Self-developed Platform Research Articles

Spatial structure digital twins: Application in intelligent health monitoring of cable dome structures

Traditional structural health monitoring (SHM) based on contact sensors has drawbacks, such as high cost, susceptibility to interference, and narrow coverage. Digital twins (DTs) present an effective intelligent solution. However, the real-time monitoring of all structural components remains challenging due to high computational cost of high-order models in underlying framework. Additionally, the vast number of spatial structures globally and their long service time indicate the potential necessity of introducing SHM based on DT. In this context, this study proposed a reduced-order model-based DT framework for spatial structures. By compressing the experimental design samples, the multidimensional high-order physical models were reduced to several approximate low-order models for constructing a DT model, thereby achieving real-time calculation covering all components. The proposed framework was applied and validated through an experiment on a scale model of cable dome structure, and real-time interactive visualisation of monitoring data was displayed on a self-developed platform. The results demonstrate that the DT model enabled real-time presentation of calculation results (the calculation speed of <0.5 s/time), the mean relative error of cable tension monitoring was <3.85%, the strut stress monitoring could be controlled within 7.5%, and the mean absolute error of nodal displacement could be maintained at around 1 mm.

A system-level multi-field coupling algorithm for regenerative cooling thrust chamber of a LOX/methane rocket engine

A systematic and hierarchical multi-physics coupling simulation method is proposed to model the combustion, combustion-gas transonic flow, coolant transcritical flow, and transient heat transfer phenomena inside the regenerative cooling thrust chamber subsystem of a LOX/methane rocket engine. By combining this algorithm with a self-developed platform, two distributed parameter modules are developed. Based on the ground test conditions of a LOX/methane engine, a verification case is established, and transient simulation studies are conducted. Comparisons with steady-state test data show that the calculated errors for coolant temperature rise and pressure drop are not higher than 4%. The simulation results reveal the locations where the transition from liquid to supercritical state occurs within the cooling jacket and provide three peril points of wall temperature on the thrust chamber. This study comprehensively reveals the occurrence of transcritical transient processes inside the cooling channels during the regenerative cooling process, laying the foundation for future entire system simulation.

VasLine: Realize online detection and augmented NIR using deep learning

The chemical information acquisition capabilities of near infrared (NIR) have attracted great interests of investigators on exploring its potential as process analytical technologies (PAT) on the traditional Chinese medicine (TCM). However, due to the operation environment of TCM is complex and noisy, the accurate online composition detection and analysis is challenging and remains unsolved. Therefore, in this paper, we propose a new online TCM composition analysis platform for high quality online spectral data acquisition. We design, VasLine, a generative framework based on deep learning to estimate the multiple critical quality attributes (CQAs). To demonstrate the feasibility of the system, the platform was applied to the extraction process of Xiao’er Xiaoji Zhike Oral Liquid (XXZOL) and the framework was trained to estimate the content of the 7 CQAs. In the evaluation, we carried out 16 batches extraction experiments and collected extensive online NIR data, offline NIR data and high-performance liquid chromatography (HPLC) data for TCM composition analysis. The results show that the estimation accuracy of VasLine outperforms the state-of-the-art regression approaches significantly for all the 7 different CQAs, e.g., the R2 metrics of VasLine for the CQAs are all higher than 0.95. This study aims to propose a novel generative framework based on a deep learning model, and the framework is applied in the self-developed platform to estimate CQAs with high-quality generative data from noisy online NIR accurately. The experiments for online TCM composition analysis show that VasLine is a new solution for the quality improvement of the actual pharmaceutical industry.

Applying XAI to an AI-based system for candidate management to mitigate bias and discrimination in hiring

Assuming that potential biases of Artificial Intelligence (AI)-based systems can be identified and controlled for (e.g., by providing high quality training data), employing such systems to augment human resource (HR)-decision makers in candidate selection provides an opportunity to make selection processes more objective. However, as the final hiring decision is likely to remain with humans, prevalent human biases could still cause discrimination. This work investigates the impact of an AI-based system’s candidate recommendations on humans’ hiring decisions and how this relation could be moderated by an Explainable AI (XAI) approach. We used a self-developed platform and conducted an online experiment with 194 participants. Our quantitative and qualitative findings suggest that the recommendations of an AI-based system can reduce discrimination against older and female candidates but appear to cause fewer selections of foreign-race candidates. Contrary to our expectations, the same XAI approach moderated these effects differently depending on the context.

Experimental research on rapid removing characteristics of carbon monoxide generated during gas explosions.

A large amount of gas, such as CO, accumulates in a coal mine after an explosion, leading to CO poisoning. In this study, a self-developed platform was used to eliminate CO from coal mines and determine the mass of the rapidly eliminated CO and its concentration in the eliminated gases. Equations were derived to calculate the amount of CO eliminated and the removing rate. The results showed that a rapid removing reagent in the form of nonprecious metal catalysts is useful for removing CO. Removing agents with larger masses facilitated the activation, irrespective of the CO concentration. For removing reagent amounts of 10, 15, 20, 25, and 30 g, the amount of CO eliminated, the removing rate, and the time required to complete catalytic oxidation increased sequentially. The CO removing process could be divided into three stages (I, II, and III) based on the variations in the CO, CO2, and O2 concentrations during CO removing. The removing reagent first chemically adsorbs CO and O2, and then desorbs CO2. The final CO concentration tends to 0, the O2 concentration remains stable, and the CO2 concentration decreases. This shows that the ablation agent has an impact on the changes in the CO and CO2 concentrations.

