為了應對氣候變化并減少碳排放，當前許多國家都依賴于風能等可再生能源。然而，風電產業也是一個資本密集型行業，這意味著風機和其他資產一樣，需要定期進行“后運維“（O&M）以防止發生意外故障。風機的平均壽命為20到25年，因此歐洲和美國在2000年之前安裝的許多陸上風機，已經達到了當初設計壽命的終點（EOD）。風電場運營商需要找到一種利潤率更高的經營方式，否則他們將面臨資產報銷和投資失敗的風險。

最新數字技術的普及為運營商提供了延長風機使用壽命和優化發電量的機會。根據WindEurope的數據，在即將達到設計壽命的22GW風電中，有18GW可以應用風機壽命延長項目（LTE）。

圖片來源：網絡

不作為的代價高昂——后運維需要更加積極主動

傳統的后運維活動主要集中在日常操作和定期維護上，這種方法依賴于被動決策，寄希望于一切都能正常運行。在最壞的情況下，未被及時發現的問題則會帶來價格高昂的補救措施。事實上，后運維相關的支出已經占據了風電項目總成本的10%到20%。此外，許多風電場的所有者還簽訂了昂貴的后運維合同，以彌補運維技能的不足。盡管像SCADA（監督控制和數據采集）這樣的監控系統已經在風電場運行了一段時間，但我們仍然缺乏在問題出現的早期階段做出精準決策的能力。從數據中提取信息并解釋結果以提早發現故障至關重要，而數字化的后運維則可以實現這一點，賦予了風電運營商對風機的更多控制權。

利用數字孿生（digital twins）和AI的力量

數字化能力通過數據來應對后運維業務的相關挑戰，而不是用數據來定義這些挑戰。傳感和捕獲準確的原始數據是關鍵的第一步。許多在過去二十年中建成的風電場并沒有配備現代風機中的傳感器。因此，我們可以通過雷達的聲學特征、無人機圖像識別等技術實現非侵入性的計算傳感，促進多傳感器的融合，從而豐富不同組件的數據采集。接下來，則是采用數據管理和轉換技術來清洗和合并數據，使其適合分析引擎進行下一步的使用。

數據工程只是整個藍圖的一部分。基于數據驅動的組件和相關過程的數字孿生（digital twins）系統將結合預測分析，并將以此生成風機性能的解析以支持主動決策和糾正維護。當結合人工智能和機器學習時，這項分析還可以從歷史數據中挖掘出新的參數和先行指標。通過先進算法跟蹤和分析這些數據，我們可以精準了解到關鍵風機組件的當前和未來狀態。然而，如果數據和分析沒有以正確的格式及時提供給操作員、工程師或業務相關者，那么這些分析將毫無意義。如今，許多致力于后運維的企業試圖將分析結果與ERP（企業資源計劃）系統連接，來實現“服務化”（servitization），這確保了信息的可操作性，也能使分析結果及時得到反饋。

向可靠的低成本風能邁進

風能發電量目前占全球總發電量的4.4%，并預計到2030年將增加到20%。隨著政府補貼的減少，風電場必須找到新的方法來降低成本并保持競爭力。數字技術一定是未來的發展方向，它將減少停機時間、降低后運維成本、提高風機的運營效率，從而最終實現低成本的清潔能源產出。

擴博智能，風電業務在全球：

擴博智能已與丹麥、巴西、美國、加拿大、越南、緬甸、泰國、希臘、羅馬尼亞、葡萄牙、意大利等不同地區、不同規模的風電廠達成合作，共覆蓋29個國家及地區，全球累計巡檢80,000+臺次，并創下最短巡檢時間15分鐘、單日陸上巡檢記錄31臺、單日海上巡檢記錄18臺等記錄。目前，擴博智能向包括運營商、主機商、葉片制造商、第三方服務提供商等提供全方位的解決方案和服務。在全球各大洲與各主要區域，我們均有專業的智能巡檢團隊可為您提供一站式服務。

英文原文

Wind Energy Gets a Much-needed Boost with Digital O&M

ORONO, Maine (AP) — As waves grew and gusts increased, a wind turbine bobbed gently, its blades spinning with a gentle woosh. The tempest reached a crescendo with little drama other than splashing water.

Many countries are relying on wind energy among other renewable resources to tackle climate change and reduce carbon emissions. However, it’s a capital-intensive sector and wind turbines like any other asset require regular operations and maintenance (O&M) to prevent unplanned breakdowns and repairs. The average shelf life of a wind turbine is 20 to 25 years. This means that many onshore wind farms in Europe and the U.S. installed before the year 2000 have already arrived at the end of design (EOD) life. Wind farm operators will need to find more profitable ways to run their business, or risk decommissioning wind assets and writing off the investment.

The proliferation of new digital technologies has given an opportunity for operators to increase the useful life of wind turbines and optimize the power yield. According to WindEurope, out of 22GW of wind power that is coming to its EOD life, 18GW will be eligible for lifetime extension (LTE) projects.

The cost of inaction is high—O&M needs to become more proactive

Traditional O&M activities centered around routine operations and scheduled maintenance, an approach that relied on reactive decision making with the hope that everything was working fine. In worst case scenarios, undetected problems would result in expensive, corrective actions. In fact, O&M accounts for approximately 10 to 20% of the total cost of energy for a wind project. In addition, many wind farm owners signed expensive maintenance contracts to fill the O&M skill gap. While monitoring systems such as SCADA (supervisory control and data acquisition) have been used for a while on wind farms, what’s missing is the sophistication needed to arrive at insightful decisions during the early stages of a problem. The ability to extract information from data and interpret outcomes for early detection of failures is critical. Digital O&M has now made this possible, giving wind operators more control over turbine performance.

Utilizing the power of data with digital twins and AI

Digital technologies use data as a vehicle to tackle O&M business challenges, as opposed to using it to define those challenges. Sensing and capturing accurate raw data is a critical first step. Many windfarms that were commissioned in the last two decades are not equipped with sensors found in modern wind turbines. Unobtrusive computational sensing through radar-based acoustic signature, drone-based imagery, and other related technologies enables multi-sensor fusion for enriching data capture across different components. The next step is employing data management and transformation techniques, including cleansing and merging the data to make it fit for consumption by analytical engines. 

Data engineering is just one piece of the puzzle. Data-driven digital twins of the components and its adjoining processes combined with predictive analytics will generate insights on performance to support proactive decision-making and corrective maintenance. When combined with artificial intelligence and machine learning, analytics can mine new parameters and lead indicators from historical data. This data can be tracked and analyzed using advanced algorithms to understand the current and future state of critical wind turbine components.

But analytics will be irrelevant, if the data and insights are not available to the operator, engineer or business stakeholder in the right format, in a timely manner. Many enterprises with a broader O&M vision are using ‘servitization’ by linking analytics with ERP (enterprise resource planning) systems. This ensures a process-driven approach in which information is actionable, and it triggers the right response.

Moving towards reliable, low-cost wind energy

Wind power represents 4.4% of the total generated power and is likely to increase up to 20% by 2030. With governments reducing subsidies, wind farms have to find new ways to cut costs and stay competitive. Digital technologies are the way forward. It will reduce downtime, cut O&M costs and improve the operational efficiency of wind turbines. The result is increased clean energy production at low costs.