调度(生产过程)
计算机科学
风力发电
实时计算
风速
环境科学
作业车间调度
天气预报
时间序列
气象学
可靠性工程
作者
Yared Brhane Tadese,Andrés Castillo-Lermanda,Manohar Chamana,Bharath Kumar Dasari,Rajendra Shrestha,Bibek Dhakal,Md Shahin Munsi,Mostafa Mohammaadour,Olatunji Adeyanju,Stephen B. Bayne
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2026-02-09
卷期号:: 1-6
被引量:2
标识
DOI:10.1109/tpec67884.2026.11513189
摘要
High wind penetration makes power-system operations increasingly dependent on short-term wind power forecasts, yet false data injection attacks on wind measurements can corrupt these forecasts. This paper quantifies how such integrity attacks propagate from forecasting into day-ahead scheduling. Wind speed is first modeled with a lightweight autoregressive predictor trained on historical measurements, and the resulting forecasts are converted to wind power using a fitted wind power model curve. A black box scaling FDIA is then introduced by corrupting a fraction of the training window with an additive multiplicative bias whose magnitude follows a normal distribution; Both the attack duration K and the mean bias$\mu$are varied to represent different adversary strengths. The attacked wind power forecasts are embedded in a coupled unit commitment → economic dispatch (UC → ED) workflow: UC is solved using the compromised forecast to decide thermal on/off schedules, while ED is solved using the true wind realization to evaluate operational consequences. Case studies on the IEEE 39-bus system with an added$\text{5 5}$-MW wind plant assess the interaction between wind penetration and FDIA intensity. Results show that while increasing wind penetration lowers total operating cost, FDIA-induced forecast bias yields a consistent cost penalty that is largest at low wind levels, reaching about 1.3 % for$\mathrm{K}=40 \%$, and falling below 1 % at higher penetrations. These findings highlight the need for cyber-resilient wind forecasting and operations-aware defenses.
科研通智能强力驱动
Strongly Powered by AbleSci AI