Computer Vision-based Framework for Power Converter Identification and Analysis

This research work is mainly focused on identifying the individual components in a hand-drawn schematic diagram, and thus performing simulative analysis of a power converter. YOLOR (You Only Learn One Representation) – the state-of-the-art deep learning-based object detection model is used to detect the electronic components i.e. resistor, capacitor, diode, etc. in a circuit diagram. A Hough transform algorithm is used to trace the horizontal and vertical wire connection, whereas KMeans clustering is used to segregate the points-of-intersection between those horizontal and vertical lines to identify the nodes in the circuit. By using all of this circuit information, a netlist of the circuit is generated – that can be fed into any spice-based circuit simulators. In this work, PySpice – an open-source python module, is used to auto-simulate the identified hand-drawn schematic diagram. In future, this work will be extended to automate the PCB design of the detected hand-drawn circuit diagram. The overall workflow algorithm of this research work is as depicted in the flowchart.

Proposed method for the automated simulation of a hand-drawn schematic of the power converters