【摘要】 通常检测到的输油管道压力信号存在着信噪比较低的问题,通过应用实数编码的改进遗传算法对最小均方误差(LM S)自适应滤波器的权值向量W进行寻优,避免了收敛步长μ选择不当所引起的误差;遗传算法的并行操作也有效地防止了W在递推过程中收敛于局部最优解.仿真结果表明,该方法简单,运行时间短,去噪效果良好,可为泄漏检测和定位提供可靠的数据.更多还原

【Abstract】 The pressure signal of oil pipelines usually detected has a rather low noise quantum. Selecting the best weigh vector of LMS adaptive filtering by means of real-coded Genetic Algorithm can eliminate the errors brought by improperly selected the convergence step μ and in addition avoid partial convergence of W. The results of simulation have shown that the method can obtain simplicity of implementation, conservation of operating time and a better reconstruction performance, and if can provide credible data for leak detection and location.更多还原