Aggregator optimal bidding method based on deep reinforcement learning and hybrid intelligent architecture

By employing an aggregator-based optimal bidding method based on deep reinforcement learning and a hybrid intelligent architecture, the problems of low accuracy and poor robustness of bidding strategies in quasi-linear demand response are addressed. This method achieves efficient decision-making and high returns in abnormal scenarios, thereby improving the accuracy and robustness of bidding strategies.

CN122367594APending Publication Date: 2026-07-10SHANGHAI UNIVERSITY OF ELECTRIC POWER
0 Cites 0 Cited by

Patent Information

Application Number
CN202610485451.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-04-14
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing aggregator bidding strategies have low accuracy and poor robustness in quasi-linear demand response, making it difficult to cope with real-time market electricity price fluctuations and user response uncertainties. This leads to a disconnect between bidding schemes and actual demand, resulting in problems such as capacity surplus or shortage, lower-than-expected returns, and low response execution rates.

Method used

We adopt an optimal bidding method for aggregators based on deep reinforcement learning and a hybrid intelligent architecture. By constructing a hybrid intelligent architecture, we utilize the improved multi-agent deep reinforcement learning algorithm (MATD3) of VMD-FFCM-LSTM-iTransformer, combined with consumer psychology models and uncertainty modeling, to design a human-machine collaborative decision-making mechanism. This enables collaboration between machine-led routine decision-making and human intervention in abnormal scenarios, thereby optimizing the bidding strategy.

Benefits of technology

It improves the accuracy and robustness of bidding strategies, enhances adaptability in abnormal scenarios, increases profit and response completion, optimizes decision-making efficiency and response speed, and meets the power grid response time requirements.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses an aggregator optimal bidding method based on deep reinforcement learning and a hybrid intelligent architecture, and belongs to the technical field of demand response scheduling and intelligent decision control. According to the method, aiming at the characteristics that load adjustment is constrained and user response is uncertain in quasi-linear demand response, an aggregator bidding model is established through load curve similarity measurement and quasi-linear constraint modeling, and a bidding strategy is trained and updated by using multi-agent deep reinforcement learning, so that the user response is more accurate. And generating a bidding strategy of demand response market bidding. And on this basis, a man-machine collaborative decision fusion mechanism is introduced to realize the output of the bidding strategy in conventional and abnormal scenes. The method has the advantages that an aggregator can be assisted in formulating a feasible demand response market bidding scheme under a quasi-linear demand response mechanism, the economical efficiency and execution reliability of bidding decision making are improved, reasonable allocation of demand response resources is promoted, and technical support is provided for stable operation of a demand response market and resource optimization configuration.
Need to check novelty before this filing date? Find Prior Art