-
摘要: 為了給數控機床故障的精準診斷提供保障，延長數控機床使用周期，以數控機床歷史維修記錄為研究對象，對數控機床設備故障領域的命名實體識別進行了研究。在分析歷史維修記錄中的故障描述特點后，提出了一種基于雙向長短期記憶網絡（Bidirectional long short-term memory, BLSTM）與具有回路的條件隨機場（Conditional random field with loop, L-CRF）相結合的命名實體識別方法。首先，對輸入語句進行分詞和標注，使用Word2vec中的Skip-gram模型對標注語料進行預訓練，將其生成的字向量通過詞嵌入層轉化為字向量序列；然后，將字向量序列輸入BLSTM學習長期依賴信息；最后將句子表達輸入L-CRF獲取全局最優序列。實驗結果表明，該方法明顯優于其他命名實體識別方法，為數控機床設備的智能檢修與實時診斷任務打下了堅實的基礎。
-
關鍵詞:
- 數控機床 /
- 設備故障 /
- 雙向長短期記憶網絡 /
- 具有回路的條件隨機場 /
- 命名實體識別
Abstract: With the advent of intelligent manufacturing and big data, the Made in China 2025 Initiative and Industry 4.0 have been paying increasing attention to automation and intelligent industrial equipment. In the background of the present times, the complexity and intelligence of computer numerical control (CNC) machine tools have been continuously improved, and the types and descriptions of CNC machine tools’ faults have increased, presenting serious challenges to equipment maintenance and diagnosis of CNC machine tools. In order to provide guarantee for accurate fault diagnosis of CNC machine tools, and to prolong the service life of CNC machine tools, it is necessary to improve the performance of named entity recognition system. Accordingly, the named entity recognition in the equipment and faults field of CNC machine tools were studied, taking the historical examinations and repair records of CNC machine tools as the research object. After analyzing the characteristics of fault description in the historical examinations and repair records, a named entity recognition method was proposed based on the combination of bidirectional long short-term memory (BLSTM) and conditional random field with loop (L-CRF). The first step is to input a sentence and segment and label the input sentence. The annotation corpus is combined with the pre-trained generated word vector by using Skip-gram model in Word2vec, and the word vector is converted into a word vector sequence through the word embedding layer. In the second step, the word vector sequence is integrated into the BLSTM layer to learn long term dependency information. The final step is to input the sentence expression into the L-CRF layer to obtain the global optimal sequence. The experimental results show that the method is superior to other named entity recognition methods, which lays a solid foundation for the intelligent maintenance and the real-time diagnostic tasks of CNC machine tools. -
表 1 句子序列標注方法
Table 1. Sentence sequence labeling method
Sentence Labeling Sentence Labeling Sentence Labeling 發 B-Dev 的 O 牙 I-Fau 動 I-Dev 螺 B-Dev 機 I-Dev 釘 I-Dev 中 O 滑 B-Fau 
下載: 導出CSV
表 2 Word2vec的Skip-gram模型參數表
Table 2. Parameter list of Skip-gram model in Word2vec
Parameter Value Window size 10 Vector dimension 200 Minimum term frequency 5 Iterations 100
下載: 導出CSV
表 3 BLSTM-L-CRF模型參數表
Table 3. BLSTM-L-CRF model parameter table
Network layer Parameter Value BLSTM Learning rate 0.002 BatchSize 20 Iterations 100 Dropout 0.68
下載: 導出CSV
表 4 不同數據集在BLSTM-L-CRF模型中的識別結果
Table 4. Experiment result of BLSTM-L-CRF models in different data set
Date set Precision/% Recall/% F-measure/% People's daily corpus（1998） 83.07 83.40 83.23 MSRA corpus 82.23 80.35 81.28 Boson NLP corpus 79.45 80.18 79.81 CNC machine dataset 86.16 83.40 84.76
下載: 導出CSV
表 5 BLSTM-L-CRF與其他模型綜合性能對比
