论文部分内容阅读
针对许多易受强随机干扰而难以用常规方法辨识的动态系统,提出一种非线性多变量动态系统的双重建模与分离辨识技术在FPGA上的实现。通过对非线性子通道特性分析,采用DDSI技术进行建模,包括构造BP神经网络、运用最小二乘辨识等算法获取数学模型,其中,建立模型的主要算法和系统参数估计的相关的FPGA结构设计通过Verilog语言的结构描述数据流进行了描述。同时采用FPGA的在线可重构技术,在运行时根据需要动态改变系统的电路结构,使硬件具有分时复用,节省逻辑资源的优良性能。这一技术的FPGA实现,使得这种非线性、多变量动态系统的特殊系统辨识技术能够广泛应用于诸如星体运动、现代控制过程、生产过程与经济管理系统等高层次科技领域中的带有强噪声干扰子通道的系统动态建模中。
Aimed at many dynamic systems which are easily disturbed by strong random disturbances and difficult to be identified by conventional methods, a dual-mode modeling and separation identification technique for nonlinear multivariable dynamic systems is proposed on FPGA. Through the analysis of the characteristics of the nonlinear subchannels, using DDSI technology modeling, including the construction of BP neural network, the use of least squares identification and other algorithms to obtain mathematical models, including the establishment of the main algorithm of the model and system parameter estimation of the relevant FPGA structure design The structure of the Verilog language describes the data flow. At the same time using FPGA’s online reconfigurable technology, the circuit structure of the system can be changed dynamically at runtime according to the needs, so that the hardware has time-sharing and saves the excellent performance of logical resources. FPGA implementation of this technology enables the special system identification technology of this nonlinear and multivariate dynamic system to be widely used in high-level science and technology fields such as astral movement, modern control process, production process and economic management system, System of Noise Interfering Subchannels in Dynamic Modeling.