A lumped kicker magnet system is being developed as part of the electron beam switchyard of Shanghai High repetition rate X-ray free electron laser (XFEL) and Extreme Light (SHINE) at Shanghai Advanced Research Institute (SARI), Chinese Academy of Sciences (CAS). SHINE is a linac-based facility with an energy of 8 GeV and a repletion rate of 1 MHz, aiming for a multi-beamline mode of operation to improve the efficiency of the user experiments. The kicker is located in a tunnel and appears to be a critical element of the defection system and determines the stability performance of the beamlines. To allow autonomous and reliable operation in a harsh environment, a dedicated control system based on the Experimental Physics and Industrial Control System (EPICS) is developed. In this work, status acquisition, control scheme, and alarm handlers are programmed via a programmable logic control (PLC) based human machine interface (HMI). A compact and modular design of the controller, including four slots (trigger, timing, external, and power supply interlock) and HMI are integrated to allow easy maintenance and operation. In addition, a waveform acquisition framework based on LabVIEW is built to ensure the intuitive demonstration of the kicker waveforms. This enables the waveform real-time diagnosis and fast-tracking if any distortions occur in the beamlines. Detailed design – overall concept, achieved performance, and initial tested results - of two modular configurations of the controllers are provided for further demonstration.

Objective This paper is aim to provide a digital evaluation method based on Multiple Electrical Performance Parameters (MEPP), to evaluate the health of the medical devices scientifically. Methods Firstly, the circuit characteristics of the medical devices were analyzed to establish the theoretical basis and evaluation index for the evaluation system. Then, based on the experience of the health status of medical equipment in the whole life cycle, the health status of the medical devices is correlated with the electrical performance parameters in this paper. Finally, this paper selected 4 different brands of monitors. The maintenance records of our hospital as result were to gain the stability of each brand, and calculate the stability value of 4 devices through the electrical performance. The health status is scored and evaluated at last. Results The experimental result shows that the scores of 4 devices are 83.9, 88.1, 89.5, 81 respectively. After checking and comparing the maintenance frequency, it is found that it conforms to the order of stability. And the experiment results of the same brand monitor also show that the stability is consistent. Conclusion Aiming at the electrical performance parameters of the monitors, this study provides a scientific evaluation for the health status of the monitors and similar medical equipment. It’s a basis for new computer technology application to traditional circuit detection.

Over the last decades, halide materials, represented by organic-inorganic hybrid perovskites, have emerged with great potential for the next-generation photonic and optoelectronic devices. In particular, their excellent photonic and optoelectronic properties can be further engineered by chemical compositions, material sizes, device structures, external fields and so on. More recently, tuning of dimensionality plays a significant role in promoting the properties of organic-inorganic hybrid perovskites. Originating from shape diversity and quantum confinement effect, the emerging different forms of perovskites, such as nanocrystals, nanowires/nanorods, nanoplatelets/nanosheets and ultra-thin films, have heralded new opportunities for novel physical phenomenon and high-performance devices. Particularly, the ultra-thin organic-inorganic hybrid perovskites, which combing the advantages of two-dimensional (2D) materials and organic-inorganic hybrid perovskites, are attracting more interest thanks to the exotic and unique properties. Considering the ultra-smooth surface and interface, ultra-thin organic-inorganic hybrid perovskites based heterostructures with van der Waals contacts can be artificially designed without considering the lattice matching. More novel properties will emerge and can be easily engineered by the strong interface coupling in ultra-thin organic-inorganic hybrid perovskites stacks. By flexible design of heterostructures based on ultra-thin organic-inorganic hybrid perovskites and their integration with diverse 2D materials, it offers many possibilities to discover new properties and device functionalities within the 2D family in the fields of green energy, sensor, integrated circuit and so on. In this presentation, I will introduce my achievements in the fields of ultrathin 2D organic-inorganic hybrid perovskites, including the materials design and preparations, heterostructure constructions, optical properties explorations and optoelectronic device designs.

Multimodal sentiment analysis has considerable potential and importance, but most of the existing methods face difficulties in adapting to complex environments and scenarios. To address these challenges, we propose a self-supervised feature fusion multimodal sentiment analysis model based on the Mamba architecture. The model abandons the traditional Transformer architecture and adopts the Mamba model, leveraging its advantages of efficiently processing long sequences and low computational complexity. Firstly, we utilize the Mamba model to extract features from text, audio, and visual modalities. To further optimize the feature fusion process, we introduce an improved cross-modal attention fusion module based on Mamba, which intelligently selects and fuses key information from different modalities using the unique selective state space model structure of Mamba when processing different types of data, thereby enhancing the accuracy of sentiment analysis. Additionally, we propose a novel feature fusion strategy that combines the powerful representation capabilities of the Mamba model with the cross-modal attention mechanism of the Transformer to more effectively integrate data features extracted by pre-trained models. To evaluate the performance of the proposed model, experiments were conducted on three public datasets: CMU-MOSI, CMU-MOSEI, and IEMOCAP. Compared with previous multimodal sentiment analysis models(such as MulT, ICCN, etc.), the experimental results demonstrate that our proposed model exhibits significant improvements in the evaluation metrics, thereby validating the effectiveness and robustness of our methodology.

