Research On Accuracy Compensation Algorithm Of New MEMS Hydrogen Sensor Based On Neural Network

With the rapid development of industry,hydrogen concentration detection has become increasingly widely used in many industrial production fields,such as chemical synthesis,safety of nuclear power reaction,and development of hydrogen energy.However,most of the hydrogen sensors commonly used in practical applications currently have defects such as poor selectivity,complex structure,and high cost,which are difficult to improve.In response to this problem,our laboratory has developed a new type of catalytic combustion hydrogen sensor combined with MEMS technology.The new MEMS hydrogen sensor has the characteristics of rapid response(response time less than 0.8s,recovery time less than 3s),high sensitivity(57.74m V/%1H₂),simple structure,and low cost.Compared with other hydrogen sensors,it has greater advantages and application prospects.However,experiments show that the new sensor also has the problem of high linearity and susceptibility to temperature interference that affect the detection accuracy.Therefore,in order to realize the high-accuracy detection of the new MEMS hydrogen sensor,this thesis proposes to use the optimized BP neural network to carry out accuracy compensation research,and design a high-accuracy intelligent detection system based on the new MEMS hydrogen sensor.The research includes the following three parts:1.Analyzing the detection principle and main characteristics of the new MEMS hydrogen sensor,it is found that the sensor has some nonlinear errors and temperature drift,which affects its detection accuracy.Therefore,it is necessary to compensate the accuracy of the new MEMS hydrogen sensor.and further analyzing the advantages and disadvantages of the commonly used accuracy compensation methods.2.The software and hardware of the hydrogen sensor intelligent detection system are designed,and the hydrogen detection experiment platform is built based on this.This hydrogen sensor intelligent detection system can perform data fusion on the different collected information,and then display the processed output results for OLED display and WIFI transmission,which realizes the real-time display and wireless transmission of hydrogen concentration.3.In response to the lack of precision in the new MEMS,the standard BP neural network accuracy compensation algorithm is first researched;Then,in view of the defect that the standard BP neural network tends to fall into the local minimum,the genetic algorithm(GA)and the particle swarm algorithm(PSO)are used to optimize the BP neural network,and the GA-BP and PSO-BP accuracy compensation algorithms are constructed.Finally,it is proposed to use genetic algorithm and particle swarm algorithm to jointly optimize the BP neural network,and construct the PSO-GA-BP accuracy compensation algorithm.The simulation experiments on four accuracy compensation algorithms are carried out,and the compensation results are analyzed from the three aspects of temperature coefficient,training accuracy,and compensation error.It is found that the PSO-GA-BP algorithm has the best compensation accuracy and efficiency.Finally,the PSO-GA-BP accuracy compensation model is applied to the actual detection experiment.And the results show that the new MEMS hydrogen sensor after compensation has been achieved the purpose of high-accuracy detection.