Real‐time vehicle detection and counting of multiple types is a difficult problem. To solve this problem, study presents an efficient method based on single shot (SSD) to construct system. The proposed named Fast‐SSD first combines the Slim ResNet‐34 with Single Shot MultiBox Detector. Then authors limit location prediction at each cell in feature map modify network. When input size picture 300 × 300, achieves accuracy 76.7 mAP PASCAL visual object classes 2007 test set. network can be...

False data injection (FDI) attack is the most common integrity attack, and it also one of serious threats in industrial control systems (ICSs). Although many detection approaches are developed with burgeoning research interests, technical capability existing methods still insufficient because stealth FDI attacks have been proven to bypass bad detector. In this paper, a novel analytical algorithm proposed identify ICSs according correlation analysis. First, we evaluate between measurements...

Wireless sensor networks are self-organized and data-centric. Sensor nodes typically deployed in unattended environment, therefore the security is a critical issue that must be resolved many applications. In order to prevent sensitive data from eavesdropping tampering, this paper proposes secure transmission scheme based on digital watermarking technique. The node calculates hash value of two time-adjacent using one-way function, then takes as watermark information embed into part sequence....

For the fault detection and diagnosis problem in large-scale industrial systems, there are two important issues: missing data samples non-Gaussian property of data. How-ever, most existing data-driven methods cannot be able to handle both them. Thus, a new Bayesian network classifier based method is proposed. At first, non-imputation presented incomplete samples, with proposed classifier, values can marginalized an elegant manner. Furthermore, Gaussian mixture model used approximate linear...

Intrusion detection is an important technique in the defense-in-depth network security framework. The IDS continuously watch activity on a or computer, looking for attack and intrusion evidences. However, host-based detectors are particularly vulnerable, as they can be disabled tampered by successful intruders. In this paper, hidden semi-Markov models method predicting anomaly events intentions of possible intruders to computer system developed based observation call sequences. BSM audit...

A recursive probabilistic principal component analysis (PPCA) based data-driven fault identification method is proposed to handle the missing data samples and mode transition in multi-mode process. This model recursively obtained by using increasing number of normal observations with partly data. First, on singular value historic matrix, whole process divided into different steady modes transitions. For modes, conventional PPCA used obtain components, impute When a transition, applied, which...

For the cases that data samples are partially missing in control systems, analysis given to determine type of mechanisms, then a dynamic Bayesian network approach is used model general fault prognosis problem after we proposed method deal with real-time nonlinear systems data. Our demonstrated on benchmark continuous stirred tank reactor (CSTR) problem, which show process constructing and use for simulation prognosis. Results though noisy missing, combined effective treatment data, networks...

As a representative complex system, the aircraft modeled very difficultly and imprecisely. This makes model-based fault detection methods degenerated. In this dissertation, nonlinear time series, which is constructed by output variables of aircraft, converted into discrete dynamic then novel series prediction method achieved adaptive observation system states. An online RBFNN used to fit nonlinearity compensate unknown disturbance. Thereby one-step-ahead proposed. By using probability...

Based on a dynamic Bayesian network with an incomplete time slice and mixture of the Gaussian outputs, data-driven fault prognosis method for model-unknown processes is proposed in this article. First, according to requirement prognosis, unknown future observed node constructed. Moreover, states are described by current measurements his historic data form conditional probability. Second, completed part historical data, parameter-learning algorithm used obtain parameters weight coefficients...

It is well known that the dynamics of chaotic system very sensitive to initial conditions state, and synchronization two identical systems only obtained, in general, with high gain control law once their are a certain large deviation. Furthermore, commonly unknown practice, which causes difficulty synchronizing systems. This paper deals unified input constraint. First, scalar sign function utilized approximate constrained non-smooth so continuous smooth nonlinear an approximated synchronized...

Recently probabilistic principal component analysis (PPCA) has been used for process monitoring and fault diagnosis, which can model the noise handle problem of missing data in framework. Nevertheless, samples are treated as components conventional PPCA method, causes estimation accuracy is largely influenced by rate. In this paper, a detection identification method based on improved proposed industry whose variables with large First, decreasing errors data, an parameters studied. By...