Energy storage systems based on batteries (e.g., Lead-acid and Lithium-ion batteries) are widely explored in cooperation with renewable energies such as photovoltaic (PV) power generation to achieve a carbon-neutral sustainable society. Among various batteries, Nickel-Zinc (NiZn) battery could be a suitable candidate for storing renewable energies due to their non-flammable materials and lower cost as compared to other battery technologies like Lithium-ion. However, the voltage window of a NiZn battery is usually smaller (around 1.3V to 1.9V) which is incompatible with power systems running at 48 Vdc or even 400 Vdc bus. To address this, series-connected cells are usually used to provide a higher bus voltage and thus a larger system capacity, which may however suffer from voltage unbalance as the number of the series-connected cells increases.

This paper proposes a switched-capacitor voltage equalizer for 30 series-connected NiZn battery cells. Due to the low-voltage feature of the NiZn battery, it uses low-voltage semiconductor devices and capacitors to automatically transfer power among NiZn cells. Moreover, the switches can operate at a high switching frequency to reduce the overall size of the reactive components. Analyses of the switched-capacitor equalizer are given, followed by simulation and design guidelines on key components. Comparisons of three types of switched-capacitor circuits are presented to discuss the pros and cons. Control strategies for the power switches are also elaborated. Experimental results of the low-voltage, low-profile switched-capacitor equalizer for a battery string consisting of 30 series-connected NiZn battery cells confirm that the proposed switched-capacitor equalizer is able to maintain similar voltages among different cells while providing a cost-effective solution for battery energy storage systems.

EEG is widely used as an efficient tool in clinical diagnosis and human cognitive studies due to its high temporal resolution. While EEG provides a wealth of physiological information about the brain, the execution of data analysis algorithms is rather time-consuming due to the massive amount of EEG data and the complexity of the algorithms used in the process. A typical example is Independent Component Analysis (ICA), which is widely used for separating unwanted noise artifacts from neural signals and in cortical source localization. ICA is essentially a statistical signal unmixing method that performs multiple iterations to update the unmixing/mixing matrix to achieve maximum statistical independence among the components, where each iteration/propagation involves numerous matrix multiplication operations.

The tensor core is a hardware unit in most modern GPUs first introduced in the NVIDIA Volta architecture, which seems to be ideal for speeding up the time-consuming matrix multiplication operations in the ICA algorithm. Similarly to the well-known CUDA core, the tensor core is also a computing unit of the Streaming Multiprocessor (SM), but the input data to the tensor cores are a set of matrixes rather than single values processed by the CUDA cores. Each Tensor Core provides a 4x4x4 matrix processing array that operates D = A * B + C, where A, B, C, and D are 4×4 matrices. Each tensor core can perform 64 floating-point FMA operations per clock cycle, 64 times more than a traditional CUDA core. Therefore, for implementing algorithms involving many matrix multiplication and addition operations, tensor cores can provide multiple times the performance of the CUDA cores.

In the presentation, we will introduce the implementation strategy and details of using the tensor core for accelerating the Infomax ICA algorithm, including performance profiling/comparison and numerical error analysis.

Large language models (LLMs), such as GPT-4.0 and Gemini have achieved excellent performance on natural-language tasks, and they also show high expectations for logical reasoning.However, in the intricate field of printed circuit board (PCB) routing, complex scenarios still largely depend on the expertise of experienced engineers, requiring considerable time and effort. The ability of large language models to handle logical problems highlights their potential for addressing PCB-routing challenges. This paper introduces an innovative approach leveraging few-shot learning and chain-of-thought prompting within LLMs to address this challenge, aiming to assist engineers in PCB design with minimal data input. By testing LLMs with a limited number of examples using zero-shot, one-shot and few-shot methods, we assess the models' performance and prove few-shot has the best effort, illustrating their potential to streamline design tasks. Furthermore, we explore fine-tuning techniques to enhance the effectiveness of the few-shot learning approach, to overcome the limitation of scarce real-world PCB cases, we employed code synthetic cases to fine-tune the model in place of actual PCB scenarios, ultimately improving the LLMs' capability to manage intricate routing tasks. The results validate the feasibility and effectiveness of this method, offering a promising avenue for reducing the manual burden in PCB design.

