Combining software defined networking and SR(segment routing)can optimize network performance, but in large-scale dynamic networks, excessive link utilization at key nodes can lead to a surge in queue delays.To address this, a SROD-LC(segment routing optimization algorithm based on deep reinforcement learning and load centrality theory)was proposed.By quantifying the importance of network nodes using load centrality theory, key nodes are identified and their link load states are monitored;utilizing a multi-agent reinforcement learning framework, distributed deep reinforcement learning agents are deployed at key nodes, coordinating routing decisions through a shared reward mechanism to achieve proactive optimization of link loads.At the same time, leveraging the flexibility of SR, segment identifier lists are dynamically adjusted to quickly reroute partial traffic, reducing local link utilization and avoiding potential congestion.Simulation experiments based on real network topologies show that when the proportion of SR key nodes is in the range of 0.3~0.5, the SROD-LC algorithm exhibits significant optimization effects, reducing the network′s maximum link utilization by 21%~35% compared to baseline algorithms.

Power semiconductor modules are the core energy conversion units in power converters.By optimizing their design, the power density can be significantly enhanced.However, current design methods lack systematic summaries.To address this, a systematic summary across four levels(material, chip, packaging and drive)was presented.This included utilizing wide bandgap materials, enhancing chip structure, adopting advanced packaging and improving gate drive design.The underlying principles behind these methods for increasing power density were summarized, and classified and compared the existing research on improving the power density of converters based on power semiconductor module design.The primary challenges in current research were combed, and the future development trend was forecasted.

AI chips face on-chip memory limits in deep learning.Current optimization methods focus on static computation graphs, leaving room to improve memory efficiency for dynamic graphs.To overcome this limitation, a memory optimization framework for control-flow computation graphs was developed.The framework realized operator-level memory reuse within subgraphs and further achieved recursive reuse across subgraphs by exploiting control-flow characteristics.In addition, a ping-pong buffering strategy for weight data was introduced to mitigate the memory wall between on-chip and off-chip memory, thereby allowing overlapping of memory access and computation operations within subgraphs.Validation on the domestic LUNA AI chip has demonstrated that the proposed framework improves on-chip memory utilization by 5.9% compared with existing methods.Moreover, the strategy effectively alleviates the memory wall problem by reducing data transfer time between on-chip and off-chip memory, resulting in execution efficiency improvements of up to 29%.

To improve the design performance of long-range guided rockets, a multidisciplinary parametric model of long-range guided rockets was first established to achieve high-precision performance simulation of guided rockets.A sequence approximation optimization method based on an improved augmented radial basis function was proposed, which enhanced the generalization ability of the augmented radial basis function model through anisotropic techniques.Recursive evolution experimental design and fast cross-validation were used to improve the efficiency of approximation modeling, and an imprecise search strategy was applied for sequence sampling.The effectiveness of the proposed optimization method was verified through numerical examples.A sequence approximate optimization design of the long-range guided rocket was carried out, and the maximum range increase by 16.7% compared to before optimization while satisfying design constraints.

To further optimize the hardware offloading of collective communication based on the network interface card in the "Tianhe" network, and to support more types of collective communication algorithms and larger message sizes, the order-preserving triggering mechanism and data buffering method for collective communication hardware offloading was investigated.An order-preserving triggering mechanism for concurrent multitasking was proposed, which meets the desired semantics of collective communication and ensures the reproducibility of floating-point computation results.A dynamic network data buffering method based on Hash tables and pulsed credit flow control was proposed to alleviate the contradiction between limited hardware buffering resources and the high demand for buffering a large amount of network data from concurrent multitasking.Experimental results show that compared with software-based collective communication operations, this method can support the hardware offloading of various algorithms for several typical collective communication operations, with significant performance improvement. Meanwhile, the hardware implementation cost is low, especially with high utilization of buffering resources.