A control strategy of normal motion and active self-rescue for autonomous underwater vehicle based on deep reinforcement learning

The survivability of autonomous underwater vehicles (AUV) in complex missions and dangerous situations is of great significance to ocean resource exploration, hydrological research, maritime rescue, and undersea military. Existing researches on motion control for the AUV mainly focus on its normal operating, but the active self-rescue method in emergency situations is hardly found. As classical control methods are not sufficient enough for complicated self-rescue missions of the AUV, this paper uses the deep reinforcement learning (DRL) algorithm to solve this problem because the DRL algorithm has the advantages in learning and decision making for complex robot control missions. In this paper, the normal motion control of the AUV based on the deep deterministic policy gradient algorithm is explored, including the yaw angle adjustment, yaw angle adjustment extension, trajectory tracking, and normal floating-up control of the AUV. Then, active self-rescue methods are successfully achieved to recover the AUV from emergencies, such as ocean water density decreasing sharply or one fin getting jammed at a random angle. What is more, real environment experiments are successfully conducted on a self-developed platform of the AUV to validate the feasibility of the proposed control methods. The results can effectively improve the survivability of the AUV and can be a reference to submarine survivability technologies.

Investigation of Multifield Characteristics in Braking of Lightweight Discs on a Self-Developed Platform

Disc brakes are widely used in various kinds of light vehicles. Currently, it has become a tendency to reduce the mass and volume of brake systems because of the increasing lightweight requirement for vehicles, especially for passenger vehicles in populated cities and for outdoor sport. However, there is often conflict between light weight and reliability. In this article, the braking characteristics of a thin disc brake were investigated experimentally, using a self-developed inertia braking platform. The platform was designed to conveniently measure multifield characteristics in real time. The goal of the experiment is to verify the special performance of a brake system with a very thin solid disc. Measurements of transient temperature, lateral vibration, and noise in the dry friction contacts of a steel solid thin disc and two resin circular pads were performed. A frequency–time analysis method was used for the lateral vibration of the disc and braking noise. The results suggest that one of the reasons for lateral vibration is the mounting error of the thin disc–shaft system, which should receive extra attention for safety reasons. In addition, the braking noise is structural borne.

Laser sintering process optimization of microstrip antenna fabricated by inkjet printing with silver-based MOD ink

Inkjet printing has been widely used for the manufacturing of patch antennas, circuit boards, biochemical sensors, frequency selection surfaces and other electronic devices. As an essential process, laser sintering is of great significance to the inkjet printing, which can enable the printed patterns to possess very good conductivity. The laser sintering process for the MOD (Metal Organic Decomposition) ink is investigated in this manuscript. First of all, the laser sintering simulation of the MOD ink is carried out based on COMSOL Multiphysics and change mechanism of the sintering temperature under different process parameters is revealed. Then, sintering experiments and process optimization are conducted on the self-developed platform, and the sintering quality differences are comparatively analyzed by measuring the widths, conductivities and thicknesses of the printed silver lines. Finally, a microstrip antenna is fabricated based on the optimized process and the performance of the manufactured antenna satisfies the use requirements. The simulation results are highly consistent with the experimental results, and both of them have verified the feasibility, reliability and effectiveness of the optimized sintering process.

High-Lift Aerodynamics Design for Large Civil Aircraft in Fudan University

Part of work on CFD simulations by the high-lift system design team in Fudan university for large civil aircraft is presented. The research on CFD simulation of the high-lift systems and some concepts and experience in three-dimensional geometry modeling are also presented, which are done on the self-developed platform of high-lift device aerodynamic calculation software and programs. For which, the design efficiency is substantially improved.

Recognition of Pixelized Chinese Characters Using Simulated Prosthetic Vision

The rehabilitation of the reading ability of the blind with a limited number of stimulating electrodes is regarded as one of the major functions of the envisioned visual prosthesis. This article systematically studied how many pixels of individual Chinese characters should be needed for correct and economic recognition by blind Chinese subjects. In this study, 40 normal-sighted subjects were tested on a self-developed platform HanziConvertor (Institute for Laser Medicine & Bio-photonics, Shanghai Jiaotong University, China) with digital imaging processing capacities to convert images of printed text into various pixelized patterns made up of discrete dots, and present them orderly on a computer screen. It was found that various complicated factors such as pixel number, character typeface, stroke number, etc., can obviously affect the recognition accuracy. It was also found that optimal recognition accuracy occurs at a specific size of binary pixel array, due to a trade-off between a strictly limited number of stimulation electrodes and character sampling resolution. The results showed that (i) recognition accuracy of pixelized characters is optimal with at least 12 x 12 binary pixels, and therefore it is recommended to apply a minimum of 150 discrete and functioning electrodes for restoring the reading ability of blind Chinese individuals in the visual prosthesis; (ii) fonts of Song Ti and Hei Ti are clearer and more effective than other typefaces; and (iii) characters with fewer strokes lead to better accuracy.