Under complex driving scenarios, power-allocating mismatches among energy sources within hybrid propulsion system of fuel cell heavy trucks could lead to energy waste and lifetime degradation of vehicle. In order to improve vehicular economy and durability, a model predictive control-based energy management strategy with a novel computation-friendly frequency-decoupling optimization solver is proposed. The solver can quickly approximate the optimization problem, which helps to simplify the solution time for non-linear problems. Within the proposed solver, the ‘Haar’ wavelet transform is introduced to decouple the power demand change rate series at three wavelet levels. The sensitivity factor is then adjusted in real time by driving state recognition and used to calculate the output power of the fuel cell system to solve the energy management optimization problem based on model predictive control. The proposed strategy is verified by the Processor-in-the-Loop testing based on a RTU-BOX platform. Testing results indicated that under MPC-based framework, the proposed multi-mode frequency decoupling method showed a similar performance in power allocation versus a quadratic programming solver but saves up to 58.95% of computation time. Moreover, the proposed strategy outperforms two benchmark strategies in terms of total operating cost by 22.86% on average, indicating its effectiveness in terms of efficiency and durability enhancement.

As multimodal large models (MLLMs) continue to advance across challenging tasks, a key question emerges: \textbf{\textit{What essential capabilities are still missing? }}A critical aspect of human learning is continuous interaction with the environment -- not limited to language, but also involving multimodal understanding and generation.To move closer to human-level intelligence, models must similarly support \textbf{multi-turn}, \textbf{multimodal interaction}. In particular, they should comprehend interleaved multimodal contexts and respond coherently in ongoing exchanges.In this work, we present \textbf{an initial exploration} through the \textsc{InterMT} -- \textbf{the first preference dataset for \textit{multi-turn} multimodal interaction}, grounded in real human feedback. In this exploration, we particularly emphasize the importance of human oversight, introducing expert annotations to guide the process, motivated by the fact that current MLLMs lack such complex interactive capabilities. \textsc{InterMT} captures human preferences at both global and local levels into nine sub-dimensions, consists of 15.6k prompts, 52.6k multi-turn dialogue instances, and 32.4k human-labeled preference pairs. To compensate for the lack of capability for multi-modal understanding and generation, we introduce an agentic workflow that leverages tool-augmented MLLMs to construct multi-turn QA instances.To further this goal, we introduce \textsc{InterMT-Bench} to assess the ability ofMLLMs in assisting judges with multi-turn, multimodal tasks.We demonstrate the utility of \textsc{InterMT} through applications such as judge moderation and further reveal the \textit{multi-turn scaling law} of judge model.We hope the open-source of our data can help facilitate further research on aligning current MLLMs to the next step.