MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202631121362 A) filed by Nit Jamshedpur on October 09, 2026, for Intelligent Machine-Learning-Based Predictive Switching Control System And Method For A Fifteen- Level Asymmetric Cascaded Multilevel Inverter.
Inventors include Mrs. Khushboo Bharti; Dr. Mrinal Kanti Sarkar; and Dr. Simanta Kumar Samal.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: The invention relates to an asymmetric cascaded H-bridge multilevel inverter and an intelligent predictive control framework for the same. Three unequal DC voltage sources having a 1:2:4 voltage relationship are coupled through a source-selection switching network and a polarity-control H-bridge to generate fifteen output states including positive, negative and zero levels. A machine-learning prediction module receives operating and switching parameters including modulation index, switching angles, source voltages, load parameters and frequency-related variables. The prediction module employs one or more of an Artificial Neural Network, Extreme Gradient Boosting and Random Forest model to estimate inverter output characteristics, harmonic distortion and/or convergence-related performance. The predicted performance is used to support switching-strategy selection, modulation optimisation and control-loop adjustment. The source study reports R² values of 0.9711/0.9635 for ANN, 0.9999/0.9830 for XGBoost and 0.9977/0.9874 for Random Forest for training/testing, respectively. The system is applicable to renewable-energy, grid- connected, electric-vehicle and industrial power-electronic applications. Fig. 1
Disclaimer: Curated by HT Syndication.
