The Study Of Measurement System Of Diffusion Silicon Pressure Sensor Based On PSO-LSSVM And BLE

Sensor technology is one of the most important high-tech in present world,and it is also an important symbol of contemporary science and technology development.With the development of the Internet of things,the sensor has gradually transitioned from the industrial field to the life field.The sensor measurement accuracy and the measurement system power consumption have become the main problems.In this paper,the factors influencing the measurement accuracy of the diffusion silicon pressure sensor are studied,and the three main causes of the influence are as follows: the stability of the power supply,the temperature drift and the input-output nonlinearly characteristics.Aiming at the problem of power supply stability,this paper designs a constant current source based on TL431 to supply the diffusion silicon pressure sensor;A temperature compensation model based on particle swarm optimization(PSO-LSSVM)is proposed to solve the problem of temperature drift.The PSO-LSSVM model was established by real-time monitoring of the experimental temperature by using a temperature sensor to perform a two-dimensional calibration experiment on the diffusion silicon pressure sensor.Because the penalty factor of the least squares support vector machine(LSSVM)and the selection of the kernel function will directly affect the prediction accuracy of the model,PSO-LSSVM model utilizes particle swarm optimization to optimize the penalty factor of the least squares support vector machine model and the parameters of the kernel function,which improves the traditional least squares support vector machine model to select the time-consuming and ineffective parameters and does not necessarily to find the global optimal solution.The experimental results show that the zero temperature coefficient and the sensitivity temperature coefficient are reduced by an order of magnitude after the compensation of the PSO-LSSVM model,and the mean square error of the predicted and calibrated values is 10-6.The temperature compensation is achieved and the prediction accuracy is improved.For the nonlinear problem,this paper uses the curve fitting method to correct the input and output characteristic curve.At the end of the paper,we built a low-power Bluetooth(BLE)sensor measurement system based on the PSoC chip.The system consists of signal amplification circuit,low-pass filter circuit,temperature monitoring module,sequence ADC module and BLE module,and it has normal work mode,sleep mode and deep sleep mode,and the system can automatically enter the low power mode when the measurement is completed.