# Intelligent Robust Control Design with Closed-Loop Voltage Sensing for UPS Inverters in IoT Devices

## TL;DR

This paper introduces a new control method for UPS inverters in IoT devices to improve performance and reduce errors during power outages.

## Contribution

The novel contribution is combining modified gray fast variable structure sliding mode control with a neural network to enhance UPS inverter performance.

## Key findings

- The proposed control technique reduces chattering and steady-state errors in UPS inverters.
- The use of a neural network compensates for prediction errors, improving dynamic response and reducing harmonic distortion.
- Simulations and DSP implementation confirm the effectiveness of the control method under various load conditions.

## Abstract

High-performance UPS inverters prevent IoT devices from power outages, thus protecting critical data. This paper suggests an intelligent, robust control technique with closed-loop voltage sensing for UPS (uninterruptible power supply) inverters in IoT (internet of things) devices. The suggested control technique synthesizes a modified gray fast variable structure sliding mode control (MGFVSSMC) together with a neural network (NN). The MGFVSSMC allows system states to speedily converge towards the equilibrium within a shorter time while eliminating the problems of chattering and steady-state errors. The MGFVSSMC may experience state prediction errors when the UPS inverter is subjected to external highly nonlinear loads or internal parameters changing drastically. This results in high harmonic distortion and inferior dynamic response of the inverter output, affecting the guarding of the IoT device. An NN by means of a learning mechanism is employed to properly compensate for the prediction error of the MGFVSSMC, achieving a high-performance UPS inverter. The suggested control technique operates with one voltage sensing, which can yield fast transience and low inverter output-voltage distortion. Both simulations and digital signal processing (DSP) implementation results demonstrate the effectiveness of the suggested control technique under a variety of load conditions.