As semiconductor integrated circuits continue to miniaturize, the chip plane miniaturization process under Moore's Law has gradually approached the limit, making the demand for high-density integration processes increasingly urgent in the field of semiconductor integrated circuits. To further improve the integration density as well as the performance and function of chips and systems, researchers have been attempting to continue the development of Moore's Law through 3D heterogeneous integration, by integrating a variety of materials like wide-band semiconductors and novel nanomaterials as well as stacking multiple types of functional devices in 3D direction. In recent years, wafer bonding technology has rapidly emerged as a promisingly pivotal solution for achieving heterogeneous integrated chips and systems with high-density interconnection and high-performance. Wafer bonding technology refers to closely connect two smooth and clean wafers together by different physical or chemical methods to assist semiconductor manufacturing processes or to form heterogeneous composite wafers with specific functions. Currently, wafer bonding technology has extensively penetrated various application fields such as microelectronics, power electronics, memories, photonics, optoelectronic integration, heterogeneous integration and advanced packaging, achieving some significant research progresses. With the continuous advancement of technology and the increasing demands, the application scope of wafer bonding technology will further expand, injecting new impetus into innovative developments across various domains. In this report, different wafer bonding methods and mechanisms will be introduced in details, such as the permanent bonding/temporary bonding, wafer level bonding/die to wafer bonding, direct bonding/indirect bonding and hybrid bonding, etc. Also, some results in our recent work will be shared about successfully developed direct bonding processes between various materials at low temperature, which greatly reduces the thermal stress and crystal deformation caused by high temperature between heterogeneous materials, making it possible to attain high-yield wafer bonding between large-size heterogeneous materials with ultra-thin atomic scale interfaces.

In this paper, the power characteristics of the electrolytic capacitor drive system is investigated. Due to the absence of high-capacity energy storage components on the DC bus, the topology of the electrolytic capacitor variable frequency drive system requires fast response of the rectifier and inverter switching tubes to avoid the instantaneous surge voltage generated during load shedding operation. Traditional control methods have fixed control parameters and cannot provide optimal control parameters for dynamic operation, making it difficult to control the switching tube to quickly achieve energy flow and threatening the stability of the system.

The experiments covering steady-state and off load operation in different power ranges be designed. A dataset of instantaneous stable operation control parameters for each loop has been obtained by debugging the optimal operating state.And a data generator is built to support the steady-state operation of the system by controlling the parameter dataset through instantaneous stable operation. At the same time, a systematic analysis of mathematical models and energy flow laws is conducted to determine the sampling frequency for data collection, including data from system start-up to steady-state and dynamic multi working conditions such as load shedding, in order to digitize the entire process and optimize and improve the database. Finally, based on the data, deep network modeling is carried out to achieve differentiated control and stable operation of the system.

A simulation model and an experimental platform are built to verify the feasibility of the control strategy of the electrolytic capacitor-less variable frequency speed control system, and it is proved that the proposed control strategy can effectively improve the reliability of the system.

To achieve real-time positioning and map construction in an unknown harsh environment with low precision and poor reliability, and realize accurate, efficient simultaneous localization and mapping (SLAM) of overloaded robots in an unknown environment, this paper proposes a SLAM algorithm to improve resampling by Rao-Blackwellized particle filtering. In the Rao-Blackwellized particle filters SLAM (RBPF-SLAM) algorithm, by using high-precision radar data, the odometer reading-based motion model is employed as the proposed distribution function, and the proposed distribution function based on the odometer reading is performed, thereby greatly reducing the number of particles. A particle weight balancing strategy was introduced during resampling to solves the problem of particle weight degradation and particle starvation caused by inaccurate grid map construction so that the map is perfectly matched with the actual environment, which greatly improves the construction efficiency of the crawler-type heavy-duty robot in an unknown environment to save computing resources. The improved algorithm and the basic RBPF-SLAM algorithm are compared in simulated environment and physical environment separately. The improved RBPF-SLAM algorithm flow is distinguished from the basic RBPF-SLAM algorithm in the resampling process. In the improved resampling process, particles are sampled according to the particle weight value, so the weight value is preprocessed and optimized.Experimental results show that the proposed RBPF-SLAM algorithm is better than the basic RBPF-SLAM algorithm in both the virtual environment and the real environment. The map is more accurate and faster, and the effectiveness of the improved algorithm is verified. The follow-up work will focus on the comparison of the improved algorithm with the latest algorithm and the verification in an unstructured experimental environment.