Search

Leader-following consensus is an essential problem for tracking control. However, the state trajectories of agents may be exposed because of frequent exchange of information. Hence, this paper considers the secure leader-following consensus with continuous-time dynamics over directed fixed topologies. For each follower, introduce a virtual node that forms a directed node pair combining with the real node, i.e., the information can only transmit from virtual node to real node in a node pair. In this case, we design a control law utilizing neighboring information, and analyze convergence and privacy preservation of the leader-following multi-agent systems. Finally, a numerical simulation is provided to illustrate our results.

In this article, we consider the event-triggered control of discrete-time nondeterministic linear systems, and instead of the continuous feedback strategies, we use the intermittent feedback strategies. Firstly, we find the optimal event generator and controller under the assumption that the closed-loop system is Hurwitz, hence, the state trajectory of the event-triggered system is close to that of the continuous feedback system with arbitrary accuracy. At the same time, a minimum number of states is sent to the controller to reduce the communication between the event generator and the controller. Furthermore, we show that the controllability of the system is sufficient to ensure that the error between the state trajectory of the event-triggered system and the state trajectory of the continuous feedback system can be arbitrarily small.

In this paper, we will briefly introduce the recent progress on machine learning based on the dynamical perspective. By combining machine learning with the dynamics, we get many continuous time algorithms. By discretizing the continuous time dynamic equations, we can not only review the existing algorithms, but also get new discrete-time algorithms. Firstly, gradient based algorithms is presented, which include the accelerated algorithm, policy gradient algorithm and stochastic gradient algorithm. Then, we come to the non-gradient algorithms, including the Pontryagins maximum principle based successive approximation algorithm and the optimal tensor algorithm. Next, continuous-time perspective of machine learning is reviewed, which to some extent establish a theoretical basis for machine learning. And in the end, we propose the advantage and the disadvantage of the above achievements, and try to give our own opinion about the application prospect and the development direction.

This paper is concerned with the open-loop time-consistent solution of constrained deterministic time-inconsistent linear-quadratic (LQ) optimal control problem. The linear constraint on control may destroy the structure of the conventional LQ formulation. We successfully transformed the equality constrained LQ problem into an unconstrained LQ problem by means of partition matrix. The necessary and sufficient conditions for the existence of an open-loop time-consistent solution are also studied in this paper. And then, we give an explicit expression of the solution of the constrained LQ problem.

In recent years, with the rapid development of unmanned surface vessel (USV), a real-time collision avoidance system is required to guarantee its safety navigation in the practical marine engineering. Therefore, this paper firstly uses wavelet transform technology to decompose and reconstruct the image collected by USV, so as to achieve the effect of denoising, and mark the foreground and background information. Then, the improved watershed segmentation method is used to segment the target object. Finally, a vector field histogram (VFH) method is proposed, which can solve the collision avoidance problem of USV from high dimension to low dimension based on the segmented image. The calculation burden is greatly low, and the real-time collision avoidance can be realized at the same time.

Driven by the advancement of science and technology, the issue of marine ship Internet of Things (IoT) users' assignment of computing tasks offloading has become more and more challenging. When faced with the complex and dynamic marine environment, considering the different quality requirements of maritime applications, we have addressed this issue in this paper. We first propose the space-air-ground-edge (SAGE) maritime communication network architecture. This novel architecture is used to offload computing-intensive applications and services for IoT users in the marine environment. At the same time, based on the Multi-Armed Bandits (MAB) problem, considering budget constraints and other marine environmental conditions such as delay and energy consumption, we propose an algorithm for selecting an edge server strategy. Simulation results of this paper show that the scheme has a better performance under different quality of service (QoS) requirements, which proves that the scheme is effective.

With the development of the ocean going intelligent surface vessel (ISV), a real-time collision avoidance system is required to guarantee the safety navigation in the practical marine engineering. In this paper, a unidimension histogram grid collision avoidance method for obstacle avoidance based on the binocular stereo vision technology is proposed. The advantages of the proposed method could map the ISV collision avoidance problem from high dimension space to low dimension space, and achieve the main collision avoidance space eigenvalues axis to collision avoidance solutions. Furthermore, the characteristics of low calculation consumption and not required accurate image depth information are also prominent. Finally the validity and superior collision avoidance performance of the proposed method is demonstrated through the smart ISV experiment.

A nonzero-sum game is considered in this paper with the cost functionals involving nonlinear terms of the conditional expectation. An open-loop equilibrium is introduced, which includes two controls of the players; one is a pre-committed control and the other one is a time-consistent equilibrium control.

