Physics-informed neural networks (PINNs) offer a promising route to bridge device-level simulations and compact circuit models. In this work, we present a hybrid modeling framework that integrates TCAD datasets with a baseline compact model and applies a PINN correction to capture stress-condition effects with high fidelity. The proposed approach achieves ≤ 2% route mean square error (RMSE) across more than 2,000 bias points, maintaining stable predictions under temperature (273–373 K) and radiation (0–100 krad) variations. Extracted Berkeley Short-channel IGFET Model (BSIM) parameters enable direct SPICE simulation, ensuring compatibility with standard circuit design workflows. For deployment, the trained PINN is exported as a quantized ONNX model, achieving sub-millisecond inference and ultra-low energy consumption (0.25 pJ/op) on a Cortex-M55 platform. This dual pathway supports both high-accuracy circuit simulation and real-time edge inference, making it suitable for embedded applications under constrained conditions. Comparative analysis with recent ANN-based models confirms that our physics-informed approach offers superior interpretability, SPICE readiness, and deployment efficiency. All datasets, code, and models are released to support reproducibility, benchmarking, and further research in compact modeling and edge-AI integration.
The increasing penetration of distributed generation (DG) significantly complicates Distribution System State Estimation (DSSE) by introducing stochasticity and uncertainty. This paper proposes a novel DSSE framework that unlike conventional methods simultaneously estimates the system state, load demands, and DGs output power through a unified constrained optimization model. The model is efficiently solved using the Whale Optimization Algorithm (WOA), whose unique balance of exploration and exploitation enables robust solution search in complex, active distribution networks. Simulation studies on standard IEEE 37-bus and 69-bus test systems reveal that the proposed WOA-based approach achieves outstanding accuracy. For the 37-bus system, WOA attains a Maximum Individual Relative Error (MIRE) of 1.15% and a Maximum Individual Absolute Error (MIAE) of 2.303 on load estimation. On the larger 69-bus system, the method further reduces these errors yielding a MIRE of 0.886% and a MIAE of 1.12 for load, and 0.73% and 1.058 for DG power estimation, respectively. Across all experiments, WOA consistently outperforms leading metaheuristics including ABC, PSO, and GA highlighting its superior accuracy, scalability, and robustness for real-world DSSE challenges.
Dynamic Network Reconfiguration (DNR) is a vital and effective technique for reducing energy loss. Due to its complexity, nonlinearity, and large-scale optimization challenge, DNR is still a very difficult problem. This paper presents a new strategy for improving the DNR's stability and reliability under Real-Time Operation Mode (RTOM). It addresses a simultaneous optimization technique within various limitations and constraints about network power flow, voltage limits, output generation of Renewable Energy Resources (RER), Distributed generation mode, and network load profile. In real-time operating mode, it optimizes Distributed Generations Sizing and Location (DG_SL) for Renewable Energy and Dynamic Network Reconfiguration (DNR). Reducing the overall daily active and reactive energy losses of the network, raising the Voltage Stability Index (VSI), distributing the load more evenly, and enhancing distribution efficiency in real-time operation mode are the primary goals. A Multi-Objective Decision-Making Approach (MODMA) based on the Analytic-Hierarchy Process (AHP) and Crow Search Algorithm (CSA). To evaluate the practicality of the proposed method, MATLAB simulations were conducted on the IEEE 33- and 69-bus networks. In the IEEE 33-bus case, the proposed AHP–CSA framework achieves up to 91.75% reduction in daily active losses and more than 90.70% reduction in daily reactive losses, with the Voltage Stability Index consistently improved toward unity. In the IEEE 69-bus case, the method delivers up to 81.78% reduction in daily active losses and 59.78% reduction in daily reactive losses, also enhancing the overall voltage stability profile. These outcomes confirm the effectiveness and robustness of the proposed approach for real-time distribution network operation with renewable DG integration.
A mobile robot must be autonomous to avoid obstacles while traveling towards the target. Dynamic obstacle avoidance remains a significant challenge in mobile robotics. Although reactive navigation strategies have been applied to address this problem, relying on the single-stage module often results in limited efficiency and restricted overall performance. This paper proposes combining an adaptive neuro-fuzzy inference system (ANFIS) and a neural network (NN). The data for obstacle severity classification were used to train the Neural Network. The relative velocity and distance between the mobile robot and obstacles determine the zone. Zone 1 is dangerous, and Zone 5 is safe. This paper uses the ANFIS to avoid obstacles during the mobile robot's motion and to avoid collisions. Based on our empirical study, three essential features have been considered in this paper: the relative speed, distance, and angle between the robot and the obstacle as inputs to the obstacle avoidance system ANFIS. The output was a suggested steering angle and speed for the mobile robot. The simulation results for the tested cases show the capability of the proposed controller to avoid static and dynamic obstacles in a fully known environment. Our results show that the ANFIS System enhances the proposed controller's performance, reducing path length, processing time, and the number of iterations compared to state-of-the-art research papers. The proposed work demonstrated better performance in path length reduction (approximately 6%) and time taken reduction to reach the target, which is reduced by about 60%.
In this paper, the decoding order error of successive interference cancellation (SIC) of multicarrier nonorthogonal multiple access (NOMA) due to the random walk of the users and position estimation deviation is considered in resource allocation. This factor extremely degrades the performance of NOMA in terms of sum rate and outage probability. Therefore, two optimal power allocation strategies for users are derived that maximize the sum rate and minimize the outage probability. The simulation results show that by considering the decoding order error in resource allocation, better performance can be achieved compared to the previous power allocation algorithms without considering this fact, which are a well-known water filling algorithm and a power allocation that maximizes the rate with minimum rate constraint.
