Objective Aiming at the problems of fragmented mechanisms, regional limitations, and insufficient targeting of prediction and regulation in the research on the impact of extreme temperature events on electrical loads, this paper systematically sorts out the impact laws, prediction methods, and regulation strategies of heatwaves and cold waves on multiple types of electrical loads, so as to provide support for constructing a resilient power system and ensuring energy security.

Method Using literature review and systematic analysis, this paper defined the concepts of heatwaves (daily maximum temperature ≥35 ℃ and duration ≥3 days), cold waves (24-hour temperature drop ≥8 ℃ and minimum temperature ≤4 ℃), and three types of loads (residential, industrial, and commercial). It integrated typical cases such as the 2022 Sichuan-Chongqing heatwave and the 2021 Texas (USA) cold wave, reviews prediction models by categorizing them into traditional statistical models, machine learning models, and deep learning models, and proposed regulation strategies combined with load characteristics while analyzing research gaps.

Result The study finds that the impact of extreme temperature events on electrical loads is essentially a chain process of "meteorological factors - user energy consumption behavior - load response" with significant differences: under heatwaves, residential and commercial air-conditioning loads surge synergistically, the load curve shows a "single-peak high platform" pattern, and the total load increases by 2%～5% for every 1 ℃ rise in temperature; under cold waves, residential heating loads dominate the sudden change, the load curve shows a "high valley, short flat peak" pattern, and the total load increases by 3%～6% for every 1 ℃ drop in temperature. Empirical evidence shows that the peak load increases by 15%～40% compared with the normal state under extreme temperatures, with higher heat sensitivity in low-latitude regions and higher cold sensitivity in high-latitude regions. Among prediction models, deep learning models such as Transformer have the highest accuracy, while traditional statistical models have strong interpretability but struggle to cope with the nonlinear changes of loads. For heatwaves, air-conditioning loads need to be suppressed through "time-of-use pricing + intelligent temperature control"; for cold waves, heating demand needs to be guaranteed by "heat storage heating + cross-regional coordination", and the implementation of regulation requires support from policy mechanisms.

Conclusion The impact of extreme temperature events on multiple types of electrical loads has significant differentiated and regional characteristics. Deep learning models are the preferred direction for load prediction, and differentiated regulation and policy guarantees can effectively stabilize load fluctuations. In the future, it is necessary to focus on breaking through research gaps such as micro load heterogeneity analysis, small-sample sudden change prediction, and cross-regional coordinated regulation to further improve the management capacity of electrical loads under extreme temperature events and provide support for the construction of a resilient power system.