Attention control with reinforcement learning for face recognition under partial occlusion

In this paper a new method for handling occlusion in face recognition is presented. In this method the faces are partitioned into blocks and a sequential recognition structure is developed. Then, a spatial attention control strategy over the blocks is learned using reinforcement learning. The outcome of this learning is a sorted list of blocks according to their average importance in the face recognition task. In the recall mode, the sorted blocks are employed sequentially until a confident decision is made. Obtained results of various experiments on the AR face database demonstrate the superior performance of proposed method as compared with that of the holistic approach in the recognition of occluded faces.     

Optimal motion planning for parallel robots via convex optimization and receding horizon

In this paper, the problem of motion planning for parallel robots in the presence of static and dynamic obstacles has been investigated. The proposed algorithm can be regarded as a synergy of convex optimization with discrete optimization and receding horizon. This algorithm has several advantages, including absence of trapping in local optimums and a high computational speed. This problem has been fully analyzed for two three-DOF parallel robots, ie 3s-RPR parallel mechanism and the so-called Tripteron, while the shortest path is selected as the objective function. It should be noted that the first case study is a parallel mechanism with complex singularity loci expression from a convex optimization problem standpoint, while the second case is a parallel manipulator for which each limb has two links, an issue which increases the complexity of the optimization problem. Since some of the constraints are non-convex, two approaches are introduced in order to convexify them: (1) A McCormick-based relaxation merged with a branch-and-prune algorithm to prevent it from becoming too loose and (2) a first-order approximation which linearizes the non-convex quadratic constraints. The computational time for the approaches presented in this paper is considerably low, which will pave the way for online applications.     

A Novel Model for the Time Dependent Competitive Vehicle Routing Problem: Modified Random Topology Particle Swarm Optimization

This paper presents a novel model for a time dependent vehicle routing problem when there is a competition between distribution companies for obtaining more sales. In a real-world situation many factors cause the time dependency of travel times, for example traffic condition on peak hours plays an essential role in outcomes of the planned schedule in urban areas. This problem is named as “Time dependent competitive vehicle routing problem” (TDVRPC) which a model is presented to satisfy the “non-passing” property. The main objectives are to minimize the travel cost and maximize the sale in order to serve customers before other rival distributors. To solve the problem, a Modified Random Topology Particle Swarm Optimization algorithm (RT-PSO) is proposed and the results are compared with branch and bound algorithm in small size problems. In large scales, comparison is done with original PSO. The results show the capability of the proposed RT-PSO method for handling this problem.     

Nano levels detection of beryllium by a novel beryllium PVC-based membrane sensor based on 2,3,5,6,8,9-hexahydro-1,4,7,10-benzotetra oxacyclododecine-12-carbaldehyde-12-(2,4-dinitrophenyl)hy

A novel poly(vinyl chloride)-based 2,3,5,6,8,9-hexahydro-1,4,7,10-benzotetra oxacyclododecine-12-carbaldehyde-12-(2,4-dinitrophenyl)hy (PBC) with sodium tetraphenyl borate (NaTPB) as an anion excluder, benzyl acetate (BA), acetophenon (AP) and o-nitrophenyloctyl ether (NPOE) as plasticizing solvent mediators was prepared and investigated as a beryllium selective sensor. The best performance was observed with the membrane having the PVC–NaTPB–NPOE–PBC composition 30%:3%:62%:5%, which worked well over a very wide concentration range (1.0×10⁻⁷ M to 1.0×10⁻¹ M). The sensor exhibits a Nernstian slope of 29.9 mV per decade of Be²⁺ activity. The detection limit of the sensor is 7.0×10⁻⁸ M (630 ppt). The proposed electrode shows excellent discriminating ability toward Be²⁺ ion with regard to alkali, alkaline earth, transition and heavy metal ions. It was successfully applied to the determination of beryllium in a mineral sample.     

Using correlation coefficients for locating partial discharge in power transformer

Most serious failure of power transformers is due to the insulation breakdown. Partial discharge (PD) that damages insulation by gradual erosion is major source of insulation failure. The effective ability of the wavelet packets analysis as a tool for disk-to-disk partial discharge faults detection and localization in transformer windings is shown in this paper. Techniques for locating a PD source are of the major importance in both the maintenance and repair of a transformer. One of the most well-known methods of PD localization in transformers is based on winding modeling and current of neutral point analysis. Since the impedance between PD location and neutral point of winding depends on the PD location in respect to neutral point, the frequency spectrum of neutral point current varies when the PD location changes. In the other word, the current components of neutral points vary according to the place where PD occurs. So in this paper, detailed model of transformer winding is modeled and the neutral point current is studied for locating PD. The used method is validated by the simulated model of transformer windings. This model produces a very acceptable current when compared to the experimental data. In this paper for locating partial discharge (PD) in transformer windings, a simulated model is developed for the transformer winding and the PD phenomenon mechanism. The impulse current test and wavelet packets transformation are used to locate PD. It is shown that the neutral current measurement of the transformer winding has useful information about PD location.     

