A new supervised multi-head self-attention autoencoder for health indicator construction and similarity-based machinery RUL prediction

Remaining useful life (RUL) prediction plays a significant role in the prognostic and health management (PHM) of rotating machineries. A good health indicator (HI) can ensure the accuracy and reliability of RUL prediction. However, numerous existing deep learning-based HI construction approaches rely heavily on the prior knowledge, and they are difficult to capture the key information in the process of machinery degradation from raw signals, thereby affecting the performance of RUL prediction. To tackle the aforementioned problem, a new supervised multi-head self-attention autoencoder (SMSAE) is proposed for extracting the HI that effectively reflects the degraded state of rotating machinery. By embedding the multi-head self-attention (MS) module into autoencoder and imposing the constraint of power function-type labels on the hidden variable, SMSAE can directly extract the HIs from raw vibration signals. As the current HI evaluation indexes don’t consider the global monotonicity and variation law of HI, two improved monotonicity and robustness indexes are designed for the better evaluation of HI. With the proposed HI, a two-stage residual life prediction framework based on similarity is developed. Extensive experiments have been performed on an actual wind turbine gearbox bearing dataset and a well-known open commercial modular aero-propulsion system simulation (C-MAPSS) dataset. The comparative results verify that the constructed SMSAE HI has better comprehensive performance than the typical HIs, and the proposed prediction method is competitive with the state-of-the-art methods.     

金字塔砂带磨损状态的声信号GA-BP识别方法

目的 金字塔砂带连续磨损会引发钝峰、材料去除能力差和产热多等问题,为避免砂带磨损造成加工效率持续降低和工件表面质量逐渐恶化,需提高金字塔砂带磨损预测能力。方法 在配有声音采集系统的力控机器人磨削系统中对钛合金工件进行了砂带磨损试验;基于Archard模型建立了金字塔砂带磨损模型,并对金字塔砂带磨损程度进行量化;然后利用短时傅里叶和小波包分解分析、提取砂带磨损相关的声音特征;基于声音信号特征建立GA-BP模型,并对金字塔砂带磨损状态进行预测。结果 Kr与R0规律相近,随着磨削速度的增大而略微增大。对磨削声音进行小波包分解,DD2频段的声音特征随磨削时间逐渐降低,相较于其他频段更具有规律性。提取DD2频段的声音信号特征建立GA-BP模型,并对金字塔砂带磨损状态进行预测。结果表明,决定系数(R²)大于0.8,平均绝对误差(MAE)小于0.04,平均偏差误差(MBE)在±0.002之间,均方误差(RMSE)小于0.05。结论 随着砂带的磨损,金字塔尖锐的胞体开始磨平,单颗胞体的局部压力逐渐减小,材料去除能力减弱,产生的微振荡越来越弱,高频信号的声音特征逐渐下降。通过DD2频段声音信号特征建立的GA-BP模型对金字塔砂带磨损状态进行预测,具有准确性和稳定性。     

Cell voltage monitoring All-in-One. A new low cost solution to perform degradation analysis on air-cooled polymer electrolyte fuel cells

In this work, a new low-cost and high-performance system for cells voltage monitoring and degradation studies in air-cooled polymer electrolyte fuel cells has been designed, built and validated in the laboratory under experimental conditions.This system allows monitoring in real time the cells’ voltages, the stack current and temperature in fuel cells made of up to 100 cells. The developed system consists of an acquisition system, which complies with all the recommendations and features necessary to perform degradation tests. It is a scalable configuration with a low number of components and great simplicity. Additionally, the cell voltage monitoring (CVM) system offers high rate of accuracy and high reliability and low cost in comparison with other commercial systems.In the same way, looking for an "All-in-One" solution, the acquisition hardware is accompanied by a software tool based on the "plug and play" philosophy. It allows in a simple way obtaining information from the cells and performing a degradation analysis based on the study of the polarisation curve. The different options and tools included in the CVM system permit, in a very intuitive and graphical way, detecting and quantifying the cell degradation without the need of isolating the stack from the system.Experimental tests carried out on the system validate its performance and show the great applicability of the system in cases where cell faults detection and degradation analysis are required.     

