项目名称: 混成系统微分不变式计算理论方法
项目编号: No.61202131
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 刘江
作者单位: 重庆绿色智能技术研究院
项目金额: 24万元
中文摘要: 如何计算微分不变式是混成系统设计、分析以及验证的一个核心问题。最近我们对代数混成系统提出了一个完备方法用之生成和检验半代数微分不变式。该方法与已有的其他方法都主要针对确定性代数混成系统。但是,实际混成系统可能由于测量误差或者控制设计等因素而包含某些参数,其模型由于刻画某种物理规律而包含非多项式的子项。参数介入引起的参数纠缠和非多项式子项对模型可计算性的破坏使得一般混成系统微分不变式的研究踟蹰不前。另外,我们的方法虽是完备的,但它的复杂度制约其应用效果。因而本项目将结合完备方法,(1)通过分析不变式边界的奇点性质研究提高该方法效率的技术,(2)探索参数混成系统以及(3)初等混成系统的微分不变式生成与检验技术。本项目的基本目标是基于我们的完备方法加强对混成系统微分不变式生成技术的研究,突破当前只能计算确定性代数混成系统微分不变式的瓶颈。研究结果有望广泛应用于混成系统的分析、设计与验证。
中文关键词: 微分不变式;变量变换;初等系统;半定规划;卡曼滤波
英文摘要: Generating differential invariants is a key point in the design, analysis and verification of hybrid systems. Recently, we invented a complete method to generating and verifying semi-algebraic differential invariants for algebraic hybrid systems. This method and others mainly focus on the deterministic parameter-free algebraic hybrid systems. However, many practical hybrid systems contain parameters derived from the measurement error or the design of system and non-polynomial terms originated from physical motion. The parametric entanglement and the destruction of computability by non-polynomial terms are the main abstacles to generate differential invariants for general hybrid systems using curent methods. At the same time, the complexity of our complete method limits its application in practice. Therefore, based upon our complete method this project will (1) improve our complete method by investigating the propoerty of sigular points of differential invariants, (2) explore the methods to generating differential invariants for hybrid systems with parameters, and (3) attack the generation of differential invariants for elementary hybrid systems. In summary, our basic target is to develop proper methods, based on our complete method, to generating differential invariants of hybrid systems for various cases. Accor
英文关键词: Differential Invariant;Variable Transformation;Elementary System;Semi-definite Programming;Kalman Filter