This paper focuses on finite-time partial topology identification of multi-group models, in which stochastic perturbation, multiple dispersal and time delay are included. Moreover, a novel finite-time partial topology identification strategy is proposed based on graph theory. Furthermore, the unknown partial topological structures are identified successfully through pinning control. Ultimately, a numerical example is provided to demonstrate the efficacy of the theoretical result.
This paper is focussed on the design issue of eventtriggered non-fragile state estimator towards interval type-2 fuzzy models confronted with bounded disturbances. The interval type- 2 Takagi-Sugeno fuzzy system model is employed to characterize the modeling uncertainties of complex nonlinear systems. A non-fragile estimator design scheme utilizing norm-bounded gain uncertainty is proposed to tackle the estimator gains changing. Meanwhile, the event-triggered correspondence mechanism is introduced for saving the limited network resources. A new fuzzy estimator synthesis design methodology is presented based on quadratic boundedness technology, the quadratic stability of the error dynamical system is guaranteed. The estimator gains are acquired in virtue of linear matrix inequality. An example is employed to verify the feasibility and applicability of the presented approach.
This paper solves the problem of distributed secure state estimation under homologous sensor attacks. Compared with the previous observer which use the data within time window, this paper proposes an improved Luenberger-like distributed observer which does use current measurement. It is also proved that the estimation error of the proposed observer converges to zero. This paper derives a necessary and sufficient condition to ensure that the estimation error is asymptotically convergent. One constructive condition is proposed to guide the design of the gain matrix. Finally, a simulation example verifies the effectiveness of the proposed observer when affected by homologous attack signal.
This paper investigates exponentially almost-sure stability (EAS stability) of dual switching linear continuous-time systems, in which only the sign of the subsystem matrices are known. This is a qualitative approach to stability analysis called sign-stability. The notion of sign-stability is natural extensions of standard EAS stability. It is called EAS sign-stability. It is remarkable the qualitative notion of stability is verified by a numerical example.
Parkinson’s disease (PD) is the second most prevalent degenerative neurological illness. Beta oscillations in electroencephalogram (EEG) are produced when a patient is suffering from PD. Deep brain stimulation (DBS) is one of the most successful treatments for PD, however, DBS control schemes still need improvement. In this paper, we test linear delay feedback control for DBS targeted the subthalamic nucleus (STN) or the internal globus pallidus (GPi) in a neural computing population model, describing dynamics of the cortex thalamocortical basal ganglia network. The numerical simulations show that both STNDBS and GPI-DBS can efficiently control beta oscillations of the PD state in term of frequency decrease.
To solve the problem that it is difficult to accurately identify bearing damage degree under strong noise, an Improved Deep Residual Network based on Normalization-based Attention Module (IDRNA) is proposed for bearing fault diagnosis. Firstly, the bearing vibration signals are preprocessed to obtain data samples, and then it is divided into training and testing subsets. After that a parallel 1x1 convolution branch is added for each convolution operation in the residual block, and Normalization based Attention Module is introduced at the end of the residual block to suppress some unimportant features to obtain the improved deep residual model. Use the training set to train the enhanced model. And the noisy signal is input to the trained model for the fault degree identification. Finally, to verify the feasibility of the model, we compare it with some other models and get good results.
As an effective means of unmanned strike or defense, missile has a higher requirement for security with its rapid development. However, the complex mechanism of missile systems determines that the accuracy of the model obtained by mechanism analysis cannot be guaranteed. While the system model obtained by algorithm based on test data has higher credibility. Aiming at the missile systems which cannot be modeled by mechanism analysis, this paper first obtains a linear model to approximate the actual missile system by the canonical variable analysis (CVA) that using the state data and input data generated in missile tests, and designs a fault diagnosis (FD) system based on the above linear model, which can predict the output of the missile system without fault. When the error between the real output and the predicted output of the missile exceeds a threshold, it can be considered that the missile has failures and corresponding measures should be taken.
The aim of this paper is to study the state estimation problem of stochastic time-varying Boolean networks (STVBNs) by using particle filtering algorithm. Firstly, the algebraic expressions of STVBNs are obtained by the semi-tensor product (STP) technique. By using Monte Carlo sampling method, an approximate expression of the posterior probability distribution is obtained. Secondly, a new method for calculating the state update formula of particles is proposed by combining the sampling process of particles and the state transition matrix of STVBNs. Thirdly, combined with the update formulas of weights and the observation matrices of STVBNs, the numerical method for the recursive formula of the weights update is presented. Based on the new update method of particles and weights, a Boolean particle filtering algorithm is designed to obtain an approximate posterior probability distribution and useful state estimation. Subsequently, the estimated state is given by using the criterion of minimum mean square error (MSE). Finally, to illustrate the feasibility of the adopted algorithm, a simple STVBN is adopted for simulation.
Piezo-driven micropositioning systems (PMSs) are prevalent in the high-precision manipulation fields, but also suffer from undesired nonlinear time-varying uncertainties, such as hysteresis, and parameter perturbation. To this end, this article proposes a novel sliding-mode-based robust control for a class of SISO nonlinear PMSs with time-varying uncertainties. Firstly, through a terminal-sliding-mode (TSM) surface with continuous, nonsingular, and finite-time convergence properties, a fast nonsingular TSM (FNTSM) control law is developed. Moreover, to avoid the shortage of the FNTSM control depending on the boundary information of system uncertainties, the Fourier series-based function estimation technique is adopted for dynamic approximation, and its approximation error is further compensated by the fuzzy logic system online. The updating laws of Fourier coefficients and fuzzy adjustable parameters are finally obtained via the Lyapunov stability theory. Numerical simulation results validate the reliability and superiority of the developed control strategy in comparison with the existing SMC methods.
