基於深度神經網絡演算法之鋰離子電池電量狀態估計及硬體實現

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：102

、訪客IP：13.58.200.16

姓名

王億豐(Yi-Fong Wang) 查詢紙本館藏

畢業系所

電機工程學系

論文名稱

基於深度神經網絡演算法之鋰離子電池電量狀態估計及硬體實現
(Li-ion Battery State of Charge Estimation Based on Deep Neuron Network Algorithm and Hardware Implementation)

相關論文

★ 應用於2.5G/5GBASE-T乙太網路傳收機之高成本效益迴音消除器	★ 應用於IEEE 802.3bp車用乙太網路之硬決定與軟決定里德所羅門解碼器架構與電路設計
★ 適用於 10GBASE-T 及 IEEE 802.3bz 之高速低密度同位元檢查碼解碼器設計與實現	★ 基於蛙跳演算法及穩定性準則之高成本效益迴音消除器設計
★ 運用改良型混合蛙跳演算法設計之近端串音干擾消除器	★ 運用改良粒子群最佳化演算法之近端串擾消除器電路設計
★ 應用於多兆元網速乙太網路接收機類比迴音消除器之最小均方演算法電路設計	★ 光耦合隔離系統之接收端晶片電路設計與實現
★ 應用於光耦合隔離系統之發送端雜訊整形類比轉數位轉換器	★ 應用於數位視頻廣播系統之頻率合成器及3.1Ghz寬頻壓控震盪器
★ 地面數位電視廣播基頻接收器之載波同步設計	★ 適用於通訊系統之參數化數位訊號處理器核心
★ 以正交分頻多工系統之同步的高效能內插法技術	★ 正交分頻多工通訊中之盲目頻域等化器
★ 兆元位元率之平行化可適性決策回饋等化器設計與實作	★ 應用於數位視頻廣播系統中之自動增益放大器及接受端濾波器設計

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 (2026-8-1以後開放)

摘要(中)

在當今的現代社會中，電池技術扮演了關鍵角色，應用範圍廣泛。包括電動車、可再生能源儲能系統以及移動設備。這些應用領域的共同特點是它們都依賴於高效且可靠的能量供應，而電池的電量狀態(State of Charge, SOC)是實現這一目標的關鍵因素之一。
以電動車為例，了解電池的SOC對於確保車輛的正常運行和提供預測的駕駛範圍至關重要，以避免電力耗盡。此外，電池的SOC也影響充電的頻率和時間，因此它關乎用戶的便利性和充電基礎設施的需求。
電池管理系統(Battery Management System, BMS)是一項專門用於監督電池的技術。BMS包含了各種感應器，在確保安全且高效利用電池內儲存的能源方面發揮著至關重要的作用。BMS的主要功能之一是估算電池的SOC。然而，由於鋰離子電池的電化學特性，使得這項功能具有相當大的挑戰性。鋰離子電池的SOC無法通過傳統感測器直接測量，但可以從電流和電壓等參數間接估算出來。本文提出了一個用於預測電動車上鋰離子電池SOC的深度神經網絡(Deep Neuron Network, DNN)模型。其目標是提高電池SOC估算的準確性以及穩定性。
本文的研究過程以Python進行演算法驗證，接著使用Verilog HDL進行模擬。隨後通過FPGA、Design Compiler和IC Compiler進行電路驗證。

摘要(英)

Due to environmental pollution, electric vehicles (EVs) are trends in transportation. EVs produce zero tailpipe pollution and use less fuel than similar conventional vehicles. However, EVs still have some problems that should be addressed, such as lack of charging station, high cost of infrastructure, limited range or range anxiety. Consequently, the estimation of parameters related to electric vehicle batteries is a critical issue.
Battery Management System (BMS) is a technology designed for the supervision of batteries. BMS plays a pivotal role in ensuring the safe and efficient utilization of energy stored within batteries, containing various sensors. One of the primary tasks of the Battery Management System (BMS) involves the estimation of the battery state of charge (SOC). Nevertheless, this task poses a considerable challenge due to the electrochemical nature of lithium-ion batteries. The SOC of lithium-ion batteries cannot be directly measured through conventional sensors, but it can be indirectly estimated from parameters such as current and voltage. This paper proposes a deep neural network model (DNN) for predicting the Li-ion SOC of EVs. The objective is to enhance accuracy and stability of battery SOC estimation.
The research process of this paper begins with algorithm validation with Python, followed by simulation using Verilog HDL. Subsequently, circuit verification is conducted through FPGA, Design Compiler, and IC Compiler.

關鍵字(中)

★ 深度神經網路
★ 電池
★ 電量狀態

關鍵字(英)