Particle transport simulations using stochastic methods face significant challenges on conventional von Neumann architectures, particularly due to random branching events and irregular memory access patterns.These limitations stem from the fundamental mismatch between probabilistic algorithms and deterministic computing paradigms.To bridge the gap between architecture and algorithms, a probabilistically tunable true random number generator was developed based on spintronic and ferroelectric devices.The physical randomness of spintronic devices was leveraged to provide a physical random source for the architecture, and the throughput of random bits was enhanced through optimized control logic and writing mechanisms.Next, programmable synapses were designed based on the memristive properties of ferroelectric devices, enabling nonvolatile continuous weight storage with tunable probabilities.The experimental results indicate that the proposed approach achieves performance improvements ranging from 171 to 1028 times compared to a general-purpose CPU when solving a sample transport problem.Furthermore, compared to existing spin-transfer torque magnetic tunnel junction based true random number generators, the developed method not only enables tunable probability random sampling but also achieves a throughput of 303 Mbit/s when generating uniformly distributed random sequences.

To achieve accurate and stable online identification of inductance parameters for PMSM(permanent magnet synchronous motor), an online inductance observation method based on virtual voltage vector excitation and current differential response was proposed, which required no additional test signal injection and was decoupled from rotor position, stator resistance, and permanent magnet flux linkage.By introducing the concept of a virtual voltage vector-oriented coordinate system, it was analytically derived and proven that the d-and q-axis inductances of a PMSM can be observed independently of the angular position in the conventional d-q synchronous reference frame.Building on this, the implementation procedure for extracting virtual voltage vectors and current differential information was discussed in detail, enabling non-intrusive inductance identification without any signal injection.The effectiveness and accuracy of the proposed method were validated by comparison with offline test procedures in IEEE standards.

For the new network communication challenges of efficient data interaction between components in open interactive environments, a novel C2N(computing and control network)was proposed.Aiming at the extreme requirements for efficiency, real-time performance, flexibility, and security, C2N adopts intelligent and simplified designs in protocol architecture, planning, application, and security design, providing high-performance and highly flexible basic network support for strong real-time collaborative fusion among heterogeneous resources.Based on a detailed investigation of relevant research work, key technologies of C2N were discussed, such as data link layer enhancement, remote direct memory access for sensor-controllers, and service-oriented sensing and control middleware.It also introduced the key technology research and test evaluation carried out by the network chip and system team of the National University of Defense Technology, and prospected future challenges and research directions to help China gain leading advantages in high-end equipment systems and innovative ecosystems.

In monocular vision-guided high-precision inter-platform pose measurement, existing methods require an accurate 3D model of the target platform and are unable to eliminate the impact of 3D model errors on pose measurement.To address this issue, iterative optimization was performed on the 3D model of the target platform and pose, and a new monocular vision measurement method was proposed.Specifically, the target platform′s 3D model was modeled using a set of sparse 3D keypoints.By leveraging multi-view geometric constraint information in sequential images, the sparse 3D keypoint set of the target and 6D pose were treated as parameters to be solved.An objective function was established to minimize object-space residuals, and through solving this optimization problem, iterative optimization of the sparse 3D keypoint set and pose was achieved.Additionally, a sliding window combined with a keyframe selection strategy was adopted to realize real-time and online high-precision monocular vision measurement.Experimental results demonstrate that, through iterative optimization of the sparse 3D keypoint set and pose, the proposed method achieves real-time, online high-precision monocular pose measurement under the condition of an inaccurate 3D model of target platform, while simultaneously improving the accuracy of the target′s 3D model.

Thermally-induced oscillatory rarefied gas flow inside a two-dimensional rectangular cavity was investigated.The effects of the Knudsen numbers and the oscillation frequency of lid temperature on the flow parameters were analyzed.The Shakhov model equation was solved numerically based on the mesoscopic approach in the near-wall region, and the macroscopic approach was adopted in the bulk flow region to reduce the computational cost.To close the numerical iteration procedure, the velocity distribution functions, served as the pseudo boundary between macroscopic and mesoscopic methods, were reconstructed using the high-order Hermite polynomials.Numerical simulations demonstrate that the temperature profile at the central vertical of the cavity predicted by the hybrid method is in good agreement with results from the mesoscopic method, with maximum error 0.23%.Besides, the computational memory cost can be saved up to about 69.91%.The hybrid approach is able to capture the nonlinear phenomenon in the thermally-induced oscillatory rarefied gas flow under high Kn numbers, where the horizontal velocity no longer obeys the law of periodic oscillating cosine function, and the rise time of the horizontal velocity is much longer than the fall time.The thickness of the viscous penetration layer and the disturbed region increases as the Kn number increases, and decreases as the St number increases.