【摘要】：隨著天然氣戰(zhàn)略資源作用和地位提升,運輸管道的運行安全越來越受到人們的重視。管道在運行維護時,需要進行實時監(jiān)測,準確判斷管道運行狀況,及時診斷泄漏隱患,避免安全事故。雖然管道泄漏檢測技術在不斷改善,但是在管道泄漏檢測中,仍然會出現(xiàn)錯報、漏報、誤報的情況。因此,本文針對這一情況,給出天然氣管道泄漏診斷的設計方案,最終實現(xiàn)管道泄漏高準確率的智能診斷。本文利用經驗模態(tài)分解EMD方法能夠將原始信號依照不同的頻率尺度下逐級分解,將這些不同尺度的波動或趨勢提煉形成本征模態(tài)函數(shù),再對能夠體現(xiàn)原始信號特性的各個本征模態(tài)函數(shù)提取能量熵和近似熵特征特征。由于傳統(tǒng)經驗模態(tài)分解方法中存在模態(tài)混疊效應的弊端而進行深入分析,最終采用總體經驗模態(tài)分解EEMD方法和近似熵、能量熵相結合方法進行特征提取。應用支持向量機對四種特征提取方法提取的特征向量組進行模式識別診斷分析,并進行識別效果對比,判斷最佳的特征提取方式。支持向量機識別診斷方法的復雜度和泛化能力由懲罰因子C和核函數(shù)參數(shù)g決定的,為了提高識別診斷的準確率,需要一個精確、快速、穩(wěn)定的方法來尋找最優(yōu)參數(shù)。本論文應用Libsvm軟件平臺分別采用網(wǎng)格搜索參數(shù)尋優(yōu)、粒子群算法、遺傳算法以及粒子群與遺傳結合算法對支持向量機管道泄漏類型分類的懲罰因子C和核函數(shù)參數(shù)g進行優(yōu)化,并進行分類準確率效果對比,最終達到高準確率模式診斷的目的。
[Abstract]:Along with the natural gas strategic resource function and the status promotion, the transportation pipeline operation safety receives the people's attention more and more. When the pipeline is running and maintaining, it is necessary to carry on real-time monitoring, accurately judge the running condition of the pipeline, diagnose the hidden danger of leakage in time, and avoid the safety accident. Although pipeline leakage detection technology is constantly improving, but in pipeline leakage detection, there will still be misreporting, misreporting. Therefore, in view of this situation, this paper gives the design scheme of natural gas pipeline leakage diagnosis, and finally realizes the intelligent diagnosis of pipeline leakage with high accuracy. In this paper, the empirical mode decomposition (EMD) method can be used to decompose the original signal step by step according to different frequency scales, and the waves or trends of these different scales can be extracted to form the intrinsic mode function. Then the energy entropy and approximate entropy characteristics of each intrinsic modal function which can reflect the characteristics of the original signal are extracted. Due to the disadvantages of modal aliasing in traditional empirical mode decomposition (EMD) methods, the general empirical mode decomposition (EEMD) method and approximate entropy and energy entropy are used to extract the features. The feature vector groups extracted by four feature extraction methods are analyzed by using support vector machine (SVM), and the recognition effect is compared to judge the best feature extraction method. The complexity and generalization ability of SVM recognition and diagnosis method is determined by penalty factor C and kernel function parameter g. In order to improve the accuracy of recognition and diagnosis, an accurate, fast and stable method is needed to find the optimal parameters. In this paper, the Libsvm software platform is used to optimize the parameters of grid search, particle swarm optimization (PSO), genetic algorithm (GA) and the combination of particle swarm optimization (PSO) and genetic algorithm (GA) to optimize the penalty factor C and kernel function parameter g of pipeline leakage type classification of support vector machine (SVM). The result of classification accuracy is compared to achieve the goal of high accuracy mode diagnosis.
【學位授予單位】：東北石油大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：TE973.6

【相似文獻】

相關期刊論文 前10條

1 徐晴晴;張來斌;梁偉;;天然氣管道泄漏聲場特性研究[J];石油機械;2011年02期

2 龐海艷;;天然氣管道泄漏及其檢測方法的幾點分析[J];中國石油和化工標準與質量;2013年17期

3 劉存貴,張英杰;煤氣管道泄漏不停產搶修[J];煤氣與熱力;2003年12期

4 張朝陽,蘇華東;用流速測試儀確定管道泄漏點的位置[J];石油規(guī)劃設計;2004年06期

5 瞿f，

本文編號：2460036