Development of ensemble machine learning approaches for designing fiber-reinforced polymer composite strain prediction model

Engineering with Computers - Over the past few decades, it has been observed a remarkable progression in the development of computer aid models in the field of civil engineering. Machine learning...     

Flow and spawning habitat relationships for Dolly Varden: Understanding habitat–population dynamics in the Canadian Western Arctic

Northern form Dolly Varden (Salvelinus malma malma) have been designated as a species of Special Concern in Canada due to declines in population abundance and potential threats. Concern over detrimental effects of low flows on population abundance prompted research on how variability in discharge regimes influence habitat availability. Habitat suitability indices for prespawning and spawning adult anadromous Dolly Varden from two streams were integrated into a two‐dimensional hydrodynamic habitat model to assess the effect of flow variability on usable habitat. Regional hydrographs were used to identify an ecologically relevant range of flows that provided optimal spawning habitat for these populations and examine the relationship between abundance and discharge. Adults spawned in the tail end of pools at moderate water depths and water velocities, and used pebble‐ to cobble‐sized substrate for building redds; whereas, prespawning adults occupied deeper pools with moderate velocities and used cobble for cover. Model outputs showed that spawning habitat availability was optimized at flow rates between 1.6 and 3.0 m³/s and between 1.0 and 6.0 m³/s in Fish Hole Creek (FHC) and Little Fish Creek, respectively. A positive relationship between flows during the fall spawning period and abundance of the FHC population suggests that higher flows coinciding with optimal habitat availability may have contributed to positive recruitment. To strengthen and refine this habitat–population relationship for Dolly Varden in this area requires investigation of a broader suite of variables associated with environmental regimes and physical habitat in reaches used for spawning.     

Towards a unified visual framework in a binocular active robot vision system

This paper presents the results of an investigation and pilot study into an active binocular vision system that combines binocular vergence, object recognition and attention control in a unified framework. The prototype developed is capable of identifying, targeting, verging on and recognising objects in a cluttered scene without the need for calibration or other knowledge of the camera geometry. This is achieved by implementing all image analysis in a symbolic space without creating explicit pixel-space maps. The system structure is based on the ‘searchlight metaphor’ of biological systems. We present results of an investigation that yield a maximum vergence error of ～6.5 pixels, while ～85% of known objects were recognised in five different cluttered scenes. Finally a ‘stepping-stone’ visual search strategy was demonstrated, taking a total of 40 saccades to find two known objects in the workspace, neither of which appeared simultaneously within the field of view resulting from any individual saccade.     

Integrating IPsec within OpenFlow Architecture for Secure Group Communication

Network securityprotocolssuch as IPsechave been used for many years to ensure robust end-to-end communication and are impor-tant in the context of SDN.Despite the widespread installation of IPsec to date,per-packet protection offered by the protocol isnot very compatible with OpenFlow and flow-like behavior.OpenFlow architecture cannot aggregate IPsec-ESP flows in transportmode or tunnel mode because layer-3 information is encrypted and therefore unreadable.In this paper,we propose using the Secu-rity Parameter Index(SPI)of IPsec within the OpenFlow architecture to identify and direct IPsec flows.Thisenables IPsec to con-form to the packet-based behavior of OpenFlow architecture.In addition,by distinguishing between IPsec flows,the architectureis particularly suited to secure group communication.     

An Efficient,Scalable Key Transport Scheme (ESKTS) for Delay/Disruption Tolerant Networks

In the past, security protocols including key transport protocols are designed with the assumption that there are two parties communication with each other and an adversary tries to intercept this communication. In Delay/Disruption Tolerant Networking (DTN), packet delivery relies on intermediate parties in the communication path to store and forward the packets. DTN security architecture requires that integrity and authentication should be verified at intermediate nodes as well as at end nodes and confidentiality should be maintained for end communicating parties. This requires new security protocols and key management to be defined for DTN as traditional end-to-end security protocols will not work with DTN. To contribute towards solving this problem, we propose a novel Efficient and Scalable Key Transport Scheme (ESKTS) to transport the symmetric key generated at a DTN node to other communicating body securely using public key cryptography and proxy signatures. It is unique effort to design a key transport protocol in compliance with DTN architecture. ESKTS ensures that integrity and authentication is achieved at hop-by-hop level as well as end-to-end level. It also ensures end-to-end confidentiality and freshness for end communicating parties. This scheme provides a secure symmetric key transport mechanism based on public key cryptography to exploit the unique bundle buffering characteristics of DTN to reduce communication and computation cost .     

