Statistically correlated multi-task learning for autonomous driving

Neural Computing and Applications - Autonomous driving research is an emerging area in the machine learning domain. Most existing methods perform single-task learning, while multi-task learning...     

Improved aquila optimization-based parameter evaluation framework and meta-heuristic algorithm in radio-over-fiber SpatialMux multi-input multi-output long-term evolution system

In heterogeneous access network, Multiple-Input Multiple-Output (MIMO) radio-over-fiber (RoF) system is an efficient approach for multiple signal transmission with low cost and complexity. The performance of RoF fronthaul system in MIMO system will be varied with different nonlinear effects. By adjusting various transmission parameters, such as the input signal power or the laser bias current, the nonlinear impacts produced by the RoF system can be reduced. In this paper, a novel algorithm Improved Aquila Optimization (IAO) is proposed to optimize transmission circumstances of MIMO RoF system. It determines the appropriate bias current for both lasers and Radio Frequency (RF) signal power in a short period. The input signals are wavelength multiplexed with Intensity Modulation and Direct Detection (IM/DD) applied. The carrier as well as transmission frequency is governed by the MIMO-Long-Term Evolution (LTE) standard. The proposed system is implemented in MATLAB, and the performance is evaluated. The experimental results show that fast convergence and trade-off between noise and nonlinearity are obtained with varying bandwidth. In the experimental scenario, the maximum Error Vector Magnitude (EVM) of 1.88, 3.14, and signal-to-noise ratio (SNR) of 3.204, and 2.698 was attained for both quadrature phase shift keying (QPSK) and quadrature amplitude modulation (QAM) modulation. [Correction added on 24 April 2023, after first online publication: the SNR values were corrected in the preceding sentence.] For 100 iterations, the processing time was reduced to 0.137 s. When compared with the conventional state-of-the-art approaches, the accuracy and computational complexity of the proposed approach are improved.     

Modified particle swarm optimization for a multimodal mixed-variable laser peening process

Optimization problems that result in shock, impact, and explosion type disciplines typically have mixed design variables, multiple optimal solutions, and high computational cost of an analysis. In the optimization literature, many researchers have solved problems involving mixed variables or multiple optima, but it is difficult to find multiple optima of a mixed-variable and high computation cost problem using an particle swarm optimization (PSO). To solve such problems, a mixed-variable niching PSO (MNPSO) is developed. The four modifications introduced to the PSO are: Latin Hypercube sampling-based particle generation, a mixed-variable handling technique, a niching technique, and surrogate model-based design space localization. The proposed method is demonstrated on the laser peening (LP) problem. The LP process induces favorable residual stress on the peened surface to improve the fatigue and fretting properties of the material. In many applications of LP, geometric configurations and dimensional integrity requirements of the component can constrain implementation of an optimal solution. In such cases, it is necessary to provide multiple alternatives to the designer so that a suitable one can be selected according to the requirements. It takes 24–72 CPU hours to perform an LP finite element analysis.     

Failure analysis of superheater tube supports of the primary reformer in a fertilizer factory

A failure analysis of superheater tube supports of the primary reformer in a local fertilizer factory is presented. A number of tube supports failed at approximately half of their designed service life. Following the failure, the factory was visited, and relevant information and samples were collected. The samples were investigated in the laboratory by chemical analysis, macro- and microhardness measurements, macro-and micrometallographic examinations, and X-ray diffractometry. The analysis showed the supports were fabricated from HH-type heat-resisting alloy and that the failure mode was high-temperature creep. The microstructure of the alloy showed the presence of massive intergranular as well as intragranular σ-phase and intragranular needle-shaped M₂₃C₆ carbides. It was also concluded from the formation of massive σ-phase in the tube that the failure was hastened because the supports were operating at approximately 800 °C. The alloy composition led to the formation of σ-phase under the operating conditions of the reformer, and the use of an alloy with a higher concentration of austenite-stabilizing element(s) could have avoided the failure.     

Optimal Automatic Generation Control of Asynchronous Power Systems Using Output Feedback Control Strategy with Dynamic Participation of Wind Turbines

Abstract—This article presents the design of optimal output feedback automatic generation control regulators for an interconnected power system with dynamic participation of doubly fed induction generator based wind turbines. The power systems consist of plants with hydro-thermal turbines and are interconnected via parallel AC/DC links. Efforts have been made to propose optimal automatic generation control regulators based on feedback of output state variables, which are easily accessible and available for the measurement. The designed optimal output feedback automatic generation control regulators are implemented, and the system dynamic responses for various system states are obtained considering 1% load perturbation in one of the areas. The dynamic performance is compared with that obtained with optimal automatic generation control regulators designed using full state vector feedback. The pattern of closed-loop eigenvalues is also determined to test the system stability.     

