Features play an important role in the performance of machine learning and classification applications. Usually, separability of classes by using raw or original features are so low, and it is necessary to use complex...
详细信息
Features play an important role in the performance of machine learning and classification applications. Usually, separability of classes by using raw or original features are so low, and it is necessary to use complex classifiers with high computational costs or use enrichment modules to increase distinctiveness of features. In this paper, a deep feature enrichment method is proposed to increase the distinguishing power of features using an adaptive neural network-based structure. Proposed method adaptively uses linear/non-linear activation functions for coding, and the dimension of the coding space adaptively adjusted to be lower, the same, or higher than the original feature space. Then the best neural network structure (number of layers and neurons per layers) and the optimum weights for the proposed neural structure are optimized using an evolutionary optimization algorithm. Optimized modules can map/code raw input features into an enriched feature space that can increase the separability of the data points among classes. In fact, our obtained enriched features can adapt themselves to the nature of the training data and they can improve the generalization power also the performance of conventional classifiers. Experimental results on popular UCI datasets such as Glass, Liver, Iris, Wine, Breast cancer and seeds show increase of significant correct recognition rates (11.63% for Glass, 4.35% for Liver, 13.34% for Iris, 27.78% for Wine, 0.72% for Breast cancer and 11.9% for seeds) and also improvement of more than 1.5% of verification rate and 2% of Identification rate for the Yale face database. (C) 2020 Elsevier Ltd. All rights reserved.
Currently, when calculating the magnetic field generated by the solenoid coil of the superconducting wire wound, we assume that the coil cross section with a uniform current density, but actual current in superconduct...
详细信息
Currently, when calculating the magnetic field generated by the solenoid coil of the superconducting wire wound, we assume that the coil cross section with a uniform current density, but actual current in superconducting wires (NbTi) in the form of a wire in channel is not evenly distributed, the current distribution only in the superconducting core, i.e., there is no current in copper, insulation, and filler, and this method of calculation will result in errors. In this paper, we model the superconducting cores of the 1.5-T superconducting magnetic resonance imaging (MRI) magnet to calculate accurate magnetic field intensity and inhomogeneity by helicoidal method in the diameter of spherical volume and find that inhomogeneity is eight times bigger than that calculated by spherical harmonic expansions, which cannot be accepted in design. Hence, in order to design a high-homogeneity MRI magnet, we amend the 1.5-T MRI magnet's original parameters by an optimization algorithm through an original interface between OPERA-3D and MATLAB according to the accurate results.
Glistenings are liquid-filled microvacuoles in intraocular lenses (IOLs) appear when the IOL is in an aquatic environment that affect the quality of vision. In our glistenings Detection method, the candidate glistenin...
详细信息
Glistenings are liquid-filled microvacuoles in intraocular lenses (IOLs) appear when the IOL is in an aquatic environment that affect the quality of vision. In our glistenings Detection method, the candidate glistenings are automatically detected by mathematic morphology methodology. Machine learning approaches, feature selection and classification are used in this paper. The 68 features are extracted and used as training data for fine segmented using the classifiers. The detected glistenings are validated by object-based with ophthalmologist's hand-drawn ground-truth. Our proposed method, Feature Selection using Fuzzy-based Firefly algorithm (FS-FFA) applied the concept of fuzzy entropy to calculating the membership of features data for in order to select good sets of the relevant features that maximize the classification performance in glistenings Detection. The proposed FS-FFA is compared with feature selection methods the standard firefly algorithm (FS-FA) and without feature selection using basic classifier k-nearest neighbor. The results have shown that the Matthews correlation coefficient (MCC) and the diagnostic odds ratio (DOR) value increase after feature selection using firefly algorithm and fuzzy entropy. Small size of features set also decreased that classification time in testing phase.
This study presents a novel data-driven approach designed to address the intricate link between climate change and agriculture, focusing on rice farming in Southeast Asia. By using advanced modelling and optimization ...
详细信息
This study presents a novel data-driven approach designed to address the intricate link between climate change and agriculture, focusing on rice farming in Southeast Asia. By using advanced modelling and optimization tools, namely ARX models and Model Predictive Control, it aims to control the temperature anomaly across fifteen world’s subregions. Using ARX models to downscale the global temperature anomaly, the approach allows the evaluation of local climate effects. The methodology is applied to evaluate the impact of climate change on rice production in Southeast Asia, projecting potential outcomes under different emission scenarios. By optimizing greenhouse gas emissions, particularly carbon dioxide and methane, the goal is to keep the temperature anomaly below critical thresholds, ensuring resilient rice production, supporting food security, and minimizing economic and social costs.
The increasing need for reducing the costs and the environmental impact of the energy supply renewed the interest on distributed generation, also favoured by the recent EU directives. Anyway, the solely installation o...
详细信息
The increasing need for reducing the costs and the environmental impact of the energy supply renewed the interest on distributed generation, also favoured by the recent EU directives. Anyway, the solely installation of efficient small- and medium-size pieces of equipment is not sufficient to achieve the expected targets, being their proper scheduling and management, of course based on the fluctuations of both the loads pattern and the energy prices, the fundamental issue determining their effectiveness. In recent times, several techniques have been proposed with the purpose of optimizing the operation of the installed generators on the basis of load predictions; even if these latter ones heavily affect the performance of the plant management, especially when considering a single prosumer whose behaviour is scarcely predictable with a good accuracy, often this aspect is neglected. The present study aims to analyse how inaccurate load predictions affect the performance of an energy plant whose generators are scheduled by an optimization tool working considering a time span of one day. Different “structures” of error will be modelled and analysed, taking as benchmark load profiles the acquired data for different periods of the year from an office building plant, equipped with a PV plant, two micro-CHP system combined with an absorption chiller, an electric chiller, a gas boiler and a reversible electric heat pump, with thermal storages. The focus will be on the comparison of both the economic and CO 2 emission impact, by considering on one side the loads prediction as perfect and on the other side with different entity of errors, with the target to stress the importance of the correct loads prediction issue.
