Modeling the stochastic mechanism of sensor using a hybrid method based on seasonal autoregressive integrated moving average time series and generalized estimating equations

• Published in: ISA transactions 2022-06, Vol.125, p.300-305
• Main Authors: Mahmoudi, Mohammad Reza, Baroumand, Salman
• Format: Article
• Language: English
• Subjects: Generalized estimating equations; Modeling; Seasonal autoregressive integrated moving average; Sensor; Time series forecasting methods
• ISSN: 0019-0578; 1879-2022
• DOI: 10.1016/j.isatra.2021.07.013

Sensor, which is one of the main components of control system, plays its vital role in measuring state and output system variables and highlights the importance of having desired statistical information about sensor output signals because it can be monitored, stored, or used as the primary input signal in other devices. However, these signals display noises (i.e. system noise and measurement noise) and even if the effects of system noises are faded away or removed from measured data, there is still some stochastic noise remained in the measurements. Even though SARIMA has been effective in modeling the stochastic noise in the sensor, the present study has found out the necessity of designing a novel approach including a combination of seasonal autoregressive integrated moving average (SARIMA) and polynomial generalized estimating equations (PGEE), to evaluate the stochastic behavior of sensors. Finally, the study tried to employ the proposed approach in real load-cell sensor data to examine its effectiveness. •It’s necessary to extract useful and desired information from general noisy sensor output signals.•Each measured data from sensors is noisy and contains some types of noise including system and measurement noise.•Even if we could remove system noise from measured data, there may still exist some stochastic noise in measurements.•A new hybrid technique, based on the combination of the polynomial generalized estimating equations (PGEE) and seasonal autoregressive integrated moving average (SARIMA) time series, is proposed to model the stochastic noise of the sensor.•The proposed strategy is applied in the implementation of load-cell sensor data.