請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88332
標題: | 利用多特定運動軌跡估測希爾式肌肉骨骼模型之肌肉肌腱參數的最佳化方法 Optimization-based Estimation of Musculotendon Parameters in Hill-type Musculoskeletal Models using Multiple Specific Kinematic Trajectories |
作者: | 林易玄 Yi-Hsuan Lin |
指導教授: | 詹魁元 Kuei-Yuan Chan |
關鍵字: | 個人化肌肉骨骼模型,希爾式肌肉模型,肌肉肌腱參數評估,參數不可識別性,最佳化, subject-specific musculoskeletal model,Hill-type muscle model,musculotendon parameter estimation,parameter non-identifiability,optimization, |
出版年 : | 2023 |
學位: | 碩士 |
摘要: | 科技日益發展,電腦模擬與分析使得研究學者在生物力學、生理訊號等研究不再局限於臨床實驗的探討,透過軟體模擬亦可得到複雜的分析結果,像是神經訊號、肌肉力量、關節扭矩等資訊,不管是在臨床醫學、醫療復健還是運動科學等領域,皆帶來了前所未有的影響。在人體動作模擬與分析中,除了模擬過程的計算評估方法會影響準確度外,模型的選擇亦影響結果重大,使用通用模型雖能減去模型建立的繁雜步驟,但模擬結果並不能完全代表受試者本人,因此在個人化的模型建立上是必要的,不過同時也充滿了挑戰性。
本研究結合生物力學軟體 OpenSim 與數學計算軟體 MATLAB,以最佳化方法來估計肌肉骨骼模型中的肌肉肌腱參數,整體研究以運動軌跡預測任務作為核心,參數估計前利用敏感度分析結果來選擇欲執行任務,再透過多預測任務的執行,以預測軌跡與目標軌跡間的誤差來尋找欲評估肌肉之參數值,而評估完成得到的最佳模型則需經過模型驗證的考驗。一連串的研究方法流程以數個模擬案例來呈現,藉由普及的上肢肌肉骨骼模型來驗證方法的可行性與有效性,此外亦探討關於參數不可識別性問題,並證實多預測任務可有效地避免其影響。綜合上述,所提出之研究方法能有效評估肌肉骨骼模型中的肌肉肌腱參數,對於未來在個人化模型的建立上,將具有實質上的幫助。 The advancement of technology has significantly broadened the possibilities of research in fields such as biomechanics and physiological signals, surpassing the limitations of clinical experiments. Through computer simulation and analysis, researchers can also obtain complex results like neural signals, muscle force, and joint torque. This progress has had unprecedented implications in various domains, including clinical medicine, rehabilitation, and sports science. However, achieving accurate results in human motion simulation and analysis relies not only on the evaluation methods but also on the choice of models. While generic models simplify the model-building process, they fail to fully capture the unique characteristics of individual subjects. Thus, the development of subject-specific models is crucial, albeit challenging. In this study, the combination of biomechanics software, OpenSim, and mathematical computing software, MATLAB, is used to estimate musculotendon parameters in musculoskeletal models through optimization methods. The primary focus of the research centers around prediction tasks. Prior to parameter estimation, sensitivity analysis is performed to determine the desired tasks to be executed. Subsequently, multiple prediction tasks are executed to quantify the discrepancy between the predicted trajectories and the target trajectories, enabling the determination of parameter values for the evaluated muscles. Finally, the optimal models resulting from the evaluation process are subjected to model validation to ensure their accuracy. The methodology is demonstrated through several simulation cases using a widely used upper extremity musculoskeletal model, confirming the feasibility and effectiveness of the proposed methods. Moreover, the study investigates the issue of parameter non-identifiability and affirms that engaging in multiple prediction tasks is an effective means to circumvent its influence. In conclusion, the proposed methodology effectively estimates musculotendon parameters in musculoskeletal models, providing substantial support for future development of subject-specific models. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88332 |
DOI: | 10.6342/NTU202302001 |
全文授權: | 同意授權(全球公開) |
顯示於系所單位: | 機械工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-111-2.pdf | 14.98 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。