values of the system modes. This asynchronous phenomenon between the system modes and controller modes will be described in this work via a hidden Markov model with partly accessible mode detection probabilities . Based on a common sliding surface, an asynchronous SMC law depending on the detector mode is synthesized to ensure the mean square stability of the sliding mode dynamics and the reachability of the specified sliding surface simultaneously. Moreover, a design algorithm for obtaining the asynchronous SMC law is established. Finally, an application of the automotive electronic throttle system is provided to illustrate the effectiveness and advantages of the proposed asynchronous sliding mode control approach.

In process industries, due to the low sampling rate of the quality variables, there are abundant unlabeled data but limited labeled data. Most data-driven quality prediction models only use the labeled data but ignore the unlabeled data, resulting in overfitting and low generalization performance. Hence, it is necessary to extract useful information from the unlabeled process data. Due to the noise in the signal transmission process and sensors, there are unlabeled data with low confidence that will mislead the training process. To tackle this problem, we propose a semi-supervised Generative adversarial network with Co-trained Generators (GCG) that utilizes the unlabeled data safely through the co-training of generators. The optimal parameters and weight coefficients of co-trained generators guarantee the ”safeness”, i.e., the performance of GCG is not worse than generators with only labeled data. The proposed method is validated by a benchmarked industrial case and the real absorption-stabilization system in Fluid Catalytic Cracking Unit (FCCU). The results suggest that the GCG method improves the generalization performance of the quality prediction model.

This paper addresses the problem of quantized [Formula: see text] filtering for multi-output discrete-time systems over independent identically distributed (i.i.d.) fading channels and Markov fading channels, respectively. The measurement outputs are quantized by a logarithmic quantizer, and then transmitted to the filter over fading channels. For the i.i.d. fading channels, the stochastic multiplicative noise form is used to model the unreliable communication environment. For the Markov fading channels, a set of Markov channel state processes is introduced to model time-varying fading channels, which characterizes various configurations of the physical communication environment and/or different channel fading amplitudes. The sufficient condition for stochastic stability with a prescribed [Formula: see text] performance is obtained by using a Lyapunov method and matrix decoupling technique. The corresponding filter design casts into a convex optimization problem. Finally, simulation results are provided to illustrate the effectiveness of our results.

Communicating via mixed wired-wireless connections is the development trend for large-scale distributed control systems. In this communication environment, due to the limited wireless resources, the communication patterns of subsystems have changed fundamentally, and the design of time-triggered and event-triggered distributed controllers should be taken into account simultaneously. The objective of this paper is to investigate mixed time/event-triggered dual-mode distributed predictive control (DPC) for constrained large-scale linear systems subject to bounded disturbances. Considering the effects of two different communication modes and introducing a prediction error between the current actual state and predicted state, the event-triggering condition is derived for each event-triggered subsystem. Based on this, a mixed time/event-triggered dual-mode DPC algorithm is proposed in view of the asynchronous coordination among subsystems. Furthermore, the sufficient conditions to ensure the recursive feasibility and closed-loop stability of mixed triggered DPC are developed. Finally, a multi-vehicle control system is provided to verify the effectiveness of the proposed approach.

This paper considers the cooperation control problem for a team of dynamically decoupled agents with resource constraints. A codesign of self-triggered mechanism and distributed model predictive control (DMPC) is proposed to achieve the cooperative objectives while efficiently exploiting communication network. The proposed self-triggered DMPC (ST-DMPC) possesses three important features. First, the communication cost is explicitly incorporated in the cost function. In this way, the triggering instant and control inputs are simultaneously optimized, and a desired tradeoff between control performance and communication cost is achieved. Second, at triggering instants, the first element of the optimal control input sequence along with the current state instead of the whole trajectory is broadcast to neighbors for cooperation, which further reduces communication load. Third, sufficient conditions on design parameters related to predictive states of neighbor agents are constructed to ensure stability of the overall system. The application of the proposed ST-DMPC to four robot manipulators validates the effectiveness of this method.

Model predictive control(MPC) poses an significant role in process industry. However, due to the characteristics of difficult to solve and poor control performance of nonlinear MPC with disturbances, this paper proposes an iterative unit-based adaptive dynamic programming (IU-ADP) aiming to solve the nonlinear MPC problem with disturbances. In the IU-ADP structure, an actor-critic scheme is adopted together with the iterative method, in which the critic network intends to approximate the performance index, the actor network with an iterative method aims to approximate the optimal control. Moreover, IU-ADP with data parallelization can achieve fast time efficiency. Finally, it is tested on the fluid catalytic cracker unit (FCCU) plant compared with the traditional MPC method. The results illustrate that IU-ADP obtains a little better control performance.

