Co-design-based event-triggered proportional integral controller utilizing fixed period and combined triggering mechanism algorithms for cyber-physical system

Abstract

Cyber-physical systems (CPSs) are integrated systems where the physical process incorporates cyber components which include computation and communication/networking. The integration is usually in the form of a feedback loop, in which the cyber component constantly monitors and controls the physical process. Conventionally, a controller is designed only to achieve a physical goal, bringing the physical output to the desired setpoint with specific performance criteria. On the contrary, CPS needs to take into account both cyber and physical performances by enhancing the integration between both elements. As a result, a co-design approach is required to support the CPS feedback controller design that has the capability to reduce the cyber energy while maintaining the physical performance as the integration enhancement criteria. Due to this benefit, the CPS has started to be implemented for control of process plants. This thesis presents two improved event-based Proportional-Integral (PI) controllers, namely Fixed Period Algorithm (FPA) and Combined Triggering Mechanism Algorithm (CTMA), as CPS feedback controllers for industrial process control. The FPA and CTMA are procedurally designed according to the new co-design framework, where the FPA is designed to reduce the control computation algorithm while the CTMA mitigates the sticking and limit cycles issues. The framework consists of controller design, trade-off design, and design's evaluation processes. A conventional PI is initially designed, then the integration’s enhancement is introduced by using the event-based strategy in trade-off design, hence producing the FPA and CTMA. The development of FPA and CTMA are based on previous event-based PI controllers, namely Durand and Marchand Saturation Algorithm (DMSA) and Durand and Marchand Hybrid Algorithm (DMHA). The CTMA is an extension of FPA that combines absolute and relative errors as a triggering mechanism. By using an improved algorithm, FPA and CTMA reduce the control computation algorithm by 25% (2.4 pJ) and more than 64% (12.8 pJ) as compared to DMSA and DMHA, respectively. The performances of FPA and CTMA in reducing control updates are also compared to DMSA and DMHA for the case with and without network delays on the lag-dominant, balance, and delay-dominant processes. Network delays are represented by constant and time-varying delays, where the maximum delay values are determined using a simple stability criteria and Monte Carlo simulation in the design’s evaluation process. It is found that CTMA reduces control updates by 50% for the lag-dominant process and 10% for the balanced process based on simulation results without the presence of delay. With the presence of delays, the superiority of the CTMA is confirmed especially for the lag-dominant process, where CTMA improves approximately 50% of the computational load reductions and 70% of the physical performance compared to DMHA. Another intriguing discovery is that the FPA can achieve comparable performance to the DMHA despite using a simpler computation algorithm. Taken together, the clear benefits of FPA and CTMA are the trade-off designs that reduce the computational energy by reducing the control updates while maintaining the physical performance. It is envisaged that FPA and CTMA can be utilised for efficient CPS feedback control in industrial process control.