Developing high-performance thermoplastic polymer and polymer composites with convenient manufacturing and recycling performance still presents a challenge. Herein, a micro cross-linking strategy was introduced to balance the relationship between processability and mechanical properties of thermoplastic polymer. Though this strategy, a polyfunctional aromatic amine containing disulfide bond (AFD) was used as chain extender to prepare the micro cross-linking thermoplastic epoxy (MTPE). By controlling the AFD content, MTPE with various cross-linking densities were successfully synthesized. Compared with linear thermoplastic epoxy (TPE), the glass transition temperature and tensile strength of MTPE, and the tensile properties of carbon fiber reinforced MTPE composites (CF/MTPE) was significantly increased by 11.5 %, 16.3 % and 7.8 %, respectively. And, due to the dynamic exchange characteristics of disulfide bonds, the processing temperature of MTPE could be consistent with that of TPE. Additionally, full-length carbon fibers could be conveniently and rapidly recovered from CF/ MTPE composites. Due to the excellent compatibility between MTPE and difunctional epoxides, the fine powder of CF/MTPE composites could be treated as a filler to enhance the mechanical properties of difunctional epoxides-based thermosets. This micro cross-linking strategy provides a new way for producing high-performance thermoplastic polymers and offers valuable insights into the recycling of composite materials.

Accurate prediction of lithium-ion battery End-of-Discharge (EoD) points is critical for optimizing battery operation and ensuring the safety and reliability of critical missions. However, the aging trajectories of cells in a battery pack exhibit strong spatiotemporal heterogeneity, and full life cycle discharge data are often unavailable under real operating conditions, making accurate generalized prediction difficult for traditional models. To address these issues, this article introduces the Artificial Intelligence Digital Twin (AI-DT) concept and proposes an AI-based aggregation modeling method for lithium-ion battery packs to predict voltage discharge trajectories and obtain EoD points. Firstly, the proposed Dynamic Time Warping Algorithm Based on Interpolation Method (IDTW) is used to capture time-series features from the battery discharge trajectory. Subsequently, a Discharge Trajectory Prediction (DTP) model is trained using the proposed Transformer-Length (Transformer-L) algorithm, which infers aging characteristics from partial discharge data to forecast full voltage trajectories. Considering the significant spatiotemporal differences in the aging of individual cells in the battery pack, Sliding Window based Drift Detection Method (SW-DDM) and Low-Rank Adaptation (LoRA) algorithm are further constructed to optimize the simulation output of the DTP model, ultimately resulting in high-precision EoD prediction. Experimental results demonstrate that the proposed modeling method exhibits strong generalization capability. With only partial discharge data of an individual battery, it can achieve high-precision and highly transferable discharge trajectory prediction for the same type of cells in the battery pack throughout its entire lifecycle, thereby enabling accurate EoD estimation and providing robust support for enhancing battery reliability.

Conventional soft sensors often suffer from challenges such as crosstalk, hysteresis, and limited sensitivity, which hinder their performance and broader applicability. This paper presents a multi-axis piezoresistive soft force sensor with a square-column-shaped sensing structure designed to reduce the spatial footprint and mitigate partial axial coupling effects. By integrating a Wheatstone bridge-based resistive compensation strategy, the sensor achieves self-decoupling in multi-axis force measurements. Furthermore, a generalized Preisach hysteresis model is implemented to effectively compensate for hysteresis-induced nonlinearities and input-output loop effects, significantly enhancing sensing accuracy and precision. Extensive experimental validations confirm the effectiveness of the proposed self-decoupling and hysteresis compensation methodologies, demonstrating notable improvements in sensor reliability and performance. The findings of this study establish a comprehensive framework for advancing multi-dimensional soft force sensing technologies, with promising implications for high-precision engineering and biomedical applications.

The electro-hydrostatic actuator (EHA) is a critical component in aerospace flight control systems, and its nonlinear dynamic characteristics pose significant challenges for accurate performance prediction. To enhance modeling precision and adaptability, this paper proposes a performance prediction and update method for EHA based on a hybrid TCN-Transformer architecture. A simulation model is constructed using multi-dimensional time-series signals including voltage, current, and rotational speed. The temporal convolutional network (TCN) is employed to capture local temporal patterns, while the Transformer extracts long-term dependencies, enabling high-precision prediction of displacement and speed. A drift detection mechanism based on a surrogate model is introduced to monitor prediction error distributions and identify system performance changes. A fine-tuning strategy is then applied to rapidly update the model to adapt to the new data distribution. Experimental results demonstrate that the proposed method achieves an average relative error below 1.5% under various operating conditions, indicating strong generalization and adaptability. This study offers an efficient and practical approach for state prediction and intelligent monitoring of EHA systems, contributing to improved safety and reliability in flight control applications.

