subject: The Worst Case Response Time for Programmable Logic Controller [print this page] The Worst Case Response Time for Programmable Logic Controller
PLC (Programmable Logic Controller) is sequence control devices that are developed to replace electro-mechanical relay hard wired controllers. Programmable logic controllers are equipped with microprocessors and control the plant by periodic actions such as sensing the state of the plant, updating actuators, and executing the control task. They are used widely in flexible manufacturing systems, transportation systems, chemical processes, etc. as factories progress in automation, PLCs accommodate more actuators and sensors distributed in vast area, which results in high wiring cost. Currently, local I/O subsystems get connected over field network to replace wire bundles. PLCs support multi tasking facilities to handle multiple sequence control programs at the same time. Consequently we have scheduling in PLCs including networked I/O, as in general purpose multi processing computers.
A popular performance metric of the programmable logic controller has been the speed of a processing unit, I/O network and a subsystem. However the more accurate metric is the worst case response time that guarantees timely control actions. The response time of a PLC is a control context and different from the conventional response time in scheduling. The worst case controller response time is the worst case delay between the state change of the plant and the corresponding action of the PLC. A PLC detects the state change of the plant on periodic execution and the worst case processing delay of the programmable logic controller. The worst case controller response time has been shorten by reducing the worst case processing delay while the period of execution remains fixed as a given term. The system bus performing bit wise operation was introduced for multi processor architecture to reduce the worst case processing delay by shortening the I? Update time. A multi processor architecture based on the data flow architecture was suggested for the instruction level multi processing to reduce the worst case processing delay. The network contention was resolved in loosely coupled multi processor PLCs to reduce the worst case processing delay.
The worst case controller response time depends on various factors as PLC become complex. A sequence control program for a large scale plant contains of many subprograms, which have precedence relations and contend for the shared resources such as a common memory, a network and a system bus. This may results in the increased worst case controller response time due to the discontinuous execution of the related sub programs and shared contention of shared resources. The proper sub programs schedule can reduce the worst case controller response time, almost all conventional scheduling methods premise a fixed given execution period. We have the worst case controller response time as a given constraint while others are design parameters in the control context. We propose a scheduling method to adjust both the worst case processing delay and the execution of period so that we may handle directly the worst case controller response time.
Another scheduling objective, efficiency is required in addition to the original objective, schedulability in scheduling task of a PLC. Efficiency achieved by lowering the required resources for guaranteeing the specified worst case controller response time. It affords to execute more other tasks, thus efficiency is a practical requirement as the PLC executes the task more efficiently. When we adjust the execution period and the worst case processing delay, there is a trade off relation between schedulability and efficiency. A schedule with the longest execution period is the most efficient one because he tasks occupy the smallest portion of resources provided the specified worst case controller response time is guaranteed. But the long execution period results in a short worst case processing delay under the specified worst case controller response time. We have to pay more cost for processing power to schedule tasks under the short worst case processing delay.