Columnar to equiaxed transition in solidification processing

The columnar to equiaxed transition (CET) is a microstructural transition that has to be controlled during certain solidification processes. CET models and microstructure selection maps, based on nucleation and dendritic growth models, are briefly described. Processing maps can be established by combining a CET model with numerical calculations of local solidification conditions. Such processing maps are helpful for the control of the solidification microstructure. As examples, two processes are discussed: the epitaxial laser metal forming (E-LMF) process where a single crystalline superalloy part is repaired in single crystal form and equiaxed grains have to be avoided; and the arc welding of aluminium alloys where equiaxed grains are the preferred solidification structure.     

Improvement in the corrosive fatigue damage to the low-pressure steam turbine blades due to unbalanced currents

This paper presents two improved methods for the corrosive fatigue damage to the blades of low-pressure steam turbines due to corrosive operating environments and power system unbalance. Two blade materials as AISI-403 and 17-4PH are chosen into the investigation, which provides an enhanced understanding on the effect of materials to reduce the corrosion fatigue damage of blades. To verify the validity of the proposed methods, simulated results of a 951 MW operating turbine generator for the corrosive fatigue damage are presented in this paper. From the simulated results, it can be found that the corrosive fatigue damage to the low-pressure steam turbine blades is significantly reduced. In addition, the effects of unbalanced currents (I₂) on the corrosive fatigue damage to the steam turbine blades are estimated when turbine generators operate in various stress concentration factors and corrosive conditions. The estimations may provide with the safe operating standard on the new or designed generator sets under corrosive environments.     

Interlaminar mechanical properties of carbon fiber reinforced plastic laminates modified with graphene oxide interleaf

By taking the advantage of the excellent mechanical properties and high specific surface area of graphene oxide (GO) sheets, we develop a simple and effective strategy to improve the interlaminar mechanical properties of carbon fiber reinforced plastic (CFRP) laminates. With the incorporation of graphene oxide reinforced epoxy interleaf into the interface of CFRP laminates, the Mode-I fracture toughness and resistance were greatly increased. The experimental results of double cantilever beam (DCB) tests demonstrated that, with 2 g/m² addition of GO, the Mode-I fracture toughness and resistance of the specimen increase by 170.8% and 108.0%, respectively, compared to those of the plain specimen. The improvement mechanisms were investigated by the observation of fracture surface with scanning electron microscopies. Moreover, finite element analyses were performed based on the cohesive zone model to verify the experimental fracture toughness and to predict the interfacial tensile strength of CFRP laminates.     

Effect of compound dimple on tribological performances of journal bearing

The effect of the compound dimple on the tribological performances of a journal bearing is studied. In doing so, the tribological performances for the bearing with the compound dimple and simple dimple are studied using a fluid structure interaction (FSI) method and compared further. Numerical results show that the compound dimple can supply the larger load-carrying capacity and lower friction coefficient due to its twice hydrodynamic action in comparison with the simple dimple. Moreover, the above improvement depends on the geometry sizes of the compound dimple, the dimple interval, and working parameters of the bearing.     

Mechanical properties of carbon nanotube reinforced polymer nanocomposites: A coarse-grained model

In this work, a coarse-grained (CG) model of carbon nanotube (CNT) reinforced polymer matrix composites is developed. A distinguishing feature of the CG model is the ability to capture interactions between polymer chains and nanotubes. The CG potentials for nanotubes and polymer chains are calibrated using the strain energy conservation between CG models and full atomistic systems. The applicability and efficiency of the CG model in predicting the elastic properties of CNT/polymer composites are evaluated through verification processes with molecular simulations. The simulation results reveal that the CG model is able to estimate the mechanical properties of the nanocomposites with high accuracy and low computational cost. The effect of the volume fraction of CNT reinforcements on the Young's modulus of the nanocomposites is investigated. The application of the method in the modeling of large unit cells with randomly distributed CNT reinforcements is examined. The established CG model will enable the simulation of reinforced polymer matrix composites across a wide range of length scales from nano to mesoscale.