As global climate change progresses and the “dual carbon” strategy advances, market-based carbon emission trading systems are of great theoretical and practical importance for green technology innovation. In this paper, A-share listed advanced manufacturing enterprises in pilot regions from 2010 to 2023 are taken as samples, and a multi-period difference-in-differences (DID) method is employed to probe into the mechanism by which this policy influences green technology innovation in China’s advanced manufacturing enterprises. Empirical analysis reveals that carbon emission trading exerts a remarkable promoting impact on green technology innovation in China’s advanced manufacturing enterprises. The study indicates that the policy’s influence on enterprises is indirect; specifically, government policies encourage enterprises to raise their R&D investment, thus facilitating green technology innovation to some degree. Moreover, carbon emission rights prices play a positive moderating role, which is vital for maintaining the policy’s incentive function in long-term green transition—within a reasonable range, carbon prices can enhance the policy’s promoting effect. In addition, enterprise-specific features like company size and asset-liability ratio have certain effects on enterprises’ green technology innovation behaviors. The research findings will offer a theoretical foundation and practical reference for optimizing China’s carbon market mechanism in advanced manufacturing and advancing the green transformation of China’s advanced manufacturing industry.

Ultra-short-term wind-power forecasting must handle both stable conditions and rapid ramp events, yet many architectures do not explicitly distinguish positive from negative ramp evidence. We evaluate whether that distinction improves multi-step predictions under a chronological wind-power forecasting protocol. RAA-Net combines a persistence-anchored baseline with a causal GRU and separate mappings for upward and downward ramps. A calibration-set blending rule substitutes persistence for selected low-volatility cases. We evaluated two turbines from the public SDWPF dataset at 40–160 min horizons, using chronological train, validation, calibration, and test partitions and three-seed neural-model comparisons against eight baselines. RAA-Net records the lowest reported RMSE at every horizon for both turbines. On the primary turbine, it also records the best reported MAE, $R^2$, nRMSE, and nMAE at each horizon. Component ablations quantify the effects of the directional module, anchor, and blending rule. These findings support direction-specific ramp representations for the evaluated turbines and protocol.