Zhejiang University Press

Soft sensors are widely used to predict quality variables which are usually hard to measure. It is necessary to construct an adaptive model to cope with process non-stationaries. In this study, a novel quality-related locally weighted soft sensing method is designed for non-stationary processes based on a Bayesian network with latent variables. Specifically, a supervised Bayesian network is proposed where quality-oriented latent variables are extracted and further applied to a double-layer similarity measurement algorithm. The proposed soft sensing method tries to find a general approach for non-stationary processes via quality-related information where the concepts of local similarities and window confidence are explained in detail. The performance of the developed method is demonstrated by application to a numerical example and a debutanizer column. It is shown that the proposed method outperforms competitive methods in terms of the accuracy of predicting key quality variables.

We investigate the solution and stability of continuous-time cross-dimensional linear systems (CCDLSs) with dimension bounded by V-addition and V-product. Using the integral iteration method, the solution to CCDLSs can be obtained. Based on the new algebraic expression of the solution and the Jordan decomposition method of matrix, a necessary and sufficient condition is derived for judging whether a CCDLS is asymptotically stable with a given initial state. This condition demonstrates a method for finding the domain of attraction and its relationships. Then, all the initial states that can be stabilized are studied, and a method for designing the corresponding controller is proposed. Two examples are presented to illustrate the validity of the theoretical results.

The input/output (I/O) pins of an industry-level fluorescent optical fiber temperature sensor readout circuit need on-chip integrated high-performance electro-static discharge (ESD) protection devices. It is difficult for the failure level of basic N-type buried layer gate-controlled silicon controlled rectifier (NBL-GCSCR) manufactured by the 0.18 µm standard bipolar-CMOS-DMOS (BCD) process to meet this need. Therefore, we propose an on-chip integrated novel deep N-well gate-controlled SCR (DNW-GCSCR) with a high failure level to effectively solve the problems based on the same semiconductor process. Technology computer-aided design (TCAD) simulation is used to analyze the device characteristics. SCRs are tested by transmission line pulses (TLP) to obtain accurate ESD parameters. The holding voltage (24.03 V) of NBL-GCSCR with the longitudinal bipolar junction transistor (BJT) path is significantly higher than the holding voltage (5.15 V) of DNW-GCSCR with the lateral SCR path of the same size. However, the failure current of the NBL-GCSCR device is 1.71 A, and the failure current of the DNW-GCSCR device is 20.99 A. When the gate size of DNW-GCSCR is increased from 2 µm to 6 µm, the holding voltage is increased from 3.50 V to 8.38 V. The optimized DNW-GCSCR (6 µm) can be stably applied on target readout circuits for on-chip electrostatic discharge protection.

In this paper we address the problem of interpolating a spline developable patch bounded by a given spline curve and the first and the last rulings of the developable surface. To complete the boundary of the patch, a second spline curve is to be given. Up to now this interpolation problem could be solved, but without the possibility of choosing both endpoints for the rulings. We circumvent such difficulty by resorting to degree elevation of the developable surface. This is useful for solving not only this problem, but also other problems dealing with triangular developable patches.

Current methods for radar target detection usually work on the basis of high signal-to-clutter ratios. In this paper we propose a novel convolutional neural network based dual-activated clutter suppression algorithm, to solve the problem caused by low signal-to-clutter ratios in actual situations on the sea surface. Dual activation has two steps. First, we multiply the activated weights of the last dense layer with the activated feature maps from the upsample layer. Through this, we can obtain the class activation maps (CAMs), which correspond to the positive region of the sea clutter. Second, we obtain the suppression coefficients by mapping the CAM inversely to the sea clutter spectrum. Then, we obtain the activated range-Doppler maps by multiplying the coefficients with the raw range-Doppler maps. In addition, we propose a sampling-based data augmentation method and an effective multiclass coding method to improve the prediction accuracy. Measurement on real datasets verified the effectiveness of the proposed method.

Although the perimeter security model works well enough when all internal hosts are credible, it is becoming increasingly difficult to enforce as companies adopt mobile and cloud technologies, i.e., the rise of bring your own device (BYOD). It is observed that advanced targeted cyber-attacks usually follow a cyber kill chain; for instance, advanced targeted attacks often rely on network scanning techniques to gather information about potential targets. In response to this attack method, we propose a novel approach, i.e., an “isolating and dynamic” cyber defense, which cuts these potential chains to reduce the cumulative availability of the gathered information. First, we build a zero-trust network environment through network isolation, and then multiple network properties are maneuvered so that the host characteristics and locations needed to identify vulnerabilities cannot be located. Second, we propose a software-defined proactive cyber defense solution (SPD) for enterprise networks and design a general framework to strategically maneuver the IP address, network port, domain name, and path, while limiting the performance impact on the benign network user. Third, we implement our SPD proof-of-concept system over a software-defined network controller (OpenDaylight). Finally, we build an experimental platform to verify the system’s ability to prevent scanning, eavesdropping, and denial-of-service attacks. The results suggest that our system can significantly reduce the availability of network reconnaissance scan information, block network eavesdropping, and sharply increase the cost of cyber-attacks.

Spectral clustering is one of the most popular and important clustering methods in pattern recognition, machine learning, and data mining. However, its high computational complexity limits it in applications involving truly large-scale datasets. For a clustering problem with n samples, it needs to compute the eigenvectors of the graph Laplacian with O(n3) time complexity. To address this problem, we propose a novel method called anchor-based spectral clustering (ASC) by employing anchor points of data. Specifically, m (m ≪ n) anchor points are selected from the dataset, which can basically maintain the intrinsic (manifold) structure of the original data. Then a mapping matrix between the original data and the anchors is constructed. More importantly, it is proved that this data-anchor mapping matrix essentially preserves the clustering structure of the data. Based on this mapping matrix, it is easy to approximate the spectral embedding of the original data. The proposed method scales linearly relative to the size of the data but with low degradation of the clustering performance. The proposed method, ASC, is compared to the classical spectral clustering and two state-of-the-art accelerating methods, i.e., power iteration clustering and landmark-based spectral clustering, on 10 real-world applications under three evaluation metrics. Experimental results show that ASC is consistently faster than the classical spectral clustering with comparable clustering performance, and at least comparable with or better than the state-of-the-art methods on both effectiveness and efficiency.