Table 5. Comparison of performance of BLSTM-L-CRF and other models
Model Precision/% Recall/% F-measure/% CRF 85.45 69.87 76.88 L-CRF 85.92 72.54 79.16 LSTM 78.90 77.84 78.37 BLSTM 80.71 79.00 79.85 CNN-LSTM 83.62 80.07 81.81 BLSTM-CRF 81.54 80.41 80.97 BLSTM-L-CRF 86.16 83.40 84.76
下載: 導出CSV
259luxu-164<th id="5nh9l"></th> <strike id="5nh9l"></strike> <th id="5nh9l"><noframes id="5nh9l"><th id="5nh9l"></th> <strike id="5nh9l"></strike> <th id="5nh9l"><noframes id="5nh9l"> <th id="5nh9l"></th> <strike id="5nh9l"><noframes id="5nh9l"><span id="5nh9l"></span> <span id="5nh9l"><noframes id="5nh9l"><span id="5nh9l"></span> <strike id="5nh9l"><noframes id="5nh9l"><strike id="5nh9l"></strike> <span id="5nh9l"><noframes id="5nh9l"> <span id="5nh9l"><noframes id="5nh9l"> <span id="5nh9l"></span> <span id="5nh9l"><video id="5nh9l"></video></span> <th id="5nh9l"><noframes id="5nh9l"><th id="5nh9l"></th> -
參考文獻
[1] Zeng J D, Wang T, Jia W J, et al. A survey on sensor-cloud. J Comput Res Dev, 2017, 54(5): 925 doi: 10.7544/issn1000-1239.2017.20160492曾建電, 王田, 賈維嘉, 等. 傳感云研究綜述. 計算機研究與發展, 2017, 54(5):925 doi: 10.7544/issn1000-1239.2017.20160492 [2] Wang T, Shen X W, Luo H, et al. Research progress of trusted sensor-cloud based on fog computing. J Commun, 2019, 40(3): 170 doi: 10.11959/j.issn.1000-436x.2019068王田, 沈雪微, 羅皓, 等. 基于霧計算的可信傳感云研究進展. 通信學報, 2019, 40(3):170 doi: 10.11959/j.issn.1000-436x.2019068 [3] Li J Y, Zhao Y K, Xue Z E, et al. A survey of model compression for deep neural networks. Chin J Eng, 2019, 41(10): 1229李江昀, 趙義凱, 薛卓爾, 等. 深度神經網絡模型壓縮綜述. 工程科學學報, 2019, 41(10):1229 [4] Bikel D M, Miller S, Schwartz R, et al. Nymble: a high-performance learning name-finder[J/OL]. arXiv preprint (1998-03-27)[2019-09-17]. https://arxiv.org/abs/cmp-lg/9803003 [5] Berger A L, Pietra V J D, Pietra S A D. A maximum entropy approach to natural language processing. Computat Linguist, 1996, 22(1): 39 [6] McCallum A, Li W. Early results for named entity recognition with conditional random fields, feature induction and web-enhanced lexicons // Proceedings of the Seventh Conference on Natural Language Learning at HLT-NAACL 2003-Volume 4. Edmonton, 2003: 188 [7] Yu H K, Zhang H P, Liu Q, et al. Chinese named entity recognition based on cascading hidden Markov model. J Commun, 2006, 27(2): 87 doi: 10.3321/j.issn:1000-436X.2006.02.013俞鴻魁, 張華平, 劉群, 等. 基于層疊隱馬爾可夫模型的中文命名實體識別. 通信學報, 2006, 27(2):87 doi: 10.3321/j.issn:1000-436X.2006.02.013 [8] He Y X, Luo C W, Hu B Y. Geographic entity recognition method based on CRF model and rules combination. Comput Appl Softw, 2015, 32(1): 179 doi: 10.3969/j.issn.1000-386x.2015.01.046何炎祥, 羅楚威, 胡彬堯. 基于CRF和規則相結合的地理命名實體識別方法. 計算機應用與軟件, 2015, 32(1):179 doi: 10.3969/j.issn.1000-386x.2015.01.046 [9] Wang L L, Aishan W, Maihemuti M, et al. A semi-supervised approach to Uyghur named entity recognition based on CRF. J Chin Inf Process, 2018, 32(11): 16 doi: 10.3969/j.issn.1003-0077.2018.11.003王路路, 艾山?吾買爾, 買合木提?買買提, 等. 基于CRF和半監督學習的維吾爾文命名實體識別. 中文信息學報, 2018, 32(11):16 doi: 10.3969/j.issn.1003-0077.2018.11.003 [10] Hochreiter S, Schmidhuber J. Long short-term memory. Neural Comput, 1997, 9(8): 1735 doi: 10.1162/neco.1997.9.8.1735 [11] Graves A, Schmidhuber J. Framewise phoneme classification with bidirectional LSTM and other neural network architectures. Neural Netw, 2005, 18(5-6): 602 doi: 10.1016/j.neunet.2005.06.042 [12] Yang H M, Li L, Yang R D, et al. Named entity recognition based on bidirectional long short-term memory combined with case report form. Chin J Tissue Eng Res, 2018, 22(20): 3237 doi: 10.3969/j.issn.2095-4344.0302楊紅梅, 李琳, 楊日東, 等. 