In order to solve the problem of labor cost and production efficiency of a production line, a truss manipulator and a 3D camera are used to design a set of friction block workpiece identification, positioning and grabbing system. Use PLC, touch screen and servo system to build a four-axis truss manipulator; collect images through 3D cameras, and use machine vision software in the industrial computer for image preprocessing, template matching and coordinate calculation; industrial computer and PLC use Modbus-TCP communication mode, The robot coordinate signal and the handshake signal are transmitted and interacted, so as to realize the positioning and grasping of the same disordered workpiece. The test results show that the system can accurately locate the workpiece, the positioning accuracy can meet the grasping requirements of the truss manipulator, and the system can meet the actual production needs of the enterprise.
The visual perception of the original RatSLAM algorithm works directly on the intensities of image pixels and thus is greatly affected by illuminations, weathers, and other factors. The algorithm may result in perception errors and mapping failures in some situations. Therefore, this paper proposes an image sequence matching algorithm based on a visual dictionary model termed as Bag of Words (BoW) and the dynamic island mechanism to realize fast and accurate scene recognition. The feature vector of scene image is generated by the ORB algorithm, and then the image matching is transformed into the calculation of the distance between image feature vectors, improving the efficiency of image matching. The dynamic island mechanism is used as a detection algorithm to realize the sequence matching of the environment, improving the accuracy of loop closure detection. Experiments are performed in different environments, showing that the visual processing performance and mapping performance of the proposed algorithm are significantly better than the existing RatSLAM algorithm.
The blast furnace ironmaking is a “black box” operation and the blast furnace hearth plays a vital role in production. In this paper, the blast furnace hearth visualisation system is based on the actual blast furnace production data and the raw data processing is completed by feature engineering. It is worth noting that firstly, based on heat transfer and finite element method, BP neural network is used to predict and simulate the erosion state of the furnace hearth. Secondly, the temperature measurement points and temperature field derived parameters of the furnace chamber area are visualised, and the XGboost algorithm is used to achieve accurate prediction of key parameters; finally, the online operation of the blast furnace hearth visualisation system is realised based on the industrial internet platform, contributing to the intelligence of blast furnace ironmaking. Finally, the online operation of the blast furnace hearth visualisation system is realised based on the industrial internet platform, contributing to the intelligence of blast furnace ironmaking.
Linear segment features contains rich geometric information and it is crucial to perform accurate linear segment detection. We propose a temporal line segment detector based on subanchor point and line segment combination approach, which gives more accurate results without adjusting the parameters. The speed remains essentially the same as the fastest running line segment detectors (EDLines) existing and is 89’ faster compared to the conventional LSD line extraction algorithm. The proposed algorithm introduces bilateral filtering and line segment combination methods based on EDLines. The bilateral filter is used after the image input to remove the noise and retain the edge information of the image well. The short line segments that exist homologous are combined to generate clearer long line segments, which overcomes the problem of broken lines in LSD algorithm and excessive filtering of short line segments in EDLines algorithm. We validate the effectiveness of the proposed algorithm in experiments of line feature extraction for multiple sets of images in different scenes.
Universal image style transfer requires not only maintaining the semantic content but also transferring arbitrary visual styles. Recent progress has been made through processing an image as a whole, but without considering fine-grained styles of different semantic regions in the image. In this paper, we propose a Fine-Grained Style Transfer (FGST) model, which renders different content image regions into different fine-grained styles, thus improving the comprehensibility and visual effect of the stylized image. Specifically, we segment the input images into different semantic regions first, and then select the style and content image with the same semantic regions for training to preserve the fine-grained style consistency. In addition, we design a new style loss function to evaluate style consistency between the output stylized image and the input style image. Compared with the state-of-the-art models, experiments show that our model obtains better visual effects.
This paper studies the estimation issue for stochastic singularly perturbed complex networks (SPCNs) under a dynamic event-triggered mechanism (ETM). The SPCN is with a Markov chain whose transition probabilities are dependent on a stochastic variable that takes values with known sojourn probabilities. A new ETM is proposed to reduce the use of network resources. We design a state estimator which ensures the estimation error dynamics to be stochastically stable with $H_{\infty}$ performance. By matrix inequality technology, the desired parameters of state estimator are obtained. The effectiveness of the event-triggered estimation method is shown via a numerical example.
The article develops a reinforcement learning control scheme for a two-degree-of-freedom (2-DOF) helicopter system to achieve robust tracking. The control framework is divided into two parts: the critic neural network (NN) and the actor NN, which are designed to evaluate the control performance and estimate system uncertainties, respectively. Besides, the gradient descent method is exploited to update the weight of radial basis function NNs. Under the proposed control strategy, the rigorous stability of the closed-loop system is analyzed and demonstrated by Lyapunov’s stability theory. Finally, the Matlab simulation results are provided to verify the efficacy of the suggested scheme.