Currently, the remaining useful life (RUL) prediction of lithium-ion batteries through data-driven methods undergoes extensive investigation. However, due to the influence of individual battery differences and dynamic environmental conditions, traditional data-driven modeling is difficult to accurately represent the actual working status of lithium batteries. This paper proposes a method for predicting the RUL of lithium batteries by dynamically updating the degradation model. It utilizes a hybrid data-driven approach combining convolutional neural networks (CNN) and long short-term memory networks (LSTM) to build a predictive model for lithium batteries. Additionally, it incorporates mechanisms for concept drift detection and dynamic model updates, enabling real-time monitoring and pinpointing of concept drifts within the actual data streams. By dynamically updating the pre-trained static models, we aim to enhance the accuracy and adaptability of predictions. Through a comparative analysis of prediction outcomes before and after the updates, we validate the effectiveness of our proposed method in addressing data drifts and improving the RUL prediction performance. This innovative technique offers a new approach for lithium battery management and maintenance, contributing significantly to the realm of monitoring battery health.

In order to effectively and accurately track the uncertainty of the aero-engine, to improve the modelling accuracy by using the model update method, and to construct a high-fidelity numerical model of the aero-engine, this paper proposes a standard modelling method for self-updating the numerical model. Firstly, the basic features of the engine sensor outputs are acquired by learning the data set from zero to the current moment of data, and the model outputs at the next moment are predicted. Secondly, the model fine-tuning updating is used in the updating method, and the number of relevant network layers is frozen, on the basis of which the static model is fine-tuned at the top level. Finally, a series of experiments on the CMAPSS dataset demonstrate the significant effect of the method in improving the modelling accuracy.

To enhance ergonomic performance by detecting operator emotional states based on grip force characteristics in remote manipulation tasks, we have developed a two-degree-of-freedom force feedback control stick equipped with multi-finger force sensing. An orthogonal serial mechanism, governed by a field-oriented impedance control strategy, was implemented to deliver high-fidelity force feedback to the operator’s hand. This setup allows the stick to generate haptic feedback by applying motor output torques proportional to its displacement. Compact custom force transducers embedded in the ergonomically designed handle simultaneously capture forces from all five fingertips (range: 0–7 N) at a sampling rate of 250 Hz. Performance validation demonstrated that the stick provides a workspace of ±40° in the anteroposterior direction and ±35° in the mediolateral direction, with peak torques of 1.19 N·m and 1.12 N·m, respectively. In a virtual reality submarine piloting task, subjective user evaluations confirmed that integrating force feedback with audiovisual scenarios indeed enhanced the operators’ sense of immersion and telepresence during manipulation simulations.

In order to improve the management of lithium-ion batteries (LiBs), it is necessary to predict the state of health (SOH) in real time. In this paper, a real-time prediction method for SOH of LiBs based on digital twin is proposed. First, we construct a future data reconstruction model, which can reconstruct the complete discharge cycle data based on real-time input data. After that, we build an SOH prediction model to predict the future battery SOH by combining historical data and current real-time reconstruction data. Experiments on MIT's public battery dataset show that the proposed method has excellent real-time prediction effect and high accuracy in the initial stage of battery cycling.

Magnetically suspended flywheels, as key actuators in aerospace attitude control systems, are widely used in highresolution remote sensing satellites, space stations, and other longlife spacecraft due to their frictionless operation and highprecision advantages. However, during long-term in-orbit operation, magnetically suspended flywheels inevitably experience performance degradation. This paper proposes an improved CNN-LSTM degradation calibration network. To improve the tracking accuracy of degradation trends in magnetically suspended flywheels, we construct a multi-scale feature representation framework. This framework integrates CNN’s local feature extraction capabilities with LSTM’s dynamic time series modeling advantages. The resulting hybrid model enhances the reliability of spacecraft in-orbit health management decisions.