基于雙向LSTM神經網絡電子病歷命名實體的識別模型. 中國組織工程研究, 2018, 22(20):3237 doi: 10.3969/j.issn.2095-4344.0302 [13] Lin B Y, Xu F, Luo Z Y, et al. Multi-channel BiLSTM-CRF model for emerging named entity recognition in social media // Proceedings of the 3rd Workshop on Noisy User-generated Text. Copenhagen, 2017: 160 [14] Bharadwaj A, Mortensen D, Dyer C, et al. Phonologically aware neural model for named entity recognition in low resource transfer settings // Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing. Austin, 2016: 1462 [15] Li C Y, Wu Y Z, Hu F H, et al. Packaging domain-based named entity recognition with multi-layer neural networks. Neu-roQuantology, 2018, 16(6): 564 [16] Yi S X, Yin H P, Zheng H Y. Public security event trigger identification based on Bidirectional LSTM. Chin J Eng, 2019, 41(9): 1201易士翔, 尹宏鵬, 鄭恒毅. 基于BiLSTM的公共安全事件觸發詞識別. 工程科學學報, 2019, 41(9):1201 [17] Chen Q Y, Cheng G, Li D, et al. Named entity recognition for mechanical design and manufacturing area. Comput Eng Appl, 2017, 53(20): 100 doi: 10.3778/j.issn.1002-8331.1604-0231陳秋瑗, 程光, 李迪, 等. 機械設計領域的命名實體識別研究. 計算機工程與應用, 2017, 53(20):100 doi: 10.3778/j.issn.1002-8331.1604-0231 [18] Zhao X N, Feng Z P. Fault diagnosis of rolling element bearing based on ensemble empirical mode decomposition and cross energy operator. Chin J Eng, 2015, 37(S1): 65趙曉寧, 馮志鵬. 基于集合經驗模式分解和交叉能量算子的滾動軸承故障診斷. 工程科學學報, 2015, 37(S1):65 [19] Zhang D, Feng Z P. Fault diagnosis of rolling bearings based on variational mode decomposition and calculus enhanced energy operator. Chin J Eng, 2016, 38(9): 1327張東, 馮志鵬. 基于變分模式分解和微積分增強能量算子的滾動軸承故障診斷. 工程科學學報, 2016, 38(9):1327 [20] Zhao W H, Zhang X, Lv D, et al. Technical status and strategies for domestic CNC machine tools. Aeron Manuf Technol, 2016, 59(9): 16趙萬華, 張星, 呂盾, 等. 國產數控機床的技術現狀與對策. 航空制造技術, 2016, 59(9):16 [21] Mikolov T, Sutskever I, Chen K, et al. Distributed representations of words and phrases and their compositionality // Advances in Neural Information Processing Systems 26(NIPS 2013). Lake Tahoe, 2013: 3111 [22] Lafferty J, McCallum A, Pereira F C N. Conditional random fields: Probabilistic models for segmenting and labeling sequence data // Proceedings of the 18th International Conference on Machine Learning 2001. Williamstown, 2001: 282 [23] Yuan S, Tang J, Gu X T. A summary of scholars' portrait techniques in the open internet. J Comput Res Dev, 2018, 55(9): 1903 doi: 10.7544/issn1000-1239.2018.20180139袁莎, 唐杰, 顧曉韜. 開放互聯網中的學者畫像技術綜述. 計算機研究與發展, 2018, 55(9):1903 doi: 10.7544/issn1000-1239.2018.20180139 [24] Zhu W Q, Liu Q. Conditional random fields with loop and its inference algorithm. Comput Eng Appl, 2008, 44(28): 180 doi: 10.3778/j.issn.1002-8331.2008.28.060朱文球, 劉強. 基于條件隨機域的上下文人類動作識別. 計算機工程與應用, 2008, 44(28):180 doi: 10.3778/j.issn.1002-8331.2008.28.060 [25] Lai S W. Word and document embeddings based on neural network approaches[J/OL]. arXiv preprint (2016-11-18)[2019-09-17]. https://arxiv.org/abs/1611.05962 -
點擊查看大圖
圖(4) / 表(5)
計量
- 文章訪問數: 2128
- HTML全文瀏覽量: 1921
- PDF下載量: 78
- 被引次數: 0
