Research On Gas Flow Control System Of Atomic Fluorescence Analyzer

Gaseous hydride of the element produced by the atomic fluorescence analyzer in a sodium borohydride reduction system can be analyzed according to the intensity of the atomic fluorescence.In the process of measurement,the gaseous hydride of the element to be measured needs to be transmitted in the functional parts of the instrument by means of the gas flow control system.If the controlled gas flow rate is unstable,it will lead to the distortion of the instrument measurement results.According to the requirements of the gas flow control system of the atomic fluorescence analyzer,this thesis designs a gas flow control system with fast response and strong robustness,which is aimed at solving the problems of traditional PID control algorithm and improving the detection performance of atomic fluorescence analyzer.In the gas flow control system,the PID control algorithm has become one of the most representative algorithms in the field of gas flow control because of its simplicity,efficiency and practical verification.After comparing and analyzing the advantages and disadvantages of positional and incremental PID control algorithms,this thesis chooses the incremental PID control algorithm and uses the incomplete differential method to suppress the proportional and differential saturation phenomena that may occur in the incremental PID controller.However,the parameter setting of the traditional PID controller is complicated and its adaptation is deficient leading to the limited potential of the PID control algorithm.Based on this phenomenon,this thesis introduces the artificial neural network PID controller,and applies the traditional PID controller,the BP neural network-based PID controller and the RBF neural network-based PID controller to the simple and complex nonlinear systems respectively.The MATLAB simulation results finally select a PID controller based on RBF neural network with better control quality as the control scheme of the system.At the same time,the momentum term is added to the RBF neural network to optimize it,so that the network can remain stable while speeding up the learning rate.In the process of hardware design,the author finished the power supply circuit design of the driving circuit of proportional solenoid valve,DA conversion circuit,flow sensor AD acquisition circuit and each module of gas flow control system.And use STM32F405RGT6 as the core controller,which is solely responsible for the control of gas flow,as a communication slave.It communicates with the atomic fluorescence analyzer master(as the host)through the serial port.In the software design part,using a custom encrypted communication protocol to ensure the security of communication data,and then equipped with FreeRTOS embedded operating system,the advantages of multi-task rotation allows the core controller to maximize its real-time control advantages.At the same time,in order to achieve a more intelligent control system,the microcontroller needs to have the function of saving data after power down.We need to store the PID parameter when the control precision has been reached in the internal FLASH of the chip.If the change command is not received,when the system is powered on,the value stored in the FLASH is used as the PID control parameter,which makes the system more intelligent.and efficient.At last,the control effect of the gas flow control system is tested experimentally and the test results show that the maximum control error of the system is less than 1% of the expected value.The gas flow control system is applied to the performance verification of the atomic fluorescence spectrometer.The verification test results are better than the performance indicators required in the evaluation outline of the atomic fluorescence spectrometer(JJF 1695-2018),which also proves that the gas flow control scheme in this thesis is effective.