Thông tin tài liệu
Journal ofScience & Technology 101 (2014) 127-133 Compensation for Closed-Loop Communication Time Delays in Can- Based Networked Control Systems Nguyen Trong Cac'*, Nguyen Xuan Hung^, Nguyen Van Khang' ''Hanoi University ofScience and Technology, No Dai Co Viet Str, Ha Noi, Viet Nam ^'Uint, 91190 Saint Aubin - France Received- December 14, 2013; accepted- Aprd 22 2014 Abstract In the context of Networked Control Systems (NCS), communication time delay strongly influences on the Quality of Control (OoC) of process control applications The goal of this paper is to propose a way to calculate the closed-loop communication time delay and we compensate this time delay using the pole placement design method m onier to improve the QoC for CAN-based NCSs We consider the implementation of several process control applications on a CAN network Then we show the interest of the proposed method by companng the QoC m cases of time delay compensation and of without time delay compensation Keywords: Communication time delay, Pole placement d 1, CAN network, Networked Control Systems Introduction Networked'Control Systems (NCSs) are feedback control systems in which controllers, sensors and actuators are interconnected usmg a common bus network The NCSs are more and more important and becommg an altemative to pomt-to-point architecture because of many advantages such as reduction of installanon and mauitenance cost, more easiness in diagnosttc and configurability However, time delays are induced due to the transmission of frames from confrollers to actuators (feed-forward channel) and from sensors to controllers (feedback charmel) These time delays degrade the QoC of the confrol system [ -6] considered the Several works compensation f d y Tl k [1] d [2] only considers d y mp f f d forward chann I h ks [ ] d [4], NCS is m d d yn hr dyn m I system (ADS) h commumcatioi d Iy t db k h I The compensa h m pi d by the approx imat f h pi pi d I d h time stamped t m d d p k Tl k [ ] mp t and [6] consid d h m d I both feed-forw d d f db k ban I 11 d d I > h closed-loop c these works considered that the time delays are constant and bounded or anticipated, if the delay is not m the expected range, the system could be crashed * Conesponding Author Tel: (+84)904369421 Email: cac nguyentrong@hust edu.vn Fig Implementation of a process control application on a CAN network The goal of this paper is to analyze and propose a compensation method for closed-loop communication time delays using pole-placement design method We consider the implementation of several mverted pendulum confrol applications on a CAN network and we show the good QoC obtained by the proposed method We use the tool TrueTime [7] for the simulation and the QoC evaluation, Tmetime is a toolbox based on Matlab/Simulink which allows to simulate real-time network-based control systems The rest of the paper is organized as follows the section II presents the model of an NCS; section III concerns the proposition of computation and compensation of the communication time delays compensation; Section IV represents the implementation of process confrol applications (inverted pendulum) on a CAN network and the QoC evaluation; the conclusions are given in section V of the paper Journal ofScience & Technology 101 (2014) 127-133 access and the transmission time of&f„ frame (noted D„) Modeling a can-based ncs 2.1 Implemeatation of a application on a CAy network process control The general model of the implementation of a process control applicanon through a network is shown in Fig The sensor receives the sampled output yt provided by the Analog Digital (AD) converter and sends il to the controller via the communication network The conttoller receives _y* from the sensor, then calculates the conttol signal «t and sends w* to the actuator via the communication network The actuator receives «*, converts Uk into analog signal iu(t)) usmg the Digital Analog (DA) converter and then duectly applies u(t) to plant The Zero Order Hold (ZOH) keeps the value u(l) until the new sampling tune We have two flows of frames, the SensorController flow which is concerned the frames going from the sensor to the confroller (noted fsc flow and call "/ic frame" a frame of this fiow), the ConfrollerActuator flow which is concerned the frames going from the controller to the actuator (noted,^^ flow and call'%o frame" a frame of this flow) The sensor task is Time-Triggered (i.e controlled by the sampler clock) The controller and actuator tasks are Event-Triggered (the events are the receptions of the fc frames and f„ frames, respectively) We call h the sampling period [8] which is defined by the following formula: 0.1 PrioJ,cJ > > > Prlojcji > PrioJ,c_n > I.e the controller has a higher priority than the sensor and durmg a sampling period, the order of medium access is ananged as Pi, Pj, , , Pn, Proposal for computation and compensation of time delays 3.1 Computation of closed-loop communication time delays The computation of closed-loop communication time delays is done by the confroller in each sampling penod Concerning the time delay Zs^-, as the controller has the knowledge of sampling instants (* (A = 0, I ,,), it can easily deduce the value of Ht, by the time difference between the reception instant of f^ frames and Ik Conceming the time delay r™, it cannot be calculated because the fa frames have not been transmitted yet However, due to the hypotlieses in Section 2 (ie thepriority of the controller is higher than this of the sensor; there is no competition between the confrollers; there is no data lost), the controller can immediately send its frame Therefore Zca IS equal To the duration ofajrame transmission (Dc^ The closed-loop time delay will be computed by the controller as followed- Journal ofScience & Technology 101 (2014) 127-133 3.2 Time delay compensation steps The compensation for time delays done by the controller in each samplmg period has the following steps: Step I: Identifying expected poles including the dominant poles and the other poles [8] which are selected equally to the real part of the dominant pole divided a, with a = -^ 10 Step 2: Computing losed-loop communication time delay Step 3: Computmg the confroller parameters according to the time delay value in order to maintain the position or the value of the expected poles Step 4: Computmg he confrol signal based on the new confrol parameters calculated in the previous step, 3.3 Code implemented in the controller The implemented code is as followed: read_inputs x(ti); calculate_closed loop delay: z- r,c + Zca', calculate_state mafrix-
Ngày đăng: 08/12/2022, 17:26
Xem thêm:
