EMI Prediction on Ripple Effect of Voltage-Controlled Ramp Waveform via ARIMA-MLR Hybrid Model

Predicting electromagnetic interference (EMI) is crucial for maintaining the functionality of electronic devices. Accurate prediction of EMI is essential for daily electronic applications, such as setting the brightness of the smartphone and controlling the volume of the television, to avoid potential damage to electronic functionality. This paper presents an analysis of the ripple effect of the voltage ramp signal under control based on frequency modulation for EMI prediction. While the time series is in static mode, this paper suggests to control the amplifier voltage as prediction task to maintain different frequencies with varying EMI radiation rather than arbitrary settings. A mathematical analysis was conducted to model EMI radiation from 29 scenarios of amplifier experiments and 751 frequencies using a combined model of auto-regressive integrated moving average (ARIMA) and multiple linear regression (MLR). To evaluate the impact of data splitting on the model performance, the model was tested with 30%, 70% data splits, and all the data sets. The findings indicate that the (7,0,4) model is the highest accuracy achieving of 88.52%. This model was enhanced with multiple linear regression, resulting in an improved accuracy of 96.37% ensuring the optimal performance and longevity of electronic devices in various scenarios and frequencies. The study highlights the enhancement of the EMI prediction model, which is primarily based on frequency modulation and three-dimensional ripple effects of ramp signals.