Experimental investigation of an indirect-type solar cooker for indoor cooking based on a parabolic dish collector

In the present research article, an indirect type solar cooking system has been developed for indoor cooking. In the proposed cooking system, a cooking pot has been placed at a distance of 5 m from the parabolic dish collector, and the heat has been transmitted from the collector to the cooking pot by means of heat transfer fluid. A gear pump of 40 W and insulated pipes have been used to circulate the fluid. A number of experiments have been performed to analyze the performance of the cooking system. During the investigation, the system achieved the temperature of the heat transfer fluid up to 175°C. The time taken for cooking the rice and the black grams has been observed 21 and 68min, respectively. The average thermal efficiency of the proposed system for the entire day has been achieved at 13.11%.     

Frequency regularization of a linked wind–diesel system using dual structure fuzzy with ultra-capacitor

Microgrids (μ-grids) are gaining increased interest around the world for supplying cheap and clean energy. In this paper, a μ-grid comprising a wind turbine generator (WTG) and diesel generator (DG) is considered. It is one of most practical and demanding systems suitable for the present energy crisis in isolated or rural areas. However, wind energy is intermittent in nature while load demand changes frequently. Therefore, a µ-grid can experience large frequency and power fluctuations. The speed governor of the DG tries to minimize the frequency and power deviations in µ-grid though its operation is slow and cannot adequately minimize system deviations. The paper proposes a novel arrangement based on a dual structured fuzzy (DSF) whose structure changes according to the switching limit with a reduced rule base. It has the capability to switch between proportional and integral actions and hence improves the frequency regularization of the μ-grid. The proposed strategy is tested in a μ-grid and the results considering step load alteration, load alteration at different instants of time, nonstop changing load request are compared with some of the well published methods to validate the effectiveness and simplicity of the present design. In addition, it shows that ultra-capacitor establishment in a μ-grid has a positive impact in minimizing system deviations with DSF for the studied cases.     

Growth and characterization of Ni:DLC composite films using pulsed laser deposition technique

The structure and surface morphology of Ni-incorporated diamond like carbon (Ni:DLC) films have been investigated. These films were deposited on Si substrates using pulsed laser deposition (PLD) technique. A KrF Excimer laser (λ = 248 nm) was used for co-ablation from multi component Ni–graphite target. The concentration of Ni was varied by ablating the Ni part of the target with various numbers of laser pulses. The SEM and AFM analysis reveals that the surface is composed of segregates of Ni which increases with the increase in Ni content during the growth process. The structural investigations by XRD and Raman spectroscopy provided information about the orientation of the incorporated constituent and the ordering of the carbon species. Maximum height of the nano structures which were observed on the surface was ∼50 nm. The G-peak of the graphite was shifted towards higher wave number due to enhancement in SP² sites which have been increased due to the increase in the Ni concentration. A small change in the surface roughness ranging from 7.78 nm to 13.1 nm due to increased Ni concentration was also observed.     

Design and development of nanoemulsion drug delivery system of amlodipine besilate for improvement of oral bioavailability

Nanoemulsion (NE) of amlodipine besilate (AB) was developed by spontaneous emulsification method with the aim to enhance the solubility and oral bioavailability of AB and to achieve localized delivery of drug at target site. Pseudoternary phase diagrams were constructed to identify the NE region. The selected formulations from NE region were subjected to droplet size analysis, partitioning study and in vitro drug release. The partition coefficient was calculated and correlated with percent dissolution efficiency as a tool to predict in vitro drug release from NEs. The release of drug from NEs was significantly higher (p < 0.01) than the marketed tablet formulation. The optimal formulation contained 15% Labrafil M, 35% [Tween 80: ethanol (2:1)], and 50% by weight aqueous phase (NE3) was characterized by transmission electron microscopy (TEM) and for thermodynamic stability. The pharmacokinetics and biodistribution studies of the optimized radiolabeled formulation (99mTc-labeled) in mice (p.o.) demonstrated a relative bioavailability of 475% against AB suspension. In almost all the tested organs, the uptake of AB from NE was significantly higher (p < 0.05) than AB suspension especially in heart with a drug targeting index of 44.1%, also confirming the efficacy of nanosized formulation at therapeutic site. A three times increase in the overall residence time of NE further signifies the advantage of NEs as drug carriers for enhancing bioavailability of AB.     

Accurate and energy-efficient boundary detection of continuous objects in duty-cycled wireless sensor networks

With the development and wide adoption of industrial wireless sensor networks to support various domain applications, the boundary detection of continuous objects has become an important research challenge, where improving the accuracy of boundary area while reducing the energy consumption are the first-class citizens to be considered. To address this research challenge, this article proposes a two-stage boundary face detection mechanism, where sensor nodes are duty-cycled and to be deployed in a dense fashion. When the occurrence of potential events are recognized using the initially activated sensor nodes, the boundary faces of continuous objects are constructed through adopting planarization algorithms. Thereafter, sensor nodes contained in certain boundary faces are examined, where their sensory data are estimated using spatial interpolation methods. Certain sensor nodes are selected to be woken up, only when their sensory data suggest that they should be more appropriate candidates of boundary sensor nodes. Consequently, the size of boundary faces is reduced, and this coarse-to-fine refinement procedure iterates, until all sensor nodes contained in the boundary faces have been examined. Experimental evaluation result shows that the boundary area can be refined significantly and be more precise, where the half of the initial boundary face area should be reduced in most situations.