Feature extraction and ML techniques for static gesture recognition

The main objective of this study is to explore the utility of a neural network-based approach in hand gesture recognition. The proposed system presents two recognition algorithms to recognize a set of six specific static hand gestures, namely open, close, cut, paste, maximize, and minimize. The hand gesture image is passed through three stages: preprocessing, feature extraction, and classification. In the first method, the hand contour is used as a feature that treats scaling and translation of problems (in some cases). However, the complex moment algorithm is used to describe the hand gesture and to treat the rotation problem in addition to scaling and translation. The back-propagation learning algorithm is employed in the multilayer neural network classifier. The second method proposed in this article achieves better recognition rate than the first method.     

Single-stage three-phase boost power factor correction circuit for AC–DC converter

This article presents a single-stage three-phase power factor correction (PFC) circuit for AC-to-DC converter using a single-switch boost regulator, leading to improve the input power factor (PF), reducing the input current harmonics and decreasing the number of required active switches. A novel PFC control strategy which is characterised as a simple and low-cost control circuit was adopted, for achieving a good dynamic performance, unity input PF, and minimising the harmonic contents of the input current, at which it can be applied to low/medium power converters. A detailed analytical, simulation and experimental studies were therefore conducted. The effectiveness of the proposed controller algorithm is validated by the simulation results, which were carried out using MATLAB/SIMULINK environment. The proposed system is built and tested in the laboratory using DSP-DS1104 digital control board for an inductive load. The results revealed that the total harmonic distortion in the supply current was very low. Finally, a good agreement between simulation and experimental results was achieved.     

Common mode voltage reduction of quasi‐Z source indirect matrix converter

Quasi‐Z source indirect matrix converter (IMC) combines both advantages of conventional IMC and quasi‐Z source inverter, for example, no direct current (DC) link capacitor, compact all‐silicon power converter, bidirectional power flow, input power factor controllable, and high voltage gain; moreover, it does not require additional input filter, because continuous quasi‐Z source network integrates LC filter function. However, there is no literature to disclose common mode voltage (CMV) issue of quasi‐Z source IMC. In this paper, for the first time, the CMV issue and reduction of quasi‐Z source IMC are investigated. Firstly, the CMV of quasi‐Z source IMC is analyzed when using current typical modulation method, which follows the brief introduction of topology and modulation method for quasi‐Z source IMC, and the factors that affect the CMV are figured out; Secondly, referring to the CMV reduction methods of conventional IMC, two solutions named as Methods I and II to reduce the CMV for quasi‐Z source IMC are developed, which are achieved in the inverter stage; the third CMV reduction method is proposed in the rectifier stage through redefining the six sectors of the rectifier stage, which can implement zero current commutation. Experimental bench is built to test three approaches for reducing the CMV of quasi‐Z source IMC. Comparative evaluation is carried out between three methods and conventional modulation method. Experimental results verify that three methods can significantly reduce the CMV of quasi‐Z source IMC, with the CMV peak value reduction of 42%, but they present different features in terms of input and output current THDs, switching loss, CMV root mean square (RMS) value, modulation index limitation, and so on. Copyright © 2015 John Wiley & Sons, Ltd.     

Electrode activation sequencing employing conductivity changes in irreversible electroporation tissue ablation

Irreversible electroporation (IRE) uses high-voltage pulses applied to tissue, which cause dielectric breakdown of cell membranes resulting in cell death. IRE is a promising technique for ablation of nonresectable tumors because it can be configured to spare critical structures such as blood vessels. A consequence of pulse application is an increase in tissue electrical conductivity due to current pathways being opened in cell membranes. We propose a novel IRE method introducing electrode switching and pulse sequencing in which tissue conductivity is first increased using preparatory pulses in order to form high-conductivity zones, which then helps provide higher electric field intensity within the targeted tissue as subsequent pulses are applied, and hence, enhances the efficiency and selectivity of the IRE treatment. We demonstrate the potential of this method using computational models on simple geometries.