The polycrystalline diamond (PCD) tools have excellent cutting performance in machining composites. However, the tool wear mechanism of PCD tools in milling C-f/SiC composites remains unclear. There are few researches on the effect of diamond grain size on the cutting performance of PCD tools and material removal mechanism. In this paper, four PCD samples with different grain sizes were prepared under high-temperature and high-pressure condition. The microstructure, flexural strength, and cutting performance of PCD samples was tested. The wear mechanism of the tools was analyzed by observing the wear morphology of the front and rear cutting surfaces of the PCD tools. The influence of diamond grain size on the cutting performance of PCD tools was investigated in terms of tool life, surface roughness and material removal mechanism. The results indicated that the highest bending strength was found in the PCD sample with the grain size of 5 mu m, as the diamond grain size increased, the bending strength of the PCD tool decreased. Tool failure was caused by fiber bundles and chip dust scratching the binder and diamond grains. In the same cutting distance, due to the difference in bonding between diamonds of different grain sizes. The 2 mu m similar to 30 mu m grain size of the PCD tool wear was the slowest, 5 mu m grain size of the fastest PCD tool wear. As the grain size decreased, the machined surface quality of C-f/SiC composites improved. The fiber removal mechanisms employed the gradual transition from shear removal to bending removal as the grain size increased.

The phthalonitrile monomer (AFPN) containing trifluoromethyl was prepared and blended with the widely used bisphenol A diglycidyl ether (DGEBF)/diaminodiphenyl sulfone (DDS) system to prepare a series of thermosets. The incorporation of AFPN endowed the thermoset with obviously promoted thermal stability and flame retardancy. The glass transition temperature (Tg) and char yield at 800 degrees C in N2 were 161 degrees C and 18.1 % respectively for the pure EP/DDS (EP0) which increased to 203 degrees C and 32.3 % for the thermoset containing 30 wt.% AFPN (EP/AF30). The detailed study on the mechanism of pyrolysis and flame retardancy was presented. The formation of graphite-like nitrogen-containing structure and center dot CF3 free radicals during combustion in the solid phase and gaseous phase, respectively, was confirmed. It was confirmed that there was synergistic effect of phthalonitrile and trifluoromethyl on the flame retardancy of the thermoset. Moreover, the dielectric constant decreased from 2.39 to 1.45 at 1 MHz which is favorable for the application as low dielectric materials.

To effectively solve the problem that the accuracy of traditional modeling methods decreases dramatically when the performance of an aircraft engine is degraded, a high-fidelity modeling method for aircraft engine degradation tracking is proposed. First, based on partial engine sensor data, a static digital model of the engine is constructed, which aims to track the output of the engine sensors at the current moment. Then, the principle and implementation of updating the engine static digital model using this framework are analyzed in details. The performance compared with engine static digital model before the update. Experimental results on the N-CMAPSS dataset demonstrate the significant effect of the proposed method on both engine model updating and degradation tracking.

To impart high flame resistance, enhanced thermal stability, and low dielectric properties to epoxy resin while maintaining good mechanical behaviors for high-end applications, a monomer (BZPN) containing the characteristic structure of benzoxazine, phthalonitrile, and trifluoromethyl was prepared and added into the Bisphenol A-type epoxy resin (DGEBA)/Dapsone (DDS) combination. The glass transition temperature (Tg) and carbon yield under a nitrogen atmosphere at 800 °C were found to significantly increase from 155 °C, 17.2% to 236 °C, 50.3%, respectively, for the neat EP/DDS and the BZPN-containing material. The UL-94 flammability rating achieved V-0 level when the BZPN content was 19.2 wt.% (EP-BZ-1). The thermal decomposition and flame retardancy mechanism were explored by TGA-FTIR, Raman, and XPS analysis. The fluorine-containing products were found in both the gas phase and the char residue, implying that the •CF3 radicals played an important role in promoting the flame-retardant behaviors through a radical trapping mechanism. The dielectric constant and dielectric loss of the materials decreased as anticipated. In addition, mechanical testing of carbon fiber-reinforced composites showed that the BZPN-containing resin presented equivalent mechanical behaviors to the neat EP/DDS resin. The synthesized BZPN was proved to be an effective and promising additive for the epoxy-based composite.