Dimirovski, Georgi M. (Dogus Author) -- Conference full title: 37th Chinese Control Conference, CCC 2018; Wuhan; China; 25 July 2018 through 27 July 2018. This paper gives an introductory review of time-inconsistent stochastic optimal control. We mainly explain the reasons that result in the time inconsistency, and review the recent progresses in time-inconsistent stochastic linear-quadratic optimal control.

In this paper, the time-consistent solution of a time-inconsistent stochastic LQ problem is investigated. Necessary and sufficient condition is obtained to the existence of the open-loop time-consistent equilibrium control, which contain the solvability of certain forward-backward stochastic difference equation systems, the stationary conditions and the convexity conditions.

A discrete-time stochastic LQ problem with multiplicative noises and state transmission delay is studied in this paper, which does not require any definiteness constraint on the cost weighting matrices. Necessary and sufficient conditions are derived for the case with a fixed time-state initial pair. A set of coupled discrete-time Riccati-like equations can be derived to characterize the existence and the form of the delayed optimal control. Furthermore, the convexity of the cost functional is fully characterized via certain properties of the solution of the Riccati-like equations.

This paper studies mean field games for multiagent systems with multiplicative noises. By solving an auxiliary limiting optimal control problems subject to consistent mean field approximations, a set of decentralized strategies is obtained and further shown to be an asymptotical Nash equilibrium. Then, we apply the result to the consensus problem for the model of single integrators with multiplicative noises. It is shown that under the proposed strategies all the agents achieve asymptotic average consensus in the mean square sense.

This paper explores the impact of the output impedance on the power-sharing stability between parallelized microgrid inverters (MGIs) operating with virtual synchronous generator (VSG) method. In these systems, a virtual output impedance is usually added to the control loop of each MGI to improve the power sharing, regardless of line impedance unbalances. The time derivative of the MGI output current is usually used to implement the virtual impedance, which makes the system highly sensitive to the output current with high slew rate. To solve this, the virtual impedance is designed based on a cascaded second order general integrator (CSOGI) scheme, which is less sensitive to the output current noise, avoids to perform the time derivative function, achieves better output voltage total harmonic distortion. The proposed strategy is demonstrated in detail and validated with a two 100kVA-MGIs parallel system under linear and nonlinear loads.

This paper is concerned with a kind of time-consistent solution of the time-inconsistent mean-field stochastic linear-quadratic optimal control. Different from standard stochastic linear-quadratic problems, both the system matrices and the weighting matrices are dependent on initial time, and the conditional expectations of the control and state enter quadratically into the cost functional. Such features will ruin the Bellman optimality principle and result in the time-inconsistency of the optimal control. Due to the dynamical nature of control problems, a kind of open-loop time-consistent equilibrium control is thoroughly investigated in this paper. It is shown that the existence of an open-loop time-consistent equilibrium control for a fixed initial pair is equivalent to the solvability of a set of forward-backward stochastic difference equations with stationary conditions and convexity conditions. By decoupling the forward-backward stochastic difference equations, the existence of the open-loop equilibrium control for all the initial pairs is shown to be equivalent to the solvability of a set of coupled constraint generalized difference Riccati equations and a set of coupled constraint linear difference equations.

Due to the large adjust inertia of the diesel generator set (DGS), the quality of power supply would decline when a large load disturbance occurs to the diesel generator system. This paper aims to improve the stability of the diesel generator system voltage amplitude and frequency under impact load conditions using a virtual synchronous generator (VSG) method. Firstly, a mathematical model of DGS is built, the dynamic performances of DGS under a variety of load conditions is analyzed. Secondly, a three-phase energy storage inverter based on VSG control is selected as the parallel power compensator to suppress the system voltage amplitude and frequency fluctuations. Finally, experimental results are given to verify the feasibility and validity of the proposed control strategy.

This paper concentrates on the load disturbance in three phase microgrid inverter (MGI) with an isolated transformer based on virtual synchronous generator (VSG) control. The transformer can provide the VSG with inductive output impedance when placed behind the LC filter. But the load disturbance may lead to steady-state error and poor dynamic performance with only capacitor voltage feedback control. The capacitor voltage differential feedback control is proposed to damp the output LC filter resonance and thus increases the system stability margin. The output current differential feedforward control is proposed to eliminate the influence of load disturbance then obtains error-free steady-state output and excellent dynamic responses. These theoretical findings have been eventually verified experimentally both on simulation and laboratory prototypes.

