摘要
为了能更加快速和精准地检测并定位矿井水的泄漏点,提出基于时间序列和压力变化特征的矿井水漏点定位方法。相空间重构矿井水历史压力时间序列,获取压力时间序列向量,并通过选择适当嵌入延迟和嵌入时间提升压力时间序列相空间重构效果;将压力时间序列向量输入模糊RBF网络进行训练,同时使用改进的差分进化(DE)算法优化模糊RBF网络连接权值,提升其学习能力和收敛速度;将实时的压力时间序列数据作为训练后的模糊RBF网络输入,获取矿井水漏点检测结果,若存在泄漏则计算出矿井水的漏点与测量位置的距离,由此确定矿井水的漏点。实验结果表明,当嵌入延迟和维度均为5时,压力时间序列相空间的重构结果最好,优化后的模糊RBF网络具备较好的学习速度和收敛速度,且矿井水泄漏检测准确度高,该方法可有效检测矿井水的多处泄漏点。
In order to detect and locate mine water leakage points more quickly and accurately,a mine water leakage point localization method based on time series and pressure change characteristics is proposed.Reconstruct the historical pressure time series of mine water in phase space,obtain the pressure time series vector,and improve the phase space reconstruction effect of pressure time series by selecting appropriate embedding delay and embedding time;Input the pressure time series vector into the fuzzy RBF network for training,and use the improved differential evolution(DE)algorithm to optimize the connection weights of the fuzzy RBF network,improving its learning ability and convergence speed;Real time pressure time series data is used as input for the trained fuzzy RBF network to obtain the detection results of mine water leakage points.If there is a leakage,the distance between the mine water leakage point and the measurement location is calculated to determine the mine water leakage point.The experimental results show that when the embedding delay and dimension are both 5,the reconstruction results of the pressure time series phase space are the best.The optimized fuzzy RBF network has good learning and convergence speed,and the accuracy of mine water leakage detection is high.This method can effectively detect multiple leakage points in mine water.
作者
李丁卯
张荣华
LI Dingmao;ZHANG Ronghua(Shanxi Luneng Hequ Electric Coal Development Co.,Ltd.,Xinzhou,Shanxi 036500,China)
出处
《自动化与仪器仪表》
2025年第10期138-141,145,共5页
Automation & Instrumentation
