Power Forecasting of Regional Wind Farms via Variational Auto-Encoder and Deep Hybrid Transfer Learning

Wind power is a sustainable green energy source. Power forecasting via deep learning is essential due to diverse wind behavior and uncertainty in geological and climatic conditions. However, the volatile, nonlinear and intermittent behavior of wind makes it difficult to design reliable forecasting models. This paper introduces a new approach using variational auto-encoding and hybrid transfer learning to forecast wind power for large-scale regional windfarms. Transfer learning is applied to windfarm data collections to boost model training. However, multiregional windfarms consist of different wind and weather conditions, which makes it difficult to apply transfer learning. Therefore, we propose a hybrid transfer learning method consisting of two feature spaces; the first was obtained from an already trained model, while the second, small feature set was obtained from a current windfarm for retraining. Finally, the hybrid transferred neural networks were fine-tuned for different windfarms to achieve precise power forecasting. A comparison with other state-of-the-art approaches revealed that the proposed method outperforms previous techniques, achieving a lower mean absolute error (MAE), i.e., between 0.010 to 0.044, and a lowest root mean square error (RMSE), i.e., between 0.085 to 0.159. The normalized MAE and RMSE was 0.020, and the accuracy losses were less than 5%. The overall performance showed that the proposed hybrid model offers maximum wind power forecasting accuracy with minimal error.

Experimental and Numerical Study on the Plume Rising in Stairwell of Buildings

In recent years, it has been observed that when a fire occurs in a multi-use facility, a toxic fire smoke rapidly rises through the vertical shaft and spreads due to the chimney effect and hot buoyancy. Generally, the fire smoke spreads rapidly through a number of evacuation passages installed for safe evacuation, which adversely affects an emergency situation. Due to the lack of this knowledge among the occupants, the majority of the occupants are evacuated using the stairwells, getting suffocated by poisonous smoke and suffering serious injuries. The present study considered the fire smoke spreading vertically through the stairwell. For this purpose, the power of the heat source and the area of the ventilation windows connected to the stairwell were modified, and the movement and diffusion of the hot plume rising vertically in the stairwell were observed. For the experiment, a 1/20 scaled-down stairwell model was employed, and the temperature ‘T’ and the vertical velocity ‘w’ of the hot plume rising inside the stairwell were measured using a 60 W-180 W heat source power. Numerical analysis was performed using FDS under similar conditions, and the results were compared with the experimental results.

Low-energy electron beam generation in inductively coupled plasma via a DC biased grid

Low-energy electron beam generation using a DC biased grid was investigated in an inductively coupled plasma (ICP). The electron beam was measured in argon gas at various pressures, ICP source powers, and substrate voltages (Vsub). At a low ICP source power (50 W), an electron beam was generated even at small values of Vsub (10 V), however at a high ICP source power (200 W), an electron beam was only generated when a higher voltage (30 V) was applied due to the short sheath thickness on the grid surface. The sheath on the grid surface is an important factor for generating electron beams because low-energy electrons are blocked. If the sheath thickness to small, a high voltage should be applied to generate an electron beam, as accelerate regions cannot exist without the sheath. At high pressure, since electrons experience numerous neutral collisions, a high substrate voltage is needed to generate an electron beam. However, if the applied substrate voltage becomes too high (40 V) at high pressure, high-energy electrons result in secondary plasma under the grid. Therefore, maintaining a low pressure and low ICP source power is important for generating electron beams.

Simulation and Implementation PID Controlling Buck Converter DC

Regulating the output voltage based on the desired set point is useful for many applications. However, getting the optimal value using fast computation with minimal error is still challenging. This paper aims to design, simulate, and implement a second-order Buck-Boost DC-DC converter circuit so that the voltage result according to the desired set point can be achieved. Initially, testing is conducted using Matlab Simulink. Then, Proteus is used to test the computation of the program on embedded systems in which the result is implemented in C. In low voltage power electronics applications, this approach has never been used to determine the output form. To determine the value of Kp, Ki, dan Kd, PID, Ziger Nichos (Guo, 2002). method is used. Meanwhile, tuning is done through Matlab. For simulation on Proteus, the output is tested by setting the setpoint values of 3.0, 2.5, and 1.7 volts. This aims to see the pattern of changes in the simulation. The simulation results with Proteus show that they have similar peak values but with different overshoot values. This is because the simulation must pass the reference voltage before it drops to the desired setpoint value. Proteus simulation can also help to prove embedded system programs are running correctly. On the other hand, the value of 1.7 volts is used as a setpoint in device implementation. This is due to the determination that the setpoint voltage in the implementation does not exceed the value of the source/power supply. The results show that for the rise time value of 378,770 ms, Overshoot and settling time are 11.798% and 0, respectively. This means the result produces an optimal value which is a return to the initial target. The optimal factor is assessed from the ability to minimize existing errors as well as having the shortest possible computational process.

Review of impedance source power converter for electrical applications

<p>Power electronic converters have been actively researched and developed over the past decades. There is a growing need for new solutions and topography to increase the reliability and efficiency of alternatives with lower cost, size and weight. Resistor source converter is one of the most important power electronic converters that can be used for AC-DC, AC-AC, DC-DC and DC-DC converters which can be used for various applications such as photovoltaic systems, wind power systems, electricity. Vehicles and fuel cell applications. This article provides a comprehensive overview of Z-source converters and their implementation with new configurations with advanced features, emerging control strategies and applications.</p>

Urban noise distributions and the influence of geometric spreading on skewness

Statistical distributions of urban noise levels are influenced by many complex phenomena, including spatial and temporal variations in the source level, multisource mixtures, propagation losses, and random fading from multipath reflections. This article provides a broad perspective on the varying impacts of these phenomena. Distributions incorporating random fading and averaging (e.g., gamma and noncentral Erlang) tend to be negatively skewed on logarithmic (decibel) axes but can be positively skewed if the fading process is strongly modulated by source power variations (e.g., compound gamma). In contrast, distributions incorporating randomly positioned sources and explicit geometric spreading [e.g., exponentially modified Gaussian (EMG)] tend to be positively skewed with exponential tails on logarithmic axes. To evaluate the suitability of the various distributions, one-third octave band sound-level data were measured at 37 locations in the North End of Boston, MA. Based on the Kullback-Leibler divergence as calculated across all of the locations and frequencies, the EMG provides the most consistently good agreement with the data, which were generally positively skewed. The compound gamma also fits the data well and even outperforms the EMG for the small minority of cases exhibiting negative skew. The lognormal provides a suitable fit in cases in which particular non-traffic noise sources dominate.

The performance of neutron diffractometers at long and short pulse spallation sources: Comparison between ESS and J-PARC

The general performance of diffractometers at the first long pulse spallation source ESS, is compared with their counterparts at J-PARC, a short pulse spallation source. The difference in the inherent pulse structure of these neutron sources presents opportunities for new concepts for instrumentation, where performance does not scale simply with source power. The article describes advantages and disadvantages of those diffractometers, adapting to the very different source characteristics. We find that the two sources offer comparable performance in flux and resolution when operating in high-resolution mode. ESS offers significant advantages in tunability and flexibility, notably in the ability to relax resolution in order to increase flux for a given experiment. The slow repetition rate of ESS favors long instruments. On the other hand, J-PARC instruments perform very well in spite of the lower source power and allow better access to epithermal neutrons, of particular interest for PDF analysis of diffraction data.