Realization Of Radiosonde Humidity Measurement System And Design Of Compensation Model

High humidity detection to improve the accuracy of weather forecasts in predicting precipitation, calculate air humidity factor of the ground, provide meteorological research data has important significance. Humidity sensor based on humidity-sensitive capacitance is the best choice of humidity measurement in high altitude, however, due to the low temperature properties of high altitude, resulting humidity sensor surface is easy to icing or condensation, the final performance for the humidity sensor has a larger measurement error. In this paper, for the purpose of reducing the moisture measurement error, from the perspective of heated twin method, computational fluid dynamics simulation, acquisition circuit design, compensation model, designed a humidity measurement system in high altitude.1. Describes the design idea of the full, analyzes the influencing factors of humidity measurement in high altitude and gives a humidity sensor integrated serpentine heated resistor, uses two humidity sensors to rotation measurement and heated-twin method.2. Using the above humidity sensor, established five kinds of sensor spacing model and four kinds of flight attitude model. Using fluid dynamics simulation method, optimal sensor spacing can be obtained by analyzing the distribution of temperature field, the optimal flight attitude angles is determined by observing wall shear stress. Using transient analysis, obtain heating and cooling time from 0 to 32km, to determine the measurement period. The simulation results show, optimal spacing is 3.5-1 Omm, the flight attitude angle is 45, the heating time is 2s and the cooling time is 2min in the rated heating power.3. By analyzing the principle of signal generation, select the multi harmonic oscillation circuit as signal conversion circuit, by using the simulation data to provide a reference for the design of hardware and software, at the same time using ZigBee wireless sensor network as data transmission. Combined with the existing experimental conditions, a experimental is designed in order to know temperature effects on humidity, which can get the influence on the capacitor becase of the change of temperature and humidity. With the purpose of minimize temperature effects, by segmented least squares fitting method, draws datas between standard humidity and humidity measurement.4. To avoid errors of nonlinear factors such as temperature impact on humidity measurement, establish regression model of humidity sensor by GA-SVM, and compare with the BP neural network, wherein the data samples are from the experiment. Simulation shows, GA-SVM is more suitable than the BP neural network in modeling of humidity sensor, the errors between prediction values of the model and humidity are at 5%RH or less at a temperature of-50 to 25, and the fitting accuracy and generalization have good results.