A New Hydrogen Sensor Fault Diagnosis Method Based on Transfer Learning With LeNet-5.

Yongyi Sun,Tingting Zhao,Bin Shen,Shuang Zhang,Zhihui Zou,Ying Yu,Hongquan Zhang,Shuxia Liu

doi:10.3389/fnbot.2021.664135

Yongyi Sun, Tingting Zhao + Show 6 more

Open Access

https://doi.org/10.3389/fnbot.2021.664135

Copy DOI

Abstract

The fault safety monitoring of hydrogen sensors is very important for their practical application. The precondition of traditional machine learning methods for sensor fault diagnosis is that enough fault data with the same distribution and feature space under the same working environment must exist. Widely used fault diagnosis methods are not suitable for real working environments because they are easily complicated by environmental conditions such as temperature, humidity, shock, and vibration. Under the influence of such complex conditions, the acquisition of sensor fault data is limited. In order to improve fault diagnosis accuracy under complex environmental conditions, a novel method of transfer learning (TL) with LeNet-5 is proposed in this paper. Firstly, LeNet-5 is applied to learn the features of the data-rich datasets of gas sensor faults in a normal environment and to adjust the parameters accordingly. The parameters of the LeNet-5 are transferred from the task in the normal environment to a task in a complex environment by using the TL method. Then, the migrated LeNet-5 is used for the fault diagnosis of gas sensors with a small amount of fault data in a complex environment. Finally, a prototype hydrogen sensor array is designed and implemented for experimental verification. The gas sensor fault diagnosis accuracy of the traditional LeNet-5 was 88.48 ± 1.04%, while the fault diagnosis accuracy of TL with LeNet-5 was 92.49 ± 1.28%. The experimental results show that the method adopted presents an excellent solution for the fault diagnosis of a hydrogen sensor using a small quantity of fault data obtained under complex environmental conditions.

Highlights

With the gradual depletion of traditional energy sources and the development of clean fuel, clean fuel plays a prominent role throughout many fields (Tsujita et al, 2005; Brown et al, 2015)
This paper proposes transfer learning (TL) with LeNet-5 method for gas sensor fault diagnosis in a complex environment, which involves two domains: the source domain and task domain
The accuracy of fault diagnosis can be improved by using TL with LeNet-5 method

Summary

Introduction

With the gradual depletion of traditional energy sources and the development of clean fuel, clean fuel plays a prominent role throughout many fields (Tsujita et al, 2005; Brown et al, 2015). As hydrogen is a clean fuel, its application range is rapidly expanding (Chalk and Miller, 2006). It is widely used in meteorological science, aerospace technology, the metallurgical industry, the electronics industry, national defense, the chemical industry, and so on, and its consumption is increasing rapidly (Poirier and Sapundzhiev, 1997; Winter, 2005; Staffell et al, 2019). Semiconductor gas sensors have been widely used in hydrogen detection based on SnO2-sensitive materials (Fedorenko et al, 2017; Zhang Q. et al, 2018). Sun et al proposed a new convolutional neural network method for hydrogen sensor fault diagnosis in 2020 (Sun et al, 2020)

Methods

Findings

Discussion

Conclusion