To effectively solve the problems of difficult to obtain data of aircraft engines and the decrease of accuracy of traditional remaining useful life prediction methods, an online prediction method of remaining useful life of engine is proposed, which mainly consists of an online model and a prediction model. The online model learns from partial running data of the engine and can be continuously updated with new running data. The purpose of the model is to keep in line with the real state of the engine. Then, the principle and implementation process of the engine RUL prediction model are analyzed in detail. Compare the performance of online prediction methods with other prediction methods. The experimental results on the dataset C-MAPSS show that the prediction performance of online methods is superior in the absence of sufficient data.

Magnetic bearing rotor systems are widely used in high-speed rotating machinery. However, conventional mechanism-based models struggle to capture system behavior. This is due to strong nonlinearities and multi-condition coupling. To address this challenge, a data-driven modeling method called LCTF-Net is proposed. It is a hybrid neural network that combines time-domain and frequency-domain features. First, frequency-domain features of control current signals are extracted using the fast Fourier transform (FFT). These features are then processed by a convolutional neural network (CNN) to learn spatial patterns. In parallel, a long short-term memory (LSTM) network captures the temporal dynamics from the original time-domain signals. The two branches are merged into a unified deep learning architecture. This structure effectively fuses time-domain and frequency-domain information. Preliminary experiments show promising results. The rotor response over the full speed range can be accurately predicted using only partial gradient speed data. The proposed method provides an accurate and generalizable solution for digital twin modeling of magnetic bearing rotors.

In recent years, modern industry has seen growing demands for high-performance rotating machinery, imposing higher requirements on the dynamic performance and reliability of rotor systems. Developing an efficient and accurate rotor dynamic model can not only provide outputs that closely match real rotor dynamic characteristics and vibration responses, but also simulate potential vibration sources and corresponding control strategies. However, when vibration sources involve the interaction of multiple complex factors or arise from processes not yet fully understood, it becomes challenging to accurately describe and quantify them through a single mechanism-based model. To address current modeling requirements for high-precision control and long-term autonomous health management, this paper proposes a mechanism-data hybrid-driven modeling method for vibration response prediction of the magnetically suspended rotor system. It fully utilizes the mechanism model, measured data, and artificial intelligence algorithm to achieve high-precision simulation of the modeling object. First, a mechanism model incorporating mechanical rotor system, magnetic bearing control system, and internal vibration sources is established. The measured vibration data are used to analyze the unbalanced parameters to be identified in the mechanism model, achieving the first stage of mechanism model calibration. Second, an attention-based data mapping network is designed to extract harmonic disturbance information caused by external vibration sources from measured vibration data, which is then supplemented to the calibrated mechanism model, achieving the second stage of mechanism model correction. Finally, the vibration response prediction performance of the constructed hybrid mechanism-data model is compared with that of the physical prototype platform, verifying the effectiveness and superiority of the proposed mechanism-data hybrid-driven modeling method.

Phthalonitrile monomer containing oxazine ring (BPS-Ph) was synthesized using bisphenol-S, aniline, paraformaldehyde, and 4-nitrophthalonitrile via a two-step method. The differential scanning calorimetry (DSC) data of the curing process of BPS-Ph monomer indicates that the peak polymerization temperature of the cyan group is about 250 degrees C and that confirms the promoting effect of oxazine ring on the curing reaction of phthalonitrile. The rheological data of E51/DDS containing BPS-Ph indicates that the BPS-Ph monomer also has promoting effect on the curing process. The activation energy was studied via the iso-conversional method principle. The activation energy corresponding to the E51/DDS crosslinking reaction is 75.79 and 60.54 kJ mol-1, respectively, for pure E51/DDS and E51/DDS/BPS-Ph. Moreover, the activation energy of the crosslinking of phthalhydrazine is 74.1 kJ mol-1 which is rather lower than that in pure BPS-Ph, and that implies the synergistically catalytic effect of the oxazine ring and the amine. The addition of BPS-Ph monomer significantly improves the thermal behaviors of the thermoset. The char yield in N2 atmosphere increased from 15.6% to 53.9% when adding 30% BPS-Ph to the E51/DDS. Moreover, the peak weight loss temperature decreased from 407 to 385 degrees C. The phase separation size of poly-BPS-Ph in E51/DDS is about tens to hundreds nanometers according to the atomic force microscopy (AFM) phase diagram. It is indicated that incorporation of phthalonitrile monomer containing oxazine ring into epoxy/amine system is an applicable and effective way to promote the thermal behavior of the thermoset. Curing mechanism and thermal behaviors of the epoxy resin containg the phthalonitrile/benzoxazine monomer. image

