金属材料矿浆输送管道内部液、固两相流电阻层析检测机理及评价方法研究

The delivery mode of mineral pulps is gradually becoming a main approach for mineral delivery with its own benefits, such as low cost, no pollution and higher transportation capabilities, etc. In our country, the transportation pipes for mineral pulps are mainly made of metal materials. With the influences of many factors, such as concentration, flow speed and volumes of particles of pulps, the metal pipes are apt to be blocked due to settlements and scales, etc., which seriously endangers the safety of mineral pulp transportation. In this project, the author puts forward an ERT (Electrical Resistance Tomography) based on conductor walls so as to visually test the two-phase flow of liquid and solid of the pulps in metal pipes. With this method, firstly establish a sensitivity field model for ERT sensors of the conducting walls to make theoretical analysis and value simulation on the characteristics of “soft field” and “edge effect”; then introduce the mode identification and smart control technologies into ERT sensor design for conducting walls and then research and develop a kind of self-adaptable multi-cross sections sensor of conducting walls; The sensor can be calibrated online and automatically optimized subject to different flow types. Meanwhile, with smart optimization methods, develop high precision ERT algorithm; finally, study on the analysis and evaluation on influence factors for multiple-phase fracture distribution ERT detection of pulps in metal pipes through the established test platform. Based on the results, a comprehensive calibration and performance evaluation method for multi-parameter ERT detection for metal pipes is proposed. With the study under this subject, the non-destructive test for pulp pipe industry shall be technically supported.

矿浆的管道输送方式，以低成本、无污染等优势成为世界矿物运输的主要途径，在我国矿浆的运输管道以金属材料为主。受矿浆浓度、流速、固体颗粒大小等因素的影响，使得管道内部出现沉积、结垢等现象，容易造成管道堵塞，严重威胁矿浆的运输安全。本课题提出一种基于导体壁的电阻层析检测方法，对金属管道内矿浆进行液、固两相流的可视化检测。该方法首先建立导体壁ERT传感器的敏感场模型，对“软场”特性与“边缘效应”进行理论分析与数值仿真；然后根据仿真结果将模式识别与智能控制技术引入导体壁ERT传感器设计，研发一种自适应式多截面导体壁ERT传感器；该传感器可以实现根据不同流型，进行在线标定与自动优化。同时利用智能优化方法，开发高精度ERT 成像算法；最后，通过所搭建的试验平台进行矿浆在金属管道内多相断层分布ERT 检测影响因素分析与评价方法研究，并根据研究结果，提出一种用于金属管道ERT检测的多参数标定与性能评价方法

由于矿浆管道输送安全的迫切需要，本项目发展一种针对导体管道液固两相流的电阻层析检测方法。已完成如下研究内容：.1提出一种自适应式多断面导体壁ERT传感器。该ERT传感器采用动态结构，并将流型识别技术引入ERT传感器设计，提出一种基于信号稀疏性的ERT流型识别方法。该方法采用信号的稀疏表示方法将ERT系统的采样电压表示为稀疏性组合，并求出其稀疏解用以实现对不同的流型进行分类。流型识别信息的引入，使得该传感器具有根据实时的流型变化，自适应地动态调整传感器结构，优化传感器性能的智能化功能。实验表明该传感器可以自动识别芯流、泡状流、层流、环流等4种典型流型，识别率分别为：88%、91%、90%、85%。针对不同流型，经动态优化后的传感器可使ERT图像重建的质量显著提高。.2提出一种多指标优化方法。该方法运用模糊数学的思想通过定义ERT指标满意度（ERTIS）和ERT指标综合满意度（OERTIS）的概念，构造满意度函数，建立一种新的科学合理的模糊评价方法。它把量纲、物理意义完全不同的ERT传感器的各项指标，统一为对ERT指标的满意度，并在满意程度上具有可比性。并通过OERTIS对多指标正交实验中ERT传感器各性能指标进行模糊分析，将多指标正交实验问题转化为单一指标的正交实验问题。为ERT传感器进行在线的多指标的优化和标定，提供了一种高效实用的新方法。.3提出一种SAP算法用ERT图像重建。首先利用LS-SVM预测不同流型时，由软场效应引起的重建误差。利用该误差构造退火粒子群算法的适应度函数,并采用退火粒子群算法进行寻优。实验结果表明该算法可以明显改善图像重建精度。.4提出基于粒子群优化核函数极限学习机（PSOKELM）的煤泥输送管道堵塞预测方法，该方将支持向量机核函数引入极限学习机，并通过粒子群算法进行参数优化。实验表明：与粒子群算法优化支持向量机（PSOSVM）预测模型和核函数极限学习机（KELM）预测模型进行比较，基于PSOKELM的预测模型在预测速度和准确性方面均优于PSOSVM预测模型，在预测精度上优于KELM预测模型。