In this work, the discrete-time sliding mode control problem for the networked control systems under fading channels is investigated. Moreover, the state signals will be quantized through a logarithmic quantizer before transmitted to the controller. Subsequently, a sliding surface relating to fading coefficient is designed to cope with the fading measurements. It is shown that the designed sliding mode controller can drive the state trajectories into the neighborhood of the specified sliding surface and the closed-loop system is asymptotically stable in the sense of mean square. In the end, a numerical simulation is given.

For a networked system with structural controllability, components will be added in the process of operation because of function increase or size increase, which may result in structural uncontrollability of the networked system under original control inputs. With network topology and original control inputs preserved, designing more control inputs or coupling relations to guarantee structural controllability is an important problem. We propose two strategies of adding minimum extra inputs or edges to render a structurally controllable system: 1) through adjusting network topology, minimum input addition problem is transformed into a minimum input selection problem of a simplified network topology, and then a set of minimum additional control inputs is chosen to ensure structural controllability; 2) combining maximum matching sets with greedy algorithm, minimum number of directed edges are added to the system topology to obtain structural controllability. The proposed strategies are able to find an optimal solution of minimum input/edge addition in polynomial time with a low time complexity. We illustrate the validity of our strategies via several graph examples.

In this paper, a multi-agent formation control strategy with collision avoidance and obstacle avoidance in complex environments is proposed. Under a virtual leader structure, the multi-agent formation is realized through path planning and formation control. In path planning, a transitional region based on prediction horizon is used to replan reference trajectories and avoid obstacles in advance. The virtual-leader alternation strategy and feasible state searching strategy are adopted to avoid one side obstacles and crowded obstacles, respectively. Then the path planning optimization problem is formulated by mixed-integer quadratic programming (MIQP) to calculate trajectories collision-free with obstacles. In formation control, a distributed model predictive control (DMPC) algorithm is developed to form and maintain formation. In DMPC optimization problem, the non-convex obstacle avoidance constraints are replaced with deviation constraints and linear constraints to design terminal invariant set and reduce online computation burden. Under the whole formation control strategy, the tasks of obstacle avoidance and collision avoidance are both achieved and the agents maintain the formation as much as possible. Finally, the example with different obstacle situations is discussed to illustrate the effectiveness of the proposed strategy.

The problem of congestion control with balanced-energy is important for application of WSNs, since the limited resources and many-to-one communication model often result in congestion and unbalanced energy consumption. In this study, a hierarchy-based congestion control and energy-balanced scheme is proposed. The network model is firstly initialised into a hierarchical topology, by which these neighbour nodes of a node will be explicitly divided into three kinds, i.e., the same hierarchical nodes, the upstream nodes, and the downstream nodes. Then, in the proposed congestion avoidance method, the node will use other lower hierarchy neighbour nodes to forward data when its downstream node will be congested. After that, the congestion control mechanism will detect the congestion via the queue length, forwarding and receiving rate, and inform its upstream nodes to find other next hop to release the congestion. The balanced energy consumption strategy will balance the energy consumption of lower hierarchy nodes by using the node with the most remaining energy. Meanwhile, by using the same hierarchy nodes, the remaining energy of all the nodes in the same hierarchy is balanced. Simulation results show that the proposed algorithm can effectively deal with the network congestion and unbalanced energy consumption.

This paper gives a new systematic framework to solve the asynchronous sliding mode control (SMC) problem for the semi-Markovian jump systems with singular perturbations, which can effectively remove the dependence on the unavailable system mode of SMC controller in some existing works. By considering the effect of the singular perturbation, this paper introduces a novel e -dependent common sliding function, and then an asynchronous SMC law is developed by just using the detected mode signal. In order to overcome the difficulties in analyzing the stability of the sliding mode dynamics and the reachability of the specified sliding surface, a novel matrix decoupling technique and a novel semi-Markovian Lyapunov function method are introduced. Besides, genetic algorithm (GA) is employed to solve two meaningful optimization problems arising from designing a desired asynchronous SMC scheme. Finally, an operational amplifier circuit with a switching positive temperature coefficient thermistor is simulated to show the practicability of the proposed asynchronous SMC approach via GA.

The recognition of grasp type is essential for a more detailed analysis of human action. In this paper, we propose a novel method for real-time grasp type recognition using Leap motion controller (LMC). Our proposal is based on the tracking data provided by the LMC sensor and a series of feature descriptors are introduced and extracted from LMC data. Combining the feature descriptors of relative positions of thumb, finger joint angles and finger directions lead to the best representation of the arrangement of the fingers. And then the grasp type classification can be achieved by using a SVM classifier. An experimental study of our approach is addressed and we show that recognition rate could be improved. The current implementation is also can satisfy the real-time requirements.

This paper studies the cooperative stabilization problem of large-scale systems composed of several dynamically coupled subsystems. To drive these subsystems towards global performance in an effective communication way, a self-triggered distributed model predictive control algorithm is proposed. In this algorithm, the triggering instants are introduced to the index function to estimate communication cost and are optimized jointly with the control input sequence by minimizing both control performance and communication cost. At its triggering instants, the local subsystem actively makes a request to all the neighbors and receives their state measurements instead of their predictive state trajectories, which is potential to improve the utilization of communication information and to relieve communication burden. Furthermore, the stability of the overall system is analyzed. Conditions for the terminal matrices and the design parameters used to produce the predictive state trajectories of neighbor subsystems are established to guarantee closed-loop system stability.

This paper focuses on a two-stage framework for economic optimization to maximize the profits of electric-vehicle (EV) microgrid. In the first stage, an economic optimization problem at the day-ahead time scale is solved to determine the power purchased from load serving entities (LSE), and make an optimal price decision (parking fee and charging fee) while considering with the EVs uncertainties. In the second stage, a real-time model predictive control strategy is proposed to meet the EVs requirement and minimize the operation cost. Through two-stage scheduling, EV microgrid can guarantee long-term safe and efficient operation, while ensuring maximum benefits. The simulation results show that the proposed method in this paper can provide reliable power supply to the EVs, increase the EV microgrid revenue and ensure the safe operation of the EV microgrid system.

This paper studies product yields forecasting for fluid catalytic cracking unit (FCCU). Conventional product yields forecasting is usually based on mechanism model, which may ignore some significant factors due to manual approximations. Deep learning methods can extract features automatically based on data without prior knowledge. Considering bidirectional temporal features and spatial features of FCCU, deep bidirectional long-short-term memory (DBLSTM) network is proposed for product yields forecasting. The bidirectional structure can capture bidirectional temporal features of FCCU by considering previous information as well as future information over a period of time. Significant spatial features of sensors at each time step can be extracted automatically through a deep structure by stacking multiple bidirectional structures. Moreover, the deep bidirectional LSTM network can deal with long-term dependencies by integrating deep bidirectional structure with LSTM cell. To avoid overfitting, regularization adopted in this paper is dropout and early stopping. Efficacy of the DBLSTM approach is demonstrated by process data from an actual FCCU in China. Through the comparison of mean squared error on product yields forecasting, the DBLSTM approach is superior to traditional regression models and other recurrent models.

In order to enhance microgrid resilience, this paper presents a hierarchical nested predictive control (HN-PC) strategy for multi-microgrids against unexpected disaster events. At nested structure layer, by distinguishing the types of loads, the microgrids with critical ones are nested in the inner level to ensure their safe operation and reduce the influence caused by microgrid under fault condition. At central scheduling layer, the power balance of microgrid in the short term can be improved and safe operation in the long term are guaranteed through the coordination of microgrids and electric vehicles (EVs). The simulation results show that the method proposed in this paper can provide reliable power supply to the fault condition of microgrid, meet the power balance, reduce the impact of the failure, and ensure the safe operation of the multi-microgrids system.

In this paper, the energy management of the combined cooling, heating and power (CCHP) microgrid with both the supply side and the demand side are considered. According to the working characteristics of the energy storage devices and other units, the energy management model of the CCHP microgrid with the load curtailment is established. On this basis, the microgrid energy management algorithm based on the model predictive control (MPC) is proposed, which can not only achieve multi-energy cooperation, but also give full play to the comprehensive scheduling ability of the supply side and load curtailment strategy of the demand side. The impact of intermittent renewable energy supply on microgrid is suppressed and the operation costs of the CCHP microgrid is reduced.

