The transformers in low-voltage distribution areas are easily influenced by environmental aging and nonlinear load in long-term operation, and the traditional on-site off-line calibration method has a high cost and low efficiency. Aiming at these problems, a model of online non-inductive calibration and health evaluation of transformer errors based on multi-source data fusion is proposed. In this study, the dataset of low- and medium-voltage energy distribution networks in Norway is used; meanwhile, the unstructured data in the format of Common Grid Model Exchange Specification is analyzed to construct a structured network topology model. Using the high-frequency active and reactive power load curves uploaded by the advanced metering infrastructure (AMI) system, combined with Kirchhoff's law, the mathematical model of power conservation and line loss correction in the substation is established; the error parameter identification equation is built by using the weighted least squares (WLS) method; the transformer ratio and phase error are iteratively solved without hardware intervention. The model further integrates meteorological environmental factors and constructs multi-dimensional state feature vectors to quantitatively evaluate the health of equipment. The improved adaptive WLS algorithm reduces the mean absolute percentage error of transformer ratio error to 0.063%, significantly better than the traditional method. Under the interference of a total harmonic distortion rate as high as 20%, the convergence success rate of the algorithm is still 92.5%, and the root mean square error of transformer ratio error is 0.253%; this shows strong robustness to nonlinear load interference. The coefficient of determination (R²) on the test set of the proposed health index state evaluation model integrating temperature, humidity and load rate data reaches 0.915. The state evaluation model with multi-dimensional environmental characteristics has a trend capture accuracy of 96.8% on the test set. Thus, it effectively realizes accurate prediction of the equipment health state.
The rapid popularization of controllable loads such as electric vehicles (EVs), battery energy storage systems (BESS), and heating, ventilating and air conditioning (HVAC) has greatly improved the flexibility of active distribution networks, but it also brings challenges to safe and efficient real-time scheduling. However, most existing control strategies rely on either static optimization or single-agent reinforcement learning, which have shortcomings in scalability, adaptability to dynamic operating conditions, and ensuring network constraint satisfaction. To remedy these deficiencies, we propose a multi-agent reinforcement learning (MARL) -based scheduling framework combined with quadratic programming (QP) safety projection, which unifies flexible resource modeling, learns adaptive cost-aware and comfort-aware control policies, and guarantees real-time physical feasibility. The case study on the IEEE 123-node test system shows the superior performance of the proposed method compared to the existing methods. Specifically, in a typical scenario with 40% PV penetration, the proposed framework achieves 19.3% peak demand reduction and 13.4% operational cost savings. Moreover, it maintains a high EV user satisfaction of 96.5% and strictly guarantees network security with a zero voltage violation rate, which indicates its potential for reliable deployment in modern active distribution networks.
The shift from fossil-based to clean-energy infrastructure has escalated the significance of wind power. Expanding the power system to accommodate wind generation remains challenging due to the natural fluctuations and uncertainty of wind speeds. Effective short-term prediction of wind speeds is key to maintain system reliability, efficient power scheduling, and dependable wind-energy integration. This work introduces a hybrid EEMD-ConvLSTM framework that hybridize Ensemble Empirical Mode Decomposition for signal decomposition and a Convolutional LSTM for spatiotemporal feature learning. Using EEMD, the raw time-dependent and structurally stochastic and highly variable wind-speed measurements are separated into a collection of intrinsic oscillatory components each obtaining meaningful time-varying sequence of the underlying dynamics. These functions are subsequently fed as input to the ConvLSTM network, which models both spatial and temporal dependencies to improve forecasting. The model's accuracy is examined using site-specific wind-speed data sourced from a power plant facility in Telangana, India. The forecasting accuracy is reflected in the MAE (0.0983 m/s), RMSE (0.1025 m/s), and MAPE (1.5472%) obtained during testing. The method performs better and runs faster than EEMD-LSTM and other conventional models, making it suitable for instantaneous utilization in smart-grid environments and renewable energy platforms
Solar drying serves as a sustainable technique for food preservation through solar technology intervention. The main purpose of this investigation is to design and construct a high-performance forced convection direct-type solar dryer for drying fish products of 20 kg capacity in a hygienic condition. For validation of experimental results, a thermal analysis of the solar dryer has been performed through CFD simulation using ANSYS Fluent. The performance indicators, such as maximum drying rate, drying efficiency, and specific energy consumption, are evaluated as 0.85 kg/hr, 63.5%, and 1.005 kWh/kg respectively, at the highest mass flow rate of 0.019 kg/s. The coefficient of determination (R2) for drying chamber temperature variation is found to be above 99%, indicating a good correlation between parameters. The highest temperature of drying chamber at 1 PM from the experiment and simulation is evaluated as 60.3 degrees C and 59.8 degrees C respectively. The equivalent solar heat flux is determined as 649.5 W/m2 and 646.244 W/m2 respectively. It was found that there is a close correlation between the simulation and experiment result of temperature, with an error rate of less than 1%, and for solar heat flux, it is between 1% to 5%. This type of dryer can be extremely beneficial in providing sustainable energy to dry various food products because of its economic effectiveness. The dryer is currently utilized by fisherwomen in the coastal regions of South Odisha.
This study aims to find out how well a PV solar facility possessing 400 MW grid-connected designed in Riyadh city, Saudi Arabia, can meet the demands of an electric load. A dispatch method that forecasts future solar production and electricity consumption was created using the HOMER program. This study investigates the optimal operation of three configurations for a solar plant with three different module sizes. The solar plant was designed using PVsyst and the results were exported to the HOMER program. The proposed system is compared to the base system, which includes load following as well as cycle charging regarding HOMER, is executed utilizing techno-economic optimization and environmental perspectives to obtain the Total Net Present Cost, the minimum Levelized Cost of Energy, and the minimum CO2 emission. The optimization procedures were executed with zero unmet loads and excess energy. The optimization findings illustrated that the suggested system with a large PV module has better energy, economic, and environmental results with the grid purchases (2,347,286,329 kWh/year), LCOE (0.0477 $/kWh), and CO2 emission (1,480,778 ton/year). The optimization procedures were executed with different load levels, and in all cases, the results confirmed that the plant designed with large panels is more economical.