★ DNN
★ battery
★ state of charge

論文目次

摘要 i
Abstract ii
致謝 iii
目錄 iv
圖目錄 vi
表目錄 ix
第一章緒論 1
1.1背景 1
1.2研究動機 2
1.3論文回顧 3
1.4論文貢獻 4
1.5論文架構 5
第二章電池管理系統介紹 3
2.1電池 6
2.2 SOC 6
2.3 BMS 8
2.4 電池充放電數據 9
2.4.1 CALCE電池數據 9
2.4.2 BMW i3實車電池數據 12
第三章深度神經網路架構與設計 16
3.1 DNN架構 16
3.1.1神經元 17
3.1.2激勵函數 18
3.1.3損失函數 21
3.1.4梯度下降法概要 22
3.1.5多層神經網路之梯度下降法 24
3.1.6優化器 27
3.2 DNN參數優化 32
3.2.1權重初始化 32
3.2.2超參數 32
3.2.3過度擬合(過度擬合) 33
3.3資料前處理 34
3.3.1移動平均法 34
3.3.2正規化 38
3.3.3輸入數據洗牌(Shuffle) 38
3.4 DNN訓練模型流程 39
第四章系統架構與模擬結果 41
4.1 DNN模型建構流程 41
4.2超參數模擬結果 42
4.3擴充輸入數據模擬結果 42
4.4 SOC估測架構及實現 43
4.5電動車實際資料之SOC估測結果 44
第五章電路架構與晶片實現 46
5.1電路設計流程 46
5.2硬體電路介紹 48
5.2.1 S形函數電路架構 48
5.2.2移動平均法電路架構 51
5.2.3 DNN電路架構 54
5.2.4其餘子電路架構 55
5.3模擬驗證 56
5.4晶片設計結果 60
5.4.1佈局圖 60
5.4.2模組分布以及面積佔比圖 62
5.4.3功率佔比圖 64
5.4.4錯誤覆蓋率 68
5.4.5關鍵路徑 69
5.4.6 LVS驗證 72
5.4.7 CHIP總結 72
第六章結論與未來展望 74
參考文獻 75

參考文獻

[1] Xiong, R., Cao, J., Yu, Q., He, H., & Sun, F., “Critical Review on the Battery State of Charge Estimation Methods for Electric Vehicles.” IEEE Trans. Ind. Vol.6, February 2018, pp. 1832-1843.
[2] How, D. N., Hannan, M. A., Lipu, M. S. H., Sahari, K. S., Ker, P. J., & Muttaqi, K. M., "State-of-Charge Estimation of Li-Ion Battery in Electric Vehicles: A Deep Neural Network Approach." IEEE Transactions on Industry Applications. Vol. 56, October 2020, pp. 5565-5574.
[3] Lipu, M. H., Hussain, A., Saad, M. H. M., Ayob, A., & Hannan, M. A.,"Improved recurrent NARX neural network model for state of charge estimation of lithium-ion battery using pso algorithm." in Proc. IEEE Symp. April 2018, pp. 354–359.
[4] Chemali, E., Kollmeyer, P. J., Preindl, M., & Emadi, A., "State-of-charge estimation of Li-ion batteries using deep neural networks: a machine learning approach." J Power Sources. Vol. 400, October 2018, pp. 242-255.
[5] Vidal, C., Malysz, P., Naguib, M., Emadi, A., & Kollmeyer, P. J., "Estimating battery state of charge using recurrent and non-recurrent neural networks. " Journal of Energy. Vol. 47, March 2022, Article 103660.
[6] Tisan, A., Oniga, S., Mic, D., & Buchman, A., "Digital Implementation of The Sigmoid Function for FPGA Circuits. " Acta Technica Napocensis. Vol. 50, no. 2, January 2009 pp. 15–20.
[7] Chiluveru, S. R., Chunarkar, S., Tripathy, M., & Kaushik, B. K., "Efficient hardware implementation of DNN-based speech enhancement algorithm with precise Sigmoid activation function. " IEEE Trans. Circuits Syst. II, Exp. Briefs. Vol. 68, no. 11, November 2021, pp. 3461–3465.
[8] Vaideeswaran, V., Bhuvanesh, S., & Devasena, M., "Battery Management Systems for Electric Vehicles using Lithium Ion Batteries." 2019 Innovations in Power and Advanced Computing Technologies. 2019, pp. 1-9.
[9] Uzair, M., Abbas, G., & Hosain, S., " Characteristics of Battery Management Systems of Electric Vehicles with Consideration of the Active and Passive Cell Balancing Process. "World Electr. Veh. J. Vol. 12, August 2021.
[10] CALCE電池數據, 取自https://calce.umd.edu/
[11] Argonne National Laboratory電池數據, 取自https://www.anl.gov/
[12] 張斐章、張麗秋，「類神經網路導論：原理與應用(第二版)(CD Inside)」，滄海書局，2015。
[13] Shiue, M.-T., Wang, Y.-F., Ou, Y.-C., Liu, B.-J., Liu, P.-L., "A Deep Neural Network Structure for Li-Ion Battery State of Charge Estimation and Hardware Implementation in Electric Vehicles. "International Journal of Industrial Electronics and Electrical Engineering (IJIEEE). Vol. 11, July 2023, pp. 1-5.
[14] Moons, B. Uytterhoeven, R. Dehaene, W. Verhelst, M. "14.5 Envision A 0.26-to-10TOPS/W subword parallel dynamic voltage accuracy frequency scalable Convolutional Neural Network processor in 28 nm FDSOI. ", IEEE, International Solid-State Circuits Conference (ISSCC), San Francisco, February 2017,pp. 246–247.
[15] Teng, C.-F.,Wu, A-Y., " A 7.8-13.6 pJ/b Ultra-Low Latency and Reconfigurable Neural Network-Assisted Polar Decoder With Multi-Code Length Support ", IEEE Transactions on Circuits and Systems I: Regular Papers, Mayday 2021.

指導教授

薛木添(Muh-Tian Shiue)

審核日期

2024-1-29

推文