一种基于FPGA的二次规划求解方法

An FPGA Implementation of Quadratic Programming

二次规划是一类重要的优化问题,许多工程上的优化问题都可以归结为二次规划的求解,如:预测控制,最小二乘等。嵌入式技术的发展使得嵌入式系统广泛地应用于人类生活的方方面面。由于二次规划的求解需要大量的计算,因此在传统基于ARM的嵌入式平台上实现比较困难。FPGA并行计算,硬件加速的特点使得在嵌入式平台上快速求解二次规划成为可能。本文主要介绍了一种基于FPGA的二次规划求解算法,首先介绍了二次规划的求解算法,然后将浮点算法改为定点算法,再采用Impulse C将定点算法变为HDL语言,最后通过实例验证了此方法的正确性。

Quadratic programming(QP) is an important optimization problem.Many engineering optimization problem can be simplified as a QP problem.The fast development of embedded technology made the embedded system exist in many aspects of our human life.Solving the QP needs lots of computation;therefore it's hard to implement the QP in the traditional embedded platform based on ARM.Due to the FPGA's characteristic:parallel computation and hardware acceleration,it's possible to solve the QP in the embedded platform.This paper mainly introduces an FPGA implementation of quadratic programming,a QP algorithm was introduced first,then the float-point algorithm was converted to fixed-point algorithm and implemented the fixed-point algorithm in HDL with Impulse C,an experiment was used to test the method last.