Chemical and infra-red sensors generate distinct responses under similar conditions because of sensor drift, noise or resolution errors. In this paper, we develop novel machine learning methods for detecting and identifying VOC and Ammonia vapor from time-series data obtained by uncalibrated chemical and infrared sensors. We process time-series sensor signals using deep neural networks (DNN). Three neural network algorithms are utilized for this purpose. Additive neural networks (termed AddNet) are based on a multiplication-devoid operator and consequently exhibit energy efficiency compared to regular neural networks. The second algorithm uses generative adversarial neural networks so as to expose the classifying neural network to more realistic data points in order to help the classifier network to deliver improved generalization. Finally, we use conventional convolutional neural networks as a baseline method. Our findings indicate that using raw time-series data obtained from uncalibrated sensors and processing them using deep-learning-based methods yield better results than using hand-crafted feature parameters.

Introduction

Ammonia and Volatile organic compounds (VOCs) are associated with numerous health problems. Although VOCs and Ammonia are naturally occurring, they can nonetheless cause serious health issues in high concentration. For example, exposure to Ammonia in high concentration causes harm to skin, lungs and eyes. Methane and other VOC compound leaks contribute to global warming. VOC compounds such as benzene and toluene are carcinogenic [1]–[۴]. In this paper, we develop novel machine learning methods for uncalibrated VOC and ammonia vapor sensors.1 We show that there is no need to use either expert-crafted features or thresholds for the purpose of interpreting the time series data that these sensors generate. This is advantageous in that it exempts system designers from extracting discriminative and robust features from the raw time-series sensory data, which is a nontrivial task. To the best of our knowledge, this is the first work that uses the raw time-series signal for VOC detection and ammonia vapor sensing using deep learning. We train the deep learning structures using only the time-series data. Our algorithms are applicable to both infrared (IR) and chemical sensor systems. In particular, they can be used for early detection and, thus, prevention of dangerous gas leaks. Mobile infrared and chemical sensors can be part of an open air cyber-physical system (CPS) [5]–[۷]. We use the time-series data obtained by the sensors in order to detect accidental and/or deliberate gas vapor leaks. The main contribution of this paper centers on the exploitation of the time-series data that sensors produce rather than conventional reliance on a single or a couple of sensor readings for leak detection.