Design Of Control System For Small Automatic Biochemical Analyzer

Biochemical analyzer is an important in vitro diagnostic equipment,commonly used for routine clinical tests such as renal function,liver function,blood glucose and blood lipid.At present,most of the domestic large hospitals and medical institutions use imported biochemical analyzers,whose price is tens of millions and the maintenance cost is high.Small and medium-sized hospitals,laboratories,clinics and other institutions are deterred.Therefore,it is of great significance to develop a low-cost,miniaturized and highly reliable automatic biochemical analyzer.Based on STM32F103ZET6 as the main controller,the control system of small automatic biochemical analyzer is designed,which mainly includes human-computer interaction software system,sample adding motion control system,temperature control system,micro pipetting system,etc.The five segment S-curve acceleration and deceleration algorithm is used to realize the sample adding motion control.The A4988 stepper motor driver is used to realize the stepper motors of pipette manipulator X direction,sample tray Y direction,pipette manipulator Z direction and incubation tank mechanism Z direction.The measured control accuracy of each direction is less than 0.2mm,which meets the system accuracy requirements.Temperature control is realized by using single neuron adaptive PID control algorithm.PT100 temperature sensor is used to collect the information and the heating plate is controlled by heating circuit to heat the incubator to a constant temperature.The steady-state error of the measured temperature control system is less than 0.2 ?,which meets the requirements of temperature accuracy.The speed position double closed-loop PID algorithm is used to realize the pipetting control.The motion control of the plunger stepper motor is realized by using TMC2208 stepper motor driver.The deviation correction of the liquid displacement is realized by linear compensation of the least square method,and the accuracy of the liquid transfer is improved.The experiment of liquid level detection and pipette monitoring are carried out effectively by using pressure method.The OV5640 was used to collect the sample images of the pipette process,and the pipette data set was established.The u-net neural network model was pruned.The model was trained and tested before and after pruning.The results showed that the model parameters(Params)and calculation(FLOPs)decreased by 93.99% and 47.30% respectively,which improved the operation speed and efficiency of the model.Based on image segmentation algorithm of pruning u-net neural network,the image of the pette area is segmented and analyzed,and the model of judging abnormal pipette is established,and the algorithm model of pipette volume detection is established combined with the geometric characteristics of pipette.Sub pixel corner detection and elm method are used to compensate the error of image segmentation and liquid displacement.In the quantitative pipette test of the whole machine,the pipette accuracy of10?L,50?L and 100?L reached 1.72%,1.36% and 1.39%,respectively,which met the requirements of precision design of micro pipette system.Image based liquid transfer monitoring can effectively judge the abnormal liquid transfer and detect the amount of liquid transfer,improve the reliability of the system,and has a certain guiding significance for the development and monitoring of micro pipette technology.