The phenolic-type phthalonitrile (PN) was added to EP/DDM system in order to enhance the thermal and mechanical performance at high temperature. The influence of the added PN on the curing process of EP/DDM was studied via DSC and the activating energy (E-alpha) was calculated based on iso-conversional method. The E-alpha values corresponding to EP/DDM crosslink reaction remained at about 60 kJ mol(-1) while it dramatically increased to 68.2 kJ mol(-1) when PN content reached 50 wt% (EP-PN50). The T-g and char yield at 700 degrees C in N-2 increased from 141 degrees C, 25.7%, for the neat EP/DDM to 226 degrees C, 68.7% for the EP-PN50. The measured char yields of the cured blend were higher than the calculated values which implies the interaction between EP/DDM and polyphthalonitrile network. The tensile and bending tests were carried out at 413 K and the modulus of EP-PN50 remains 2.3 Gpa. On the meantime, the cyano-functionalized SiO2 (CN-SiO2) was prepared to further promote the mechanical behaviors of this resin blend in high temperature. The contact angles of raw SiO2, KH560-SiO2, CN-SiO2 with EP-PN50 are 59.3, 52.6, 49.7 degrees, respectively, which confirms the better wettability of CN-SiO2 to the EP/PN blend. Furthermore, the tensile and bending tests conducted at 413 K confirmed that the CN-SiO2 was more efficient on enhancing the mechanical performance of this EP/DDM/PN system at high temperature.

For the 5-DoF gimballing magnetically suspended flywheel (MSFW), a three-degrees-of-freedom (DoF) active magnetic bearing (AMB) with a conical gap is used as an electromagnetic actuator to suspend the flywheel (FW) rotor, so the gyro moment of gimballing MSFW is improved by actively deflecting FW rotor with the conical AMB. Firstly, the dynamic models of the FW rotor with the 3-DoF conical AMB are investigated, and the magnetic forces using the equivalent magnetic circuit and the FEM model are analyzed. Moreover, the coupling effect among three controllable DoFs of conical AMB is investigated when the position of the FW rotor sets at different values. Furthermore, the negative torques of conical AMB are studied when the FW rotor is stably suspended at the equilibrium position. The experimental results indicate that the controllable deflection angle of 3-DoF conical AMB reaches 1°, the coupling ratio among three controllable DoFs is reduced to 9%, and the negative torque during the active deflecting process is 0.02 Nm. Compared to the normal AMB with the planar gap, the designed 3-DoF conical AMB has a wide deflection angle, minor coupling effect, and tiny negative torque. Therefore, the 3-DoF conical AMB causes slight influences on the FW rotor during the suspension and deflection control of gimballing MSFW, and it is more suitable for the gimballing MSFW.

The deep and narrow groove structure has wide applications in micro-devices such as high-frequency circuits, micro heat pipes, microchannels, etc.. It has been involved in various fields, including aerospace, electrocommunication, and medicine. However, as the deep and narrow groove structure shrinks, the processing accuracy and surface quality become challenging factors limits its further development. The radius of the micromilling cutter edge plays an important role in precision and surface quality. Therefore, this paper conducts a systematic experimental study on the edge radius using custom-made PCD micro-milling cutters and commercially available cemented carbide milling cutters. This study investigated the influence of three different cuttingedge radius, 1.73 mu m, 2.39 mu m, and 2.97 mu m, of PCD micro-milling cutters and varying feed rates per tooth on milling surface morphology and burr formation. A comparison was also made between the cutting performance and machining quality of PCD micro-milling cutters and cemented carbide micro-milling cutters. The results indicated that the machining quality of deep, narrow grooves is significantly related to the cutter's cutting-edge radius, tool material, and milling parameters. PCD tools with smaller cutting-edge radii produced micro-grooves with an optimal surface roughness of approximately 40 nm and minimal surface burrs under the same parameters. Therefore, using PCD micro-milling cutters with a smaller cutting-edge radius is crucial in improving the quality of deep and narrow groove machining.