一种闭环控制系统的快速开发.pptVIP

THE RAPID DEVELOPMENT OF A CLOSED-LOOP CONTROL SYSTEM 一种闭环控制系统的快速开发 1.3. Closed-loop control systems Although system functionalities may be differed from control systems to control systems,the control algorithms (PID, MPC, etc.) employed are the core function modules.Furthermore, the development cycles and performance of the systems are also mainly determined by the implementation of those algorithms[6]. When those algorithms are implemented on the reconfigurable hardware, the first issue raised is how to reach high performances of algorithms with limited hardware resources. The second issue raised is how to have a general chip-level design model, in which modules and algorithms can be reconfigured. 各种控制系统的功能可能不同,不过控制算法的配置都是控制系统中的核心功能,而且控制算法影响开发周期和系统性能..控制算法应用到可重构硬件的时候,第一要考虑怎样在有限的硬件资源上达到高性能,二是怎么设计一般性的芯片级设计模型,使模块和算法可以在上面重构. Input: one matrix A with n x l elements and one matrix B with l x m elements Output: one matrix C with n x m elements Begin Stepl. Choose Nc. (number of computing cells) Set j=1 (initialize clock cycle count) Step2. For each clock cycle j Xl - Bj;Xk+1 - Xk;j - j+1;k=1,2,...,j; If Ncj，then stop Step3. Set j=1 (initialize clock cycle count) For each clock cycle j Cj -A1xXj;Ck -A j-k+1x Xk;j } j+1;k=1,2,...,j; If lj，then stop End If it is assumed that the n is the maximum of three variables: n, m and 1, then the computation complexity in the conventional ways is O(n3 ).The maximum computation complexity after the proposed algorithm is O(n2一n+1) and then can be deduced asO(n2).For example, if the clock rate is 100 clock cycles/ps, simulated on the testing platform, the computational delay of the proposed algorithm, for the two 5x5 matrices, is 0.29 ps similar to the theoretical results, while the computational delay of the conventional algorithm is 1.40 ps. It can be seen that the computation speed by the proposed algorithm is almost five times faster than tha