In the operation of urban water distribution networks (WDNs), the supply pressure to users should be kept low while meeting the lower bound to satisfy users’ water demand, reduce leakages and power consumption. In this paper, a multi-model based pressure optimization method for large-scale WDNs is proposed, where the pressure of the whole WDNs is represented by the pressure of key nodes and key pump station nodes based on the calculation of pressure transfer relationship and a developed network simplification method, the multiple linear models are constructed to approximate the pressure transfer relationships, and the models between the selected key nodes and key pump station nodes are used for the pressure optimization of the whole WDNs. The proposed methods are operation data based, which overcomes the difficulty in constructing accurate and complicated hydraulic models for large-scale WDNs. The network simplification reduces data requirements in multiple model construction and ensures real-time control. In the simulation case, which is part of Shanghai’s WDNs, the supply pressure to users is optimized to lower and more homogeneous value without violating the lower bound by applying the proposed approach, which is significant in reducing water leakages and energy consumption.

Studying from the characteristics of the two stagesin the high-tech industry, combining with the practice anddeficiency of the actual measurement, wesummarize China's high-tech industry innovation efficiency calculationmethodsand explore the one of DEA methodin depth to construct a new index system, on which a new internetDEA model (a methodological innovation) is constructed. This new model can be applied to get overallefficiency and the efficiency of the two phases in high-tech industry R&D activities. In the end, the articleshows the application of this model: the ability to analyze China's high-tech industrial innovation efficiency andthe efficiency of high-tech industry's R&D activity in different regions and will have implications for thegovernmental policy-making.

In this paper, a new control strategy was proposed to deal with the discrete-time multi-agent consensus problem. Two types of protocols are discussed in this paper: i) networks of single-integrators without delay under centralized event-triggered control and ii) networks of single-integrators with delay under distributed event-triggered control. For each consensus protocol, we prove that the multi-agent network will achieve consensus asymptotically. Numerical examples are provided to demonstrate the effectiveness of the obtained theoretical results.

This paper proposes a strategy of event-triggered distributed predictive control (DPC) for large-scale systems with dynamic couplings. The event-triggering condition which only involves local information of the subsystems has been derived based on the input to state stability theory. In the propose scheme, all subsystems optimize with decoupled cost functions and constraints only when the event-triggering conditions are satisfied. The dynamic couplings as well as disturbance can be handled through a robustness constraint in the local optimization. In addition, a dual-mode control scheme is adopted to further save computation resources. Several sufficient conditions are developed to ensure the recursive feasibility and close-loop stability of event-triggered DPC. Finally, the effectiveness of the proposed approach is illustrated via four-tuck systems.

This paper studies the routing algorithm based on the stability in mobile AdHoc[1][2][3] network and presents a routing mechanism based on neighbor stability. We put the mechanism in multicast routing protocol MAODV[4] and propose a improved routing algorithm NBS-MAODV. Simulation results show that NBS-MAODV algorthm reduces the link fracture times and the total overhead of the network. It also improves the packet delivery ratio.

This paper deals with a seam tracking system of welding of Tailored Weld Blanks based on active laser-triangulation. The system consists of DSP based vision camera and stripe-type laser diode. The total system is assembled into a compact module which can be attached ahead of welding torch and it has high dynamic performance in guiding the robot arm welder accurately over the seam. its structure, basic principle of this system and the main function are introduced . The images taken by the camera are analyzed using image processing algorithms: setting the range of interest, the median filtering and bi-level thresholding. Furthermore, the middle line of the laser stripe is extracted. Finally, a search method is carried out to examine the mid-point. A prototype sensor system has been developed and experimental results show its effectiveness and good performance.

Based the use of Generalized Stochastic Petri Nets (GSPN) a process model has been built up on the Flexible Manufacturing System (FMS). The modeling method has been analysed in this paper. Stochastic variables and the exponential distribution have been used to describe the uncertainty properties of the activities, behaviors and states. Also the state space of the research graph has been compressed based on Markov theory to get the steady probability. This model can be used to analyse the utility of the resource, the probability of a fault, the waiting time of a state and the throughout of the modeling in the FSM. This model also provides evidences to evaluate and optimize FMS. In the end, a laser blank welding system has been analysed and realized based on this model.