A bending vibrations hybrid type ultrasonic motor with a bonded-type structure is proposed and designed in this work. Four pieces of PZT plates are bonded on a metal beam to form the proposed motor. In this design, elliptical movements of the driving feet, the significant motion to realize the actuating of a rotor, is produced by the superimposing of two first bending vibrations orthogonal in space. The structural parameters' sensitivities of the bending resonance frequency and the electromechanical coupling factor are calculated and discussed by modal analysis and harmonic analysis respectively. The most sensitive parameters are obtained. Then, the vibration characteristic of the proposed motor is investigated by transient analysis; another ultrasonic motor with the same length but no horns is also calculated to give a comparison. Different with previous motor, the proposed one drive the runner by two transversal displacements of the driving tip; this new actuating mode has merits of higher resonance frequency, larger vibration amplitude and velocity, better movement trajectory uniformity and shorter response time. The study results can guide the fabrication of linear or rotary ultrasonic motor with small size.

A novel longitudinal-bending (L-B) hybrid ultrasonic motor using vibration coupling is proposed. Different with the previous L-B ultrasonic motors, only longitudinal PZT plates are used; and the longitudinal-bending hybrid vibrations of the motor are excited synchronously. Fixed boundary condition is applied on one side surface of the flange, which locates at the wave node of a longitudinal transducer. The working principle of the motor is analyzed. By using finite element method, comparison between the proposed ultrasonic motor with fixed boundary condition and a transducer with free boundary condition are developed by modal analysis, harmonic analysis and transient analysis. It is found that the longitudinal-bending hybrid vibrations can be produced by the longitudinal piezoelectric elements under a fixed boundary condition of the flange. It should be noted that the bending vibration is not generated by the bending piezoelectric elements, but excited by a vibration coupling effect under an unsymmetrical boundary condition. The change of the boundary condition also shows minute effects on the resonance frequency and the electromechanical coupling factor. Oblique elliptical movements are formed on the two end tips of the transducer. These motions can push the runners linearly by frictional forces if preloads are applied.

A cross-shaped piezoelectric motor with length of 43mm and diameter of 11 mm was proposed in this paper. Two third bending modes were used to generate ellipse trajectory at the driving foot located at the vibrator's center. Sixteen pieces of piezoelectric ceramic plates with sizes of 5mm×3mm×1mm were used to excite two third bending modes. Sine and cosine voltages with effective value of 100V at the frequency of 54.696 kHz were applied to the piezoelectric ceramic plates in right and left sides of the vibrator. Symmetrical structure of the motor solved the problem of mode frequency degeneracy. Two excitation methods were put forward to excite two third bending vibrations. And effectiveness of the two excitation methods was compared by the transient analysis with finite element method. Transient analysis results showed that motion trajectories of nodes on the driving feet were three-dimensional and that ellipse motion in the XOY plane was fit to drive the motor, and the new excitation method improves the vibration amplitude in the OX and OY by 2 times and 4 times, respectively compared with the vibration amplitude excited by the traditional excitation method.

This paper presents a new idea for constructing ring-type linear ultrasonic motor. The stator of proposed motor contains two driving feet, sixteen pieces of piezoelectric ceramic plates and three rings-inner ring, middle ring and outer ring. The proposed ultrasonic motor forms elliptical movements on the driving feet by the superposition of stretching and bending vibration of the metal rings. This paper presents the working principle of proposed motor. By using FEM method, the proposed motor is designed and analyzed. By adjusting the structure parameters, the resonance frequencies of two working modes were tuned to be close. And by transient analysis, the elliptical motion trajectories of nodes on the driving feet are obtained and verify the theoretical feasibility.

This paper proposed the common mode model of three-level T-Type inverter. Then a kind of improved LCL filter, where the capacitor common point was linked to the DC neutral point, was used to eliminate the leakage current. However, this improved LCL filter changes the mechanism of the neutral-point potential. Traditional algorithms, such as the distribution factor algorithm, are no longer applicable. In order to control the neutral-point potential, a new general distribution factor algorithm was proposed in this paper. Experimental results demonstrate the availability and correctness of the theoretical analysis and algorithm.

