Abstract

Sensor drift is a major problem in chemical sensors that requires addressing for reliable and accurate detection of chemical analytes. In this paper, we develop a causal convolutional neural network (CNN) with a Discrete Cosine Transform (DCT) layer to estimate the drift signal. In the DCT module, we apply soft-thresholding nonlinearity in the transform domain to denoise the data and obtain a sparse representation of the drift signal. The soft-threshold values are learned during training. Our results show that DCT layer-based CNNs are able to produce a slowly varying baseline drift signal. We train the CNN on synthetic data and test it on real chemical sensor data. Our results show that we can have an accurate and smooth drift estimate even when the observed sensor signal is very noisy.

Highlights

  • Chemical sensors have been used for detecting and identifying chemical analytes in a wide array of industrial and safety applications [1]

  • Despite the fact that chemical sensor technology provides practical solutions, sensor responses may degrade with time, resulting in inconsistent results.This phenomenon is known as sensor drift

  • One particular variant of convolutional neural networks, the temporal convolutional neural network (TCNN), has been shown to outperform recurrent neural networks on many benchmark data sets [12, 13]. This motivates us to employ a TCNN-based framework to address the problem of drift correction in chemical sensor data

Read more

Summary

Introduction

Chemical sensors have been used for detecting and identifying chemical analytes in a wide array of industrial and safety applications [1]. The chemical sensory system becomes unreliable over time, if the sensor drift signal is not properly estimated. One particular variant of convolutional neural networks, the temporal convolutional neural network (TCNN), has been shown to outperform recurrent neural networks on many benchmark data sets [12, 13] This motivates us to employ a TCNN-based framework to address the problem of drift correction in chemical sensor data. Our results show that the the nonlinear denoising and smoothing in transform domain using the DCT-based structure removes the high-frequency features, i.e., induces smoothness in the features and, subsequently, the constructed output baseline drift signal.

Causal DCT-based Real-Time Sensor Drift Estimation Framework
TCNN with a DCT layer
DCT-based Layers
Experimental Results
16 TCNN-DCT
Conclusion

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.