關(guān)于大壩監(jiān)測(cè)數(shù)據(jù)質(zhì)量評(píng)價(jià)因子及算法研究

摘" 要： 大壩監(jiān)測(cè)數(shù)據(jù)是判斷大壩運(yùn)行安全的主要依據(jù)，為了鑒別數(shù)據(jù)優(yōu)劣并選擇出可信度較高的數(shù)據(jù)，文中構(gòu)建一個(gè)大壩監(jiān)測(cè)數(shù)據(jù)質(zhì)量評(píng)價(jià)框架。針對(duì)測(cè)點(diǎn)之間的相關(guān)性、監(jiān)測(cè)項(xiàng)目及儀器的特點(diǎn)，利用Kshape算法找出具有強(qiáng)相關(guān)性的測(cè)點(diǎn)，再通過(guò)相對(duì)偏移率、相對(duì)平滑率、周期波動(dòng)程度和精度修正率等評(píng)價(jià)因子對(duì)大壩監(jiān)測(cè)數(shù)據(jù)進(jìn)行評(píng)價(jià)；其次，結(jié)合混合蝙蝠算法優(yōu)化后的長(zhǎng)短期記憶網(wǎng)絡(luò)對(duì)大壩監(jiān)測(cè)數(shù)據(jù)進(jìn)行分類，構(gòu)建了大壩監(jiān)測(cè)數(shù)據(jù)質(zhì)量評(píng)價(jià)算法流程。以新疆某大壩監(jiān)測(cè)數(shù)據(jù)為研究對(duì)象進(jìn)行試驗(yàn)，結(jié)果表明所提出的大壩監(jiān)測(cè)數(shù)據(jù)質(zhì)量評(píng)價(jià)算法的準(zhǔn)確率為94.33%，可為評(píng)價(jià)大壩監(jiān)測(cè)數(shù)據(jù)質(zhì)量提供有效的解決方法。

關(guān)鍵詞： 大壩監(jiān)測(cè)數(shù)據(jù)； 評(píng)價(jià)因子； 數(shù)據(jù)質(zhì)量評(píng)價(jià)； 長(zhǎng)短期記憶網(wǎng)絡(luò)； 測(cè)點(diǎn)聚類； 相關(guān)性分析

中圖分類號(hào)： TN919.2?34； TV698.2" " " " " " " 文獻(xiàn)標(biāo)識(shí)碼： A" " " " " " " " " " " 文章編號(hào)： 1004?373X（2025）02?0090?07

Research on evaluation factors and algorithms of dam monitoring data quality

FENG Yuyang1， LI Denghua2， 3， FANG Boya4， DING Yong1

（1. School of Science， Nanjing University of Science and Technology， Nanjing 210094， China; 2. Nanjing Hydraulic Research Institute， Nanjing 210029， China;

3. Key Laboratory of Reservoir Dam Safety， Ministry of Water Resources， Nanjing 210029， China; 4. Huashe Testing Technology Co.， Ltd.， Nanjing 211100， China）

Abstract： Dam monitoring data is the main basis for judging the safety of dam operation. In order to identify the data quality and select the data with high reliability， a dam monitoring data quality evaluation framework is constructed. According to the correlation between measuring points and the features of monitoring items and instruments， Kshape algorithm is used to find out the measuring points with strong correlation， and then the dam monitoring data is evaluated by means of the evaluation factors such as relative offset rate， relative smoothness rate， periodic fluctuation degree and accuracy correction rate. In combination with the LSTM （long short?term memory network） optimized by hybrid bat algorithm， the dam monitoring data is classified， and the algorithm flow of dam monitoring data quality evaluation is constructed. The test is conducted by taking a dam monitoring data in Xinjiang as the research object. The results show that the accuracy of the proposed dam monitoring data quality evaluation algorithm is 94.33%， which can provide an effective solution for evaluating the quality of dam monitoring data.

Keywords： dam monitoring data; evaluation factor; data quality evaluation; long short?term memory network; measuring point clustering; correlation analysis

0" 引" 言

大壩監(jiān)測(cè)數(shù)據(jù)是判斷大壩運(yùn)行安全的主要依據(jù)，目前在智能化趨勢(shì)下，大壩傳統(tǒng)的數(shù)據(jù)質(zhì)量評(píng)價(jià)方法存在效率低下、受主觀影響的缺點(diǎn)，為解決上述問(wèn)題，國(guó)內(nèi)外學(xué)者提出新的解決思路。

文獻(xiàn)[1]對(duì)現(xiàn)有的隨機(jī)森林進(jìn)行優(yōu)化，提出了6項(xiàng)評(píng)價(jià)因子及相關(guān)評(píng)價(jià)標(biāo)準(zhǔn)構(gòu)成的安全監(jiān)測(cè)歷史數(shù)據(jù)質(zhì)量評(píng)價(jià)方法，建立了基于隨機(jī)森林的大壩安全監(jiān)測(cè)歷史數(shù)據(jù)質(zhì)量評(píng)價(jià)算法。文獻(xiàn)[2]提出了一種基于格拉姆角場(chǎng)（GAF）的卷積神經(jīng)網(wǎng)絡(luò)算法，實(shí)現(xiàn)了基于卷積神經(jīng)網(wǎng)絡(luò)的實(shí)時(shí)圖像處理，以及對(duì)實(shí)時(shí)監(jiān)測(cè)結(jié)果的自動(dòng)評(píng)估。……