利用機器學習RFID混合模型的運維檢修移動作業應用

趙水忠 王劍 顧曄

摘? 要： 針對傳統的運維檢修方法無法準確定位出故障點的缺點，提出利用機器學習RFID混合模型的運維檢修移動作業應用研究。首先，將射頻識別技術與決策樹算法結合，建立機器學習RFID混合模型;在模型的基礎上，采用運維檢修移動作業裝置中的RFID讀寫器，通過射頻識別技術識別出與運維檢修點相對應的標簽，完成運維檢修點的數據采集;通過去重過濾算法，去除采集到的重復數據;通過監督學習，訓練數據，對故障點與非故障點進行精確分類，再對其做運維檢修處理，完成利用機器學習RFID混合模型的運維檢修移動作業方法的設計。通過與傳統的運維檢修方法作對比實驗，實驗結果表明，提出的利用機器學習RFID混合模型的運維檢修移動作業方法具有更高的定位精度。

關鍵詞： 運維檢修; 機器學習; 數據采集; RFID混合模型; 重復數據去除; 故障點定位

中圖分類號： TN99?34; TP391? ? ? ? ? ? ? ? ? ? ? 文獻標識碼： A? ? ? ? ? ? ? ? ? ? 文章編號： 1004?373X（2020）09?0157?04

Application of operation and maintenance mobile assignment

using machine learning RFID hybrid model

ZHAO Shuizhong1， 2， WANG Jian2， GU Ye2

（1. Shanghai Jiao Tong University， Shanghai 200240， China; 2. State Grid Zhejiang Electric Power Company， Hangzhou 310000， China）

Abstract： An application research of the operation and maintenance mobile assignment using the machine learning RFID hybrid model is proposed to overcome the shortcomings of the traditional operation and maintenance method that can not accurately locate the fault point. The RFID technology and the decision tree algorithm are combined to establish a machine learning RFID hybrid model. On the basis of the model， the tag corresponding to the operation and maintenance inspection point is identified with the RFID technology and the RFID reader in the mobile operation device to complete the data acquisition of the operation and maintenance inspection point. The deduplication filtering algorithm is used to remove the collected duplicate data. The supervised learning and the training data are used to accurately classify the fault point and the non?fault point， and then the operation and maintenance processing is adopted for the classification result to achieve the design of the operation and maintenance mobile assignment method using the machine learning RFID hybrid model. A contrast experiment of the method designed in this paper and the traditional operation and maintenance method was carried out. The experimental results show that the designed operation and maintenance mobile operation method using machine learning RFID hybrid model have higher positioning accuracy.

Keywords： operation and maintenance overhauling; machine learning; data acquisition; RFID hybrid model; duplicate data removing; fault point positioning

0? 引? 言

近年來，隨著運維檢修工作量的不斷加大，傳統的運維檢修方法已無法滿足實際需求，需要開發新的技術和方法提高運維檢修技術的水平。目前，一些企業采用移動作業的方式，改進了運維檢修的模式，為運維檢修工作提供了便利[1]。然而，現有的運維檢修移動作業方法無法及時、準確地識別出需要做運維檢修處理的目標位置，導致運維檢修的效率大大下降[2?4]。隨著科技水平的不斷提高，機器學習和射頻識別技術得到了不斷的發展。……