Research On Infrared CO₂ Gas Sensor System Based On Neura Network Algorithm Compensation

The gradual increase of CO2 gas concentration in the atmospheric environment has brought many negative impacts on the global ecological environment,climate,and people's lives.Therefore,it is extremely urgent to accurately monitor the CO2 concentration in real-time.The non-dispersive infrared absorption method has the advantages of good selectivity and stability,wide measurement range,as well as fast response,which has gradually become the main method for on-line monitoring of CO2 gas concentration.The main work of this article includes:The relationship between the infrared radiation absorption efficiency in the sensor cavity and the cavity length of the gas chamber was emphatically studied,by using theoretical analysis and fluid dynamics simulation software.The calculated results show that the cavity length has an optimal value of 8 cm when the cavity diameter is 1 cm.As this time,the sensor's infrared absorption efficiency reaches its maximum and the sensor has the highest detection sensitivity.Based on the above analysis,the hardware and software modules of the CO2 sensor are designed.In the circuit,ICL7650 is used to amplify the electrical signal of the detector,and STM32 controls the ADC to perform high-speed signal acquisition.This circuit can effectively suppress the noise,which improves the signal to noise ratio of the system.Based on the simulation results,a CO2 infrared gas sensor was developed and it was calibrated and tested with different cavity lengths in the range of 0?2000 ppm.The experimental results show that the output signal of the CO2 sensor with 8cm cavity is the la gest and has high detection sensitivity,which is consistent with the simulation results.The experimental system was used to test CO2 concentration and compared with that of the referenced value.The average relative error of the three groups of experiments was 3.99%,5.1%and 3.55%,respectively,and the repeatability and stability error were 4.51%and 3.55%.In addition,we also conducted temperature-influenced experiments on the device.The results show that the higher the temperature the greater the error of the measurement results of the experimental system,in the range of 5 to 45?.Finally,the experimental results at different temperatures were compensated in the range from 0 to 2000 ppm.The absolute error between the measured and the standard concentration before compensation is up to 2500 ppm,while it is within±70 ppm after data fusion of the wavelet neural network modified by the genetic algorithm.Therefore,the measurement accuracy of the CO2 gas sensor is improved,which solves the problem of temperature drift of the sensor.The study of this article is exactly in line with the country's current demand for real-time monitoring of the atmosphere,and it is of great significance for the treatment of the atmospheric environment and the protection of human health.