The nuclear magnetic resonance (NMR) probe, one of the most important components in the spectrometer, has a considerable impact on overall instrument sensitivity. By the concept of a NMR analyzer with low-cost and portable was proposed in recent years, the earlier NMR probes were not suitable for the manufacture cost and cycle time. So, in this paper, a NMR probe based on the printed circuit board (PCB) planar microcoil was developed for time domain NMR measurement. When the Larmor frequency is 28.1 MHz, the NMR transverse relaxation time T2 of a deionized water sample was detected by the 2mm inner diameter PCB microcoil with six turns, and results of T2 measured value were stable under different scan times. In addition, relaxation times of used frying oil and three edible oils were analyzed by the PCB microcoil, both T22 and T23 components of used frying oil, obviously less than three edible oils, these results are agreed with the measurement of a commercial instrument Bruker minispec mq60.

Microgrid should be operated in both grid-connected and islanded mode to ensure high voltage quality and reliability. In the case of continuous uninterrupted power supply, seamless transfer is important between the two modes,and synchronization of the voltage of the point of common coupling (PCC) and utility grid should be finished first to achieve the goal. In this paper varies of synchronization strategies used in different microgrid control structures from islanded mode to grid-connected mode are summarized, and a new method based on droop control is proposed. This method does not affect power sharing during synchronization procedure. Simulation and experiment results are presented to show the feasibility of the method.

This paper proposes a new design method of LCL-filter for three-phase PWM voltage source inverter. Maximum converter-side current ripple is calculated by defining different modulation section of the adopted PWM strategy. In this way, the method can be easily extended to different converter topology and PWM strategy. Then the result is used to determine converter-side inductance. Once it is fixed, current attenuation passing from converter to grid at switching frequency and reactive power limitation equations are synthesized to design the minimum grid-side inductance and the proper range of resonance frequency. Then according to adopted PWM converter voltage spectrum, choose final resonance frequency to avoid the harmonic frequency spectrum distribution. A design example is given and the simulation and experiment results demonstrate the effectiveness of the method.

In this paper, a general active damping method based on capacitor voltage detection for stable control of LCL-filter grid-connect inverter is proposed. Three active damping methods are derived from the general form. Root locus and bode diagrams are used to analysis these three active damping methods. And in the case of different control delay, robustness of three active damping methods are compared. Integrated the advantages and disadvantages of each method, and eventually come to a relatively better active damping method. Finally, each active damping method is verified by simulation and experiment.

many software industrial association established in all provincial capital cities in China. Providing service for talents and company is their main responsibilities. The evaluation of software talents skills is multi-factor, multi-objective and multi-criteria decision-making problems. The current selection and evaluation method which the company used commonly is single, non-objective, non-universal and low input-output ratio. Base on a lot of investigation and research in typical software industry, the method of AHP (analytic hierarchy process) and FCEM (fuzzy comprehensive evaluation method) are used to evaluate the skills level of junior-level software talents. It makes evaluation results more objective and reasonable.

Case based reasoning(CBR) is very important task in data mining, but privacy information will be disclosed easily in CBR. This paper presents random locally linear embedding(LLE) on encrypted case based reasoning method. In order to be ensure the security of the CBR, the parameters nearest neighbor number k and embedded space dimension d of LLE algorithm are selected randomly. Further we embed the sensitive attribution into random dimension space using random LLE, thus the sensitive attributes are encrypted and protected. Because the transformed space dimension d and nearest neighbor number k are both random, this algorithm is very secure. In addition, LLE can keep the inherent shape of dataset, so the precision change of CBR after encryption can be controlled in a small scope. The experiment show that if we select appropriate parameters, then nearest neighbors of every case may be almost consistent. The present algorithm can guarantee that the security and the precision both achieve the requirements.

We studied the effects of parameters on the stability of passive dynamic walking by simulations and experiments. For simulations, we quantified the stability as the largest amount of lowing that the virtual prototype could still recover from and continue walking to the end. We used the ADAMS software to obtain this largest amount lowing. For experiments, we quantified the stability of each walker by observing 100 passive walking trials down a gentle slope of finite length and recording the fraction of trials which successfully walked to the end. This study provides a guide for designing passive dynamic walkers with high stability. This might also be used for designing prosthetic legs.

Root mean square error is presented as a quantitative measure for the similarity between human walking and passive dynamic walking. Normalized root mean square error of the hip angle and angular velocity between both walkers in a walking cycle is 6.9% and 12.2% on average, respectively. It shows a strong similarity between human walking and passive dynamic walking. A comparison is made between human walking and passive dynamic walking in four phases divided from a gait cycle. The reasons for resulting in the differences are studied.