Study of real-time control in the optimization of human specimen fluid exchange process using gradient descent algorithm in adaptive regulation mode

The human specimen, due to its non-renewable nature, requires the liquid exchange process in adaptive regulation mode to realize precise control.In this paper, a stochastic parallel perturbationbased gradient descent algorithm (SPGD) is introduced into the real-time control system for human specimen liquid exchange.The SPGD algorithm is used to assist the real-time control system to monitor the liquid concentration and regulate the liquid exchange power in real time, so as to keep the liquid exchange speed fast and stable and reduce the risk of specimen damage.The advantages of SPGD algorithm and real-time control system in the process of human specimen liquid exchange are verified through several experiments.The results show that there is a correlation between different liquid concentrations, real-time power and liquid exchange speed, and the combination of SPGD algorithm and real-time monitoring of the changes between the three can improve the stability of liquid exchange.The SPGD algorithm with the introduction of stochastic parallel perturbations reaches an evaluation function value of 0.83 at 31 iterations and a convergence accuracy of 0.9698 after 150 iterations.The convergence speed and accuracy are better than the unimproved SPGD algorithm.During the real-time control process, the relative error of pipetting is not more than 1.5%, and the repeatability deviation is less than 1%.Within the time range of 0-120ms, the real-time control system introducing SPGD algorithm can give the changing power of liquid exchange, which guarantees the specimen safety in the process of liquid exchange.