Compensation Algorithm Research Of Temperature And Humidity Sensor Based On BP Neural Network

The HMP45D temperature and humidity sensors are manufactured by Vaisala Finland and is often used for meteorological observations in recent years.However,in practical applications,it is easily affected by some environmental factors,such as temperature,humidity,pressure and radiation.These results lead to a great error in the measurement results,which greatly reduce the accuracy of the measurement.Therefore,it is urgent to compensate the measurement error of the temperature and humidity sensors by taking a suitable method.In order to solve the problem that the HMP45D type temperature and humidity sensors are easily susceptible by external environmental factors in practical business applications,this article will combine the method of artificial fish swarm algorithm(Artificial Fish Swarm Algorithm)with back propagation BP neural network algorithm.In addition,improved simplified particle Swarm Algorithm(simple particle Swarm optimization)combined with BP neural network will also be applied to compensate the temperature and humidity sensor separately.The main tasks include the following:The analysis of the environmental factors of the temperature and Humidity Sensor and the study of the influence degree.Firstly,this part analyzes the interference caused by environmental factors to temperature and humidity sensors,find out the main influence factors.Then verify the results through the relevant standard experiments.Finally,it is concluded that temperature is the biggest environmental impact factor which interferes with humidity sensor measurement results,and humidity also brings great error to the measurement of temperature sensor through the analysis of the experimental data and diagram.The establishment of temperature compensation model of humidity sensor.This part is pay attention to study the Artificial fish Swarm algorithm(AFSA)and use the global optimization ability of AFSA to find the right of the optimal threshold value of BP neural network,then establish AFSA-BP neural network model to simulate temperature compensation.According to the measured data of the temperature influence experiment,this model is compared with the traditional BP neural network method.The results show that the AFSA-BP neural network effectively avoids the defect of the BP neural network to the local minimum,and greatly improves the accuracy of the measurement of humidity sensor.The establishment of humidity compensation model for temperature sensor.The basic particle swarm optimization(PSO)algorithm is analyzed in this part and find the simplify particle swarm optimization(SPSO).Then,establish the improved SPSO-BP neural network model to simulate humidity compensation by adopt the linear decreasing inertia weight.According to the measured data of the humidity influence experiment,this model is compared with the traditional BP neural network method.The results illustrate that the improved SPSO-BP neural network model has the higher compensation accuracy and faster convergence speed,and effectively reduces the influence of humidity on the temperature sensor.