臺灣博碩士論文加值系統

本論文模擬工業4.0的智慧工廠資料收集與分析，也就是架設物聯網感測器以收集智慧工廠的環境資料，並將收集到的資料進行分佈值變數的處理和分析。首先在物聯網感測器的部分，分別開發溫溼度感測器和灰塵感測器，並將它們布置在不同地點，並設定每隔一段時間會透過Wifi將資料上傳至雲端。收集到資料後需要經過前處理，再將資料轉換為分佈值變數。其次開發新的K-MEAP分群法，進而針對分佈值變數轉成分位數函數(quantile function)進行比對和計算的其Wasserstein 距離來建立新的分佈值變數資料間的相似度矩陣，藉由相似度矩陣進行可以指定生成特定群數的K-MEAP分群法，找到被集結成群的資料的關聯性。最後我們開發一種非監督式的學習-生成對抗網路(Generative adversarial network)來學習真實的物聯網感測器資料，並開發真實分佈值變數資料為基底來生成新的資料，以克服分群資料不足時的問題。

This thesis simulated data collection and analysis for smart plant of industry 4.0. That is, the design of the internet of things (IoT) sensor to collect environment data of smart plant, and the collection of data can be transformed into distributional-valued variable to do analysis. Firstly, we design IoT sensor are arranged in different places and set to upload data through Wifi every once in a while to the cloud. After collecting data, data need to be pre-processing, and then convert data to the distribution-valued variable. Secondly, a new K-meap clustering approach is proposed. Then similarity matrix of the data can be established by using Wasserstein distance of the comparison and calculation of the quantile function on the distribution-value variable. And the K-meap clustering method for generating the specific number of groups may be specified by similarity matrix. Clustering can help to find the relationship between the data which are clustered in a group. Finally, we also proposed a non-supervised approach– Generative adversarial network (GAN) to learn real IoT sensor data from smart plant. GAN can simulate the distribution-value
ii
variable of real IoT sensor data to generate some new data that can overcome insufficient data in clustering.

摘要……i
Extended Abstract……ii
Acknowledgements……iv
Table of contents……v
List of table……vii
List of figures……viii
Chapter 1 Introduction……1
1.1 Review of IoT……1
1.2 Review of Clustering……1
1.3 Review of GAN……3
1.4 Review of Symbolic Data……5
1.5 Motivation and Contributions……6
Chapter 2 Mathematical Background……8
2.1 Multiple Exemplars Affinity Propagation with Specified K Clusters……8
2.2 Wasserstein Distance……13
2.3 Generative Adversarial Network……15
Chapter 3 Data collection from Internet of Things……19
3.1 System Architecture……19
3.3 Upload Results……22
Chapter 4 Multiple Exemplars Affinity Propagation with Specified K Clusters with
Wasserstein Distance……25
4.1 Detail of K-meap……25
4.2 Results……28
Chapter 5 Generative Adversarial Network with Wasserstein Distance……60
vi
5.1 Wasserstein GAN……60
5.2 Results……63
Chapter 6 Conclusion……64
References……65
Extend Abstract……68

