基于演化算法的数字滤波器优化设计

发布时间：2018-06-15 20:12

本文选题：数字滤波器 + 结构进化　；参考：《河南大学》2015年硕士论文

【摘要】：目前,滤波器在电子系统中已经成为一种不可或缺的组成部分,广泛应用于通信技术、图像识别、军事雷达、航空领域、医疗设备和语音等众多领域。卓越的信号处理能力使其具有越来越大的市场应用价值。模拟滤波器若要满足较高的精度或者多个技术指标,不仅设计过程复杂、元器件个数多,结构庞大,而且不一定能达到目标要求。随着电子计算机技术和大规模集成电路的不断发展,数字滤波器可用计算机软件或者大规模集成数字硬件实时实现。计算机技术的快速发展为数字滤波器的设计与实现创造了条件。但是,数字滤波器的功能要求不断提高,数字滤波器的结构越来越复杂,普通的设计方法很难满足要求。因此,借助计算机设计数字滤波器起到越来越重要的作用。近些年,许多研究者提出各种各样的算法应用于数字滤波器的设计。包括遗传算法(Genetic Algorithm,GA),模型退火算法(Simulate Anneal Arithmetic,SAA),禁忌搜索(Taboo Search,TS),蚁群最优化算法(Ant Colony Optimization,ACO),神经网络算法(Neural Network Algorithm,NEA)和人工免疫算法(Immune Clonal Selection Algorithm,ICSA)等等。但是这些方法都是先根据目标特性要求标定传输函数的系数,然后再考虑数字滤波器的结构,这种设计只保证了传输函数在系数标定阶段是最优的,但是在整个数字滤波器的设计过程中可能不是最优。本文主要提出一种基于演化算法的数字滤波器优化设计方法,利用遗传算法优化数字滤波器结构,在得到最优滤波器结构之后,再利用差分算法和步长变化算法优化滤波器系数,最终得到数字滤波器最优解。该方法能够根据目标特性要求直接设计数字滤波器结构,无需标定传输函数系数。本文主要进行以下几个方面的研究工作:(1)综述了数字滤波器设计方法,并着重阐述了数字滤波器的演化设计方法。(2)利用遗传算法设计数字滤波器结构。因为遗传算法的性能和效率主要由交叉率和突变率的取值决定,所以本文对这两个参数进行了深入的分析,并得出该算法最优的输入参数集。(3)利用差分算法和步长变化算法继续优化数字滤波器系数。与遗传算法优化结构所得实验结果进行比较,得出系数的优化对阻带最小衰减有明显的提高。(4)全文最后对研究内容进行总结并做出展望。通过本文的实验结果可以证明,基于演化算法的数字滤波器优化设计方法能够获得较好的实验效果。根据目标特性直接设计滤波器结构,无需滤波器阶数、传递函数等先验知识,为滤波器的设计提供了一种有效的设计方法,同时该方法也有广泛的适用性,可以解决其他类似的优化问题。
[Abstract]:At present, filters have become an indispensable part of electronic systems, widely used in communications technology, image recognition, military radar, aviation, medical equipment and voice and many other fields. The outstanding signal processing ability makes it has more and more market application value. If analogue filter is to satisfy higher precision or more technical indexes, not only the design process is complicated, the number of components is large, the structure is huge, but also the target requirement is not always met. With the development of computer technology and large-scale integrated circuits, digital filters can be realized in real time by computer software or large-scale integrated digital hardware. The rapid development of computer technology creates conditions for the design and implementation of digital filters. However, the functional requirements of digital filters are increasing, the structure of digital filters is becoming more and more complex, and the common design methods are difficult to meet the requirements. Therefore, the design of digital filters by computer plays an increasingly important role. In recent years, many researchers have proposed a variety of algorithms for digital filter design. It includes genetic algorithm, simulated Anneal algorithm, Tabu search algorithm, Ant Colony Optimization algorithm, Neural Network algorithm and artificial immune algorithm, and so on. However, these methods first calibrate the coefficients of the transfer function according to the target characteristics and then consider the structure of the digital filter. This design only ensures that the transmission function is optimal in the calibration stage of the coefficients. But it may not be optimal in the whole design process of digital filter. In this paper, a digital filter optimization design method based on evolutionary algorithm is proposed. Genetic algorithm is used to optimize the digital filter structure, and the optimal filter structure is obtained. Then the difference algorithm and step size algorithm are used to optimize the filter coefficients, and the optimal solution of the digital filter is obtained. This method can design the digital filter structure directly according to the target characteristics, without calibrating the transfer function coefficient. In this paper, the following aspects of the research work: 1) the design method of digital filter is summarized, and the evolutionary design method of digital filter is described emphatically. 2) the structure of digital filter is designed by genetic algorithm. Because the performance and efficiency of genetic algorithm are mainly determined by the values of crossover rate and mutation rate, the two parameters are deeply analyzed in this paper. The optimal input parameter set of the algorithm is obtained. The difference algorithm and step size algorithm are used to continue to optimize the digital filter coefficients. Compared with the experimental results obtained by genetic algorithm optimization, it is concluded that the optimization of the coefficient can significantly improve the minimum attenuation of the stopband. 4) at the end of this paper, the research content is summarized and the prospect is made. The experimental results in this paper show that the optimal design method of digital filter based on evolutionary algorithm can obtain better experimental results. The filter structure is designed directly according to the target characteristics without prior knowledge such as filter order, transfer function and so on. It provides an effective design method for filter design, and it also has wide applicability. Other similar optimization problems can be solved.
【学位授予单位】：河南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN713.7

【相似文献】