建立太陽能電廠清洗決策以提升清洗成本效益

NCU Institutional Repository > 管理學院 > 工業管理研究所碩士在職專班 > 博碩士論文 > Item 987654321/97278

請使用永久網址來引用或連結此文件: https://ir.lib.ncu.edu.tw/handle/987654321/97278

題名:	建立太陽能電廠清洗決策以提升清洗成本效益
作者:	吳達秉;Da-Bing
貢獻者:	工業管理研究所在職專班
關鍵詞:	太陽能模組;髒汙效應;發電損失;PVsyst 模擬;清洗決策;PV modules;soiling effect;power loss;PVsyst simulation;cleaning strategy
日期:	2025-07-09
上傳時間:	2025-10-17 11:04:58 (UTC+8)
出版者:	國立中央大學
摘要:	本研究針對太陽能電廠模組之髒汙問題，建構一套以發電損失為核心，並納入區域差異性與經濟效益考量的清洗決策模型。相較於傳統固定週期清洗方式，本模型可因應模組污染的空間不均性，提升清洗作業的資源配置效率，進而強化發電效益與清洗成本效益。研究方法首先透過 PVsyst 軟體模擬模組於潔淨狀態下之發電性能表現，結合實測溫度與模組衰退條件進行日尺度修正，建立符合實際運行條件之模擬基準。再藉由實測性能比與模擬結果之比對，量化模組髒汙所致的發電損失，並推估每日經濟影響，進一步發展設備層級之清洗時機判斷邏輯。相關資料涵蓋 2023 年 11 月至 2025 年 5 月之模擬與實測記錄，並結合現地觀察進行模組污染型態分類與區域策略設計。於 2024 年 12 月至 2025 年 5 月期間共觀測 13 次自然降雨事件，結果指出當日雨量達 2.5 L/m² 以上時雖具一定清洗潛力，惟難以達到人工清洗效果，特別對於鳥糞等附著型污染幾乎無法產生實質改善。模組區域位置與污染類型對發電損失具有明確影響，反映污染分布之高度異質性，亦突顯差異化管理之必要。清洗策略方面，本研究以單台變流器清洗成本作為經濟閾值，實測期間採取分批清洗方式，針對六台設備執行清潔作業，相較全區清洗方案可降低總支出 31.9%。另三台設備至觀測期末仍未達清洗門檻，驗證延後處理在經濟性上具可行性。整體結果顯示，結合模擬基準、損失量化與污染辨識條件所建構之清洗決策邏輯，能有效提升判斷精度與成本效益，適用於污染分布不均之太陽能系統場域。 ;This study addresses the issue of performance degradation in photovoltaic modules caused by surface soiling and proposes a cleaning decision-making model centered on power loss, incorporating spatial pollution variability and economic efficiency. Compared to conventional fixed-interval cleaning strategies, the proposed model accounts for the uneven distribution of soiling across module areas, thereby enhancing the efficiency of cleaning resource allocation and ultimately improving both energy yield and cleaning cost-effectiveness. The methodology begins by using PVsyst software to simulate the ideal performance of clean modules, integrating measured module temperature and degradation conditions to generate a temperature-corrected daily baseline. By comparing simulated performance ratios with actual measured values, this study successfully quantifies power losses caused by soiling and estimates daily economic impact. A cleaning threshold model is then developed at the inverter level, supported by observational classification of pollution types across module zones. The analysis is based on daily-resolution data collected from November 2023 to May 2025. Between December 2024 and May 2025, thirteen natural rainfall events were recorded. Analysis shows that rainfall above 2.5 L/m² may provide partial cleaning effects but generally fails to achieve the efficacy of manual cleaning, particularly in zones affected by bird droppings or other adhesive contaminants. Observations further confirm that pollution type and module location significantly influence accumulated power loss, highlighting the need for differentiated regional management strategies. For cleaning strategy evaluation, this study adopts the cleaning cost per inverter as the economic threshold. During the observation period, a selective cleaning approach was applied, targeting six inverters, which reduced total costs by 31.9% compared to a full-system cleaning strategy. The remaining three inverters did not reach the cleaning threshold by the end of the study, validating the cost-effectiveness of deferred cleaning. Overall, the proposed strategy—based on simulated baselines, quantified soiling losses, and pollution classification—demonstrates improved accuracy and economic benefit, making it applicable to PV systems with non-uniform pollution distribution.
顯示於類別:	[工業管理研究所碩士在職專班 ] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	11	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....