[1] Frey.B.J. (2007), “Clustering by passing messages between data points,” Science. pp. 972-976. [2] Wang.C.D., Lai.J.H., Suen.C.Y., and Zhu.J.Y.(2013), “Multi-exemplar affinity propagation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, pp. 2223-2237. [3] Zhang.X.L., Wang.W., Norvag.K., Sebag.M.(2011), “K-AP: Generating Specified K Clusters by Efficient Affinity Propagation,” 2010 IEEE International Conference on Data Mining. [4] Wang.Y. and Chen.L.(2015), “K-MEAP: Multiple Exemplars Affinity Propagation With Specified K Clusters,” IEEE Transactions on Neural Networks and Learning System, vol. 27, pp. 2670-2682. [5] Goodfellow.L.J., Pouget-Abadie.J., Mirza.M. and Xu.B.(2014), “Generative Adversarial Networks,” arXiv:1406.2661, Cornell University. [6] Billard.L.(2006), “Symbolic data analysis: what is it?,” Compstat 2006 - Proceedings in Computational Statistics pp 261-269. [7] Mahalanobis.P.C.(1936), “On the generalised distance in statistics,” India,. [8] Bock.H.H. and Diday.E.(2000), “Analysis of Symbolic Data,” Exploratory Methods for Extracting Statistical Information from Complex Data,” Springer. [9] Bertrand.P. and Goupil.F.(2000), “Descriptive Statistics for Symbolic Data,” Analysis of Symbolic Data, pp 106-124.
[10] Billard.L.(2007), “Dependencies and Variation, Components of Symbolic Interval-Valued Data,” Selected Contributions in Data Analysis and Classification, pp 3-12. [11] Carvalho.F.D and Lechevallier.Y.(2009), “Dynamic Clustering of Interval-Valued Data Based on Adaptive Quadratic Distances,” IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans pp. 1295-1306. [12] Arjovsky.M., Chintala.S. and Bottou.L.(2017), “Wasserstein GAN,” arXiv:1701.07875, Cornell University. [13] https://thingspeak.com/, Cloud service platform. [14] https://erdb.epa.gov.tw, Environmental Protection Administration environmental resource database. [15] Li.Y.C.(2018), “Improved AP Clustering and Its Application on Distributional-Valued Variable,” National Formosa University. [16] Jain.A.K.(2010), “Data clustering: 50 years beyond K-means,” Pattern Recognit. Lett., vol. 31, no. 8, pp. 651-666. [17] Xie.B., Wang.M., and Tao.D.(2011), “Toward the optimization of normalized graph Laplacian,” IEEE Trans. Neural Netw., vol. 22, no. 4, pp. 660-666. [18] He.Z., Xie.S., Zdunek.R., Zhou.G., and Cichocki.A.(2011), “Symmetric nonnegative matrix factorization: Algorithms and applications to probabilistic clustering,” IEEE Trans. Neural Netw, vol. 22, no. 12, pp. 2117-2131. [19] Huang.X., Ye.Y., and Zhang.Z.(2014), “Extensions of kmeans-type algorithms: A new clustering framework by integrating intracluster compactness and intercluster separation,” IEEE Trans. Neural Netw. Learn. Syst., vol. 25, no. 8, pp. 1433–1446.
[20] Jing.L., Ng.M.K., and Zeng.T.(2013), “Dictionary learning-based subspace structure identification in spectral clustering,” IEEE Trans. Neural Netw. Learn. Syst., vol. 24, no. 8, pp. 1188–1199. [21] Hou.C.,Nie.F.,Yi.D., and Tao.D.(2015), “Discriminative embedded clustering: A framework for grouping high-dimensional data,” IEEE Trans. Neural Netw. Learn. Syst., vol. 26, no. 6, pp. 1287–1299. [22] Ta¸sdemir.K., Milenov.P., and Tapsall.B.(2011), “Topology-based hierarchical clustering of self-organizing maps,” IEEE Trans. Neural Netw., vol. 22, no. 3, pp. 474–485. [23] Kingma.D.P. and Welling.M.(2014), “Auto-Encoding Variational Bayes,” In proceedings of the International Conference on Learning Representations (ICLR). [24] Radford.A., Metz.L. and Chintala.S.(2016), “Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks,” arXiv:1511.06434, Cornell University. [25] Zhu.J.Y, Park.T., Isola.P., and Efros.A.A.(2018), “Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks,” arXiv:1703.10593, Cornell University. [26] Smith.E. and Meger.D.(2017), “Improved Adversarial Systems for 3D Object Generation and Reconstruction,” arXiv:1707.09557, Cornell University. [27] Wang.Z.Q.(2018), “Study on Distributional-Valued Variable Analysis Using Machine Learning Approached,” National Formosa University. [28] Krizhevsky.A., Sutskever.I.(2012), and Geoffrey E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” Advances in Neural Information Processing System 25 (NIPS).