Our paper presents chosen computational algorithms for solution of finite element models with structural uncertainties. An application of the chosen approaches will be presented–the first one, a simple combination of...
详细信息
Our paper presents chosen computational algorithms for solution of finite element models with structural uncertainties. An application of the chosen approaches will be presented–the first one, a simple combination of only inf-values or only sup-values; the second one presents full combination of all inf-sup values; the third one uses the optimizing process as a tool for finding out an inf-sup solution and last one is the Monte Carlo method as a comparison tool.
Wind power generation has strong volatility. Accurate wind speed forecasting can not only avoid the waste of power resources, but also facilitate the development of clean energy and promote the energy transition world...
详细信息
Wind power generation has strong volatility. Accurate wind speed forecasting can not only avoid the waste of power resources, but also facilitate the development of clean energy and promote the energy transition worldwide. However, previous research has predominantly focused on the accuracy of wind power prediction, while ignoring the reliability of wind speed prediction system. In this research, a hybrid forecasting system with both accuracy and reliability of wind power forecasting is proposed. Firstly, a hybrid adaptive decomposition denoising algorithm is proposed to solve the unreasonable decomposition and residual noise. To improve the search performance, the seagull algorithm is optimized by chaotic system and Cauchy operator, and then the parameters of long short-term memory model are adjusted. Finally, based on data enhancement theory, an interval prediction model combined with kernel density estimation is proposed. The model is verified by the historical data of Sotavento wind farm in Spain and Eman wind farm in China. The average absolute percentage error values of wind speed point prediction are 2.87% and 8.01%, respectively. At the same confidence level, the interval prediction model proposed has narrower widths compared to the comparative model, with higher average interval scores. The results indicate that the point prediction model proposed in this research exhibits higher accuracy, while the interval prediction model demonstrates greater stability and reliability. These findings provide technical support for wind power forecasting.
Blockchain, a popular technology, remains the decentralized data management framework approved for use by many industries. The application to the insurance industry needs to offer mobility using the wireless network. ...
详细信息
Blockchain, a popular technology, remains the decentralized data management framework approved for use by many industries. The application to the insurance industry needs to offer mobility using the wireless network. The wireless network has many limitations to overcome. This paper focuses on such problems and introduces three levels of a solution to the problem. The first level is resolved using the edge computers as storage at the agencies and the partners. The second level of economic operation is solved by introducing a D2D network solution. The third level of high transactions over the network is considered using a two-stage optimization method. The introduced optimization algorithms are simulated, and results are compared with a classical step-by-step calculation method that is not feasible under real-time application. The optimization methods successfully determine the maximum channel rate with the interferences influencing the operation of such a system.
Modern (micro) grids host inverter-based generation units for utilizing renewable and sustainable energy resources. Due to the lack of physical inertia and, thus, the low inertia level of inverter-interfaced energy re...
详细信息
Modern (micro) grids host inverter-based generation units for utilizing renewable and sustainable energy resources. Due to the lack of physical inertia and, thus, the low inertia level of inverter-interfaced energy resources, the frequency dynamic is adversely affected, which critically impacts the stability of autonomous microgrids. The idea of virtual inertia control (VIC), assisted by battery energy storage systems (BESSs), has been presented to improve the frequency dynamic in islanded microgrids. This study presents the PD-FOPID cascaded controller for the BESS, a unique method for enhancing the performance of VIC in islanded microgrids. Using the firefly algorithm (FA), the settings of this controller are optimally tuned. This approach is robust to disruptions due to uncertainties in islanded microgrids. In several scenarios, the performance of the suggested approach is compared with those of other control techniques, such as VIC based on an MPC controller, VIC based on a robust H-infinite controller, adaptive VIC, and VIC based on an optimized PI controller. The simulation results in MATLAB show that the suggested methodology in the area of VIC is better than previous methods.
Large-scale grid integration of variable renewable energy is crucial for achieving decarbonized development. However, this integration requires frequent regulation of flexible power sources for complementary operation...
详细信息
Large-scale grid integration of variable renewable energy is crucial for achieving decarbonized development. However, this integration requires frequent regulation of flexible power sources for complementary operation, which can lead to wear-and-tear and fatigue damage to key components. This poses potential risks to flexible power sources. Existing studies have primarily focused on limiting unit startups, while have neglected the risk of frequent power regulation. Thus, this work proposes a risk-averse short-term scheduling method for a Wind Solar-Cascade hydro-Thermal-Pumped storage hybrid energy system to balance frequent regulation risk, cost, and carbon emission: (1) a risk-averse short-term scheduling model is proposed, considering multilayer constraints;(2) a multi-objective hybrid African vulture optimization algorithm is proposed to effectively solve the scheduling problem including continuous and discrete variables. A case study in the Songhua River basin, China shows that: (1) compared with traditional models, the proposed model reduces the risk by 31.4% and enhances the comprehensive performance in balancing the three objectives by 22.4%;(2) the proposed algorithm performs robustness and search capability advantages, with improvements of 33.01% and 21.44% respectively, in solving the problem of challenging constraints and mixed decision variables. Overall, this work contributes to enhancing the management of large hybrid energy systems.
暂无评论