Research On Multi-sensor Array Data Fusion Technology Based On Dangerous Chemical Gas Detection

At present,under the Made in China 2025 plan,China’s industrialization is at an important stage of transformation.The leakage of hazardous chemicals gas in industrial production has become one of the major dangers that threaten the safety of human life.Therefore,monitoring and detecting the concentration of hazardous chemical gases has become an indispensable safety precaution in people’s daily lives.Because the single gas sensor is characterized by poor selectivity and cross-sensitivity,it cannot meet the requirements for the detection accuracy of the gas concentration of hazardous chemicals.Therefore,research on multi-sensor data fusion technology for gas detection of hazardous chemicals is of great significance.In this paper,by understanding the most common hazardous chemical gas environment,the hazardous chemical gas composed of a mixture of CO gas and CH4 gas is selected as the detection and analysis object,and the appropriate semiconductor gas sensor is selected through the analysis of the principle characteristics of the sensor,and the multi-sensor 2 * 2 array and 2 *3 array planning.According to the requirements of the sensor and the array planning,the construction of the gas detection platform for hazardous chemicals was completed.This paper the BP neural network algorithm and generalized dynamic fuzzy neural network(GD-FNN)as the data fusion algorithm,GD-FNN algorithm not only can the input variable of fuzzy rules and the importance of an assessment,so that each rule of the width of the input variables can be carried out according to the size of its contribution to the system performance online adaptive adjustment.Four multi-sensor array data fusion models were constructed based on two algorithms The experimental detection data are used as training samples to train and test the four models.Finally,the advantages and disadvantages of different arrays and different data fusion algorithms are analyzed through the fusion results.The experimental results show that the gas concentration value error of the mixed hazardous chemical products obtained by the traditional calibration method is very large,and it is not suitable as a data processing method for the detection of mixed hazardous chemical gases.By comparing the fusion results of different models of the test samples,it can be found that the 2* 3 array fusion model has the same error level as the 2 * 2 array fusion model,and there is no obvious array advantage.The fusion result of the dangerous gas sensor data fusion model established by BP neural network has a large error,while the fusion result of the dangerous gas gas multi-sensor data fusion model established by the generalized dynamic fuzzy neural network(GD-FNN)algorithm has a small error.It can well fit the fusion value and the real value,and is more suitable as the multi-sensor data fusion algorithm of the gas detection environment in this paper.