New Approaches in Monitoring Multivariate Categorical Processes based on Contingency Tables in Phase II

In some statistical process control (SPC) applications, quality of a process or product is characterized by contingency table. Contingency tables describe the relation between two or more categorical quality characteristics. In this paper, two new control charts based on the WALD and Stuart score test statistics are designed for monitoring of contingency table‐based processes in Phase‐II. The performances of the proposed control charts are compared with the generalized linear test (GLT) control chart proposed in the literature. The results show the better performance of the proposed control charts under small and moderate shifts. Moreover, new schemes are proposed to diagnose which cell corresponding to different levels of categorical variables is responsible for out‐of‐control signal. In addition, we propose EWMA–WALD and EWMA–Stuart score test control charts to improve the performance of Shewhart‐based control charts in detecting small and moderate shifts in contingency table parameters. Meanwhile, we compare the performances of two proposed EWMA‐based control charts with the ones of three existing control charts called EWMA–GLT, EWMA–GLRT and an EWMA‐type control chart for multivariate binomial/multinomial processes along with the ones of the corresponding Shewhart‐based control charts. A numerical example is given to show the efficiency of the proposed methods. Finally, the effect of parameter estimation in Phase I based on m historical contingency table on the performance of the Shewhart‐based control charts is studied. Copyright © 2016 John Wiley & Sons, Ltd.     

Experimental investigation of the flow-induced motion of a square-section cylinder

The flow-induced motion(FIM)of an elastically mounted square-section cylinder is experimentally investigated over a wide range of Reynolds numbers(1.5×104＜Re＜7....     

Experimental studies on the catalytic behavior of alloy and core-shell supported Co-Ni bimetallic nano-catalysts for hydrogen generation by hydrolysis of sodium borohydride

Monometallic (Co) and bimetallic (Co-Ni and Co-Cu) oxides catalysts supported on the almond based activated carbon (AC) were prepared by the heterogeneous deposition-precipitation method. The activity of these catalysts was evaluated as a function of reaction temperature, NaOH, and NaBH₄ concentration. Several analysis methods including XRD, XPS, FTIR, TEM, FESEM, ICP-OES, and BET were applied to characterize the structure of prepared samples. Well-dispersed supported bimetallic nano-catalysts with the size of particles below 20 nm were formed by using nickel and copper oxides as a promoter which was confirmed by XRD and TEM techniques. Surface composition of alloy and core-shell cobalt-nickel oxides catalysts was analyzed by ICP-OES which was in a good agreement with nominal content during catalyst preparation. The performance of bimetallic cobalt-nickel oxides catalysts indicated the synergic effect between cobalt and nickel in comparison with monometallic and bimetallic cobalt-copper samples for hydrogen production. Maximum hydrogen generation rate was measured for the supported core-shell catalyst as named Ni1/Co3/AC. The reaction rate increased with increasing the temperature of the alkaline solution as a significant parameter while other operating conditions were kept constant. The optimal values for NaOH and NaBH₄ content were calculated to be 10 wt % for both variables at 30 °C. Hydrogen production rates were calculated to be 252.0, 310.8 and 658.8 mL min^?1.g^?1 by applying Co3/Ni1/AC, Co3-Ni1/AC (alloy) and Ni1/Co3/AC at 30 °C in 5 wt % NaBH₄ and 5 wt % NaOH solutions, respectively. Obtained activation energy (50 kJ mol^?1) illustrated that the suitable catalysts were synthesized for hydrogen generation. The experimental study showed that the hydrolysis of NaBH₄ was a zero-order type reaction with the respect to the sodium borohydride concentration. A semi empirical kinetic model was derived at the various temperatures and NaOH concentrations.     

Taguchi Optimization of Solvent-Antisolvent Crystallization to Prepare Ammonium Perchlorate Particles

The sphericity and size of ammonium perchlorate (AP) particles significantly influence the properties of composite propellants. As the AP particles become more spherical, the accumulation coefficient increases, the viscosity during casting decreases, and the particle loading and burning rate increase. Hence, the production of micronized AP particles with an average size between 1 and 20 μm is important to increase the loading percentage of AP in the composite propellant. Here, the Taguchi experimental design was used to optimize the solvent-antisolvent crystallization (SAC) process for the preparation of micronized AP particles with higher sphericity. SAC parameters such as the type of antisolvent, the solvent-to-antisolvent ratio, the antisolvent temperature, the stirring speed, and the retention time were investigated at four levels. The type of antisolvent and the solvent-to-antisolvent ratio were found to mainly contribute to improving the sphericity and size of the AP particles, respectively.