Nowadays, with rapid and increasing demands of industrial and consuming products, mass production has been a new industry trend. Many manufacturing machines (processes) are operating simultaneously and in parallel to produce products. Process control design in mass production remains in the stand-alone mode by two main kinds of strategies: feedback control and feedforward control, in which each machine (process) make use of information only from itself. As communication techniques are more and more mature and advanced, real time information sharing between the machines (processes) becomes possible. This provides good chances for process control design to be shifted from the conventional standalone mode to a peer-wise information sharing mode for enhancing control performances of each in...[ Read more ]

Nowadays, with rapid and increasing demands of industrial and consuming products, mass production has been a new industry trend. Many manufacturing machines (processes) are operating simultaneously and in parallel to produce products. Process control design in mass production remains in the stand-alone mode by two main kinds of strategies: feedback control and feedforward control, in which each machine (process) make use of information only from itself. As communication techniques are more and more mature and advanced, real time information sharing between the machines (processes) becomes possible. This provides good chances for process control design to be shifted from the conventional standalone mode to a peer-wise information sharing mode for enhancing control performances of each individual machine (process).

This research focuses on one of the common scenario in mass production: multiple identical processes are operating simultaneously under the same environment, whose process dynamics is time-invariant and the uncertainty part of dynamics can be approximated by Linear-in-unknown-parameters (LIP) formulae, As an exploration of peer-wise sharing mode, this research treats each machine (process) as an ‘agent’, defines peer-wise sharing mode as the ‘inter-agent learning’ method, selects the adaptive control strategy as the design platform for the algorithm development. According to the property of unknown parameters of LIP formulae to be online estimated, this work further separates the problem into two cases: constant or ‘pseudo time-varying’ parameters. For each case, a series of inter-agent learning adaptive control methodologies have been developed in this research, theoretical analyses and numerical simulations are conducted to prove the superior performance of the developed adaptive control method using peer-wise information sharing over conventional adaptive control.