Research Quartz Flexible Accelerometer Temperature Characteristics Model Identification Methods

The paper mainly focuses on the characteristics of quartz flexible accelerometeroutput value, which varies with the external ambient temperature. In the backdrop ofmicrogravity environment of space, sudden temperature changes, the input ofaccelerometer ranges from-1g to1g，The regularity between the accelerometers outputvalue and the external environment is studied. The suitable model for the accelerometersinput and output is established. The temperature static and dynamic experimental studieswere conducted and then the identification for the static and dynamic models is taken.Based on the model identification, accelerometer model temperature compensation issuesare researched in order to improve the flexibility quartz accelerometer output accuracygreatly in the research background. Temperature compensation is conducted in the staticand dynamic temperature. The combination of these two compensation methods, themethod to compensate the accelerometer output value in the whole environmentstemperature (-20℃~+50℃) is given as well as the Compensation contrast. The mainsteps are as follows:1. Through studying the characteristics of the accelerometer structure in theory, thereasons to cause the temperature drift error of quartz flexible accelerometer output signalare analyzed. By temperature static experiments, the temperature of various componentsinside the case accelerometer node distribution is analyzed. Combined with thetemperature coefficient of components in theory, the temperature error scope of theaccelerometer is calculated quantitatively. This is reason why the zero bias and scale factorfor quartz flexible accelerometer are drifting from one temperature to another.2. The flexible quartz accelerometer temperature characteristics of the experiment areimplemented, including static, dynamic temperature experiments. Experimental method isin1g gravity field, the points turning accelerometers in (2points,4points,12points)experimental methods. Ambient temperature ranges from-20℃to50℃. Experimentalplatform software real-time collects and storages the accelerometer output value andtemperature in dynamic and static temperature. Collection and measurement minimumcircuit is based on chip of C80c51f300series.24bits A/D conversion circuit is adopted inminimum circuit. The resolution of conversion circuit is0.149μV. The frequency ofconversion circuit ranges from1Hz~10Hz. Digital temperature sensor DS18B20is selected for sampling internal temperature. Minimum circuit can effectively complete datareal-time acquisition and processing and provide reliable data for the research ofcompensation scheme.3. The method to identification model in the static temperature characteristicexperiment is researched. Static temperature model researches temperature characteristic ofaccelerometer output value in various equinoctial points when the static temperature rangesfrom-20℃to55℃. According to the different experimental tumbling points under1ggravity field in the static temperature characteristics experiment, the two-pointidentification model method, the four-point identification model method, multi-pointidentification model method are studied, respectively4. The method to identification model in the dynamic temperature characteristicexperiment is researched. Dynamic temperature identification model resolves the problemwhich the output value lags temperature when it changes rapidly. In the paper, threedynamic temperature conduct model, differential transmission model, dynamic weightedaverage model and neural network model are introduced, respectively. Given thecompensation coefficients, each model is used to compensate and verify the output valueso that the optimal parameters in each model can be found. Taken the optimal modelparameters, the compensation effect of three kinds of model is contrasted by four items.5. Temperature static and dynamic model are combined for model comparison. Fouritems, the accuracy of output value, fluctuation of index factor, consumption of hardwareresources and compensation of time, are adopt to compare and evaluate the models.