Using a Neural Network to Minimize Pressure Spikes for Binary-coded Digital Flow Control Units

Abstract

A unique method of improving energy efficiency in fluid power systems is called digital flow control. In this paper, binary coding control is utilized. Although this scheme is characterized by a small package size and low energy consumption, it is influenced by higher pressure peaks and larger transient uncertainty than are other coding schemes, e.g., Fibonacci coding and pulse number modulation, consequently resulting in poor tracking accuracy. This issue can be solved by introducing a delay in the signal opening/ closing of the previous or subsequent valve, thus providing sufficient time for state alteration and valve processes. In a metering-in velocity control circuit, a feedforward neural network controller was used to create artificial delays according to the pressure difference over the digital flow control unit (DFCU) valves. The delayed signal samples fed to the controller were acquired through the genetic algorithm method, and the analysis was performed with MATLAB software.