Anomaly detection in unlabelled and highly imbalanced high-dimensional monitoring data is one of the most urgent and challenging industry problems in the energy industry. Based on the powerful high-dimensional data analysis capabilities of autoencoders, the use of autoencoders for anomaly detection is becoming more and more popular. This paper proposes an anomaly detection method based on deep learning and support vector data description. First, feature engineering is built based on an optimized serial deep autoencoder; second, different feature combinations are studied and compared; finally, anomaly detection based on support vector data description. In this paper, experiments are carried out on the actual operating data of a real steam turbine to verify the effectiveness and accuracy of the proposed method.

The complexities of wind turbine wake evolution necessitate the estimation of turbulence and horizontal and vertical wake expansion. The standard approach for developing a Gaussian wake model based on Gaussian similarity assumes that the turbine wake profile is symmetric around the hub height and approximates a normal distribution. The impacts of Atmospheric Boundary Layer (ABL) and ground surface roughness cause the wake to deviate from radial symmetry, resulting in a nonsymmetric wake around hub height. We develop the Bivariate Gaussian (B-Gaussian) Model to estimate the nonsymmetrical wind turbine wake expansion due to ABL effects. The B-Gaussian model factors the correlation coefficient (Ω) to compare the horizontal and vertical expansions. The radial profiles result for the B-Gaussian shows a faster wake decay for the unstable ABL because of faster mixing at higher-turbulence intensities, and slower wake decay is due to slower mixing at lower-turbulence intensities for the stable ABL. The radial wake profiles for the unstable ABL transition from vertical elliptical-shaped in the near wake region to double circular-shaped profiles in the far-wake region as Ω decreases. Finally, compared to preceding Gaussian models, the B-Gaussian outperforms previous Gaussian wake models and has the least overall and downstream entrainment errors.

Due to the intricacies of wind wake transition, it is necessary to estimate both the atmospheric boundary layer (ABL) contributions as well as the momentum deficit and thrust balancing. Wake superposition models such as Linear and Katic produce inconsistent wake predictions that are empirically derived. Designing an explicit cumulative model based on the momentum deficit and thrust budget analysis is not feasible. We use the analytical and machine learning approach to predict and correct inconsistencies between momentum deficits and thrust. The Mass Flow Conservation Superposition (MFCS) model is derived using the Linear Momentum equation, and the Mass Flow and Thrust Conservation Superposition (MFTCS) model is derived by fitting the Gaussian Process Regression (GPR) model to Large Eddy Simulation (LES) data. Noticeably from our results, the MFCS underestimates the thrust magnitude and the wake recovery rates. The MFTCS model based on the GPR model fit correctly estimates the momentum deficit and thrust balance. The GPR model assesses the smoothness and differentiability of the LES data to predict wake recovery. The MFTCS outperforms Katic and MFCS results but is more accurate than the LES data based on the goodness of fit criteria.

Understanding the wake impacts of a wind farm is critical for layout planning and turbine design. The main goal of this study is to model a Mass Flow Conserving Superposition (MFCS) and compare the Linear Wake superposition (LWS) while considering the wake and boundary interaction. This study factors the changes in thrust and momentum inside a wind farm cross-sectional area as the wake develops flows and interacts across five spans. We derived the mass flow conserving wake superposition approach that considers the mass continuity across the wind farm area to calculate the wake evolution between spans. The results for the wake profiles depict a quasi-periodic behavior that peaks at the last span in both the MFCS and LWS approaches. However, the MFCS approach has a more significant shift wake profile between spans than the LWS approach because it accounts for the thrust and boundary interactions changes. The MFCS approach's centerline wake velocity deficit reduces 0.47 to 0.34 m s−1 while the LWS approach's centerline wake velocity deficit remains constant around 0.6 at m s−1. A substantial shift in the MFCS approach's velocity deficit compared to the LWS approach can be attributed to changes in thrust and momentum.

The numerical simulation of wind turbine generators was carried out to investigate the relationship between wake expansion and distance of two wind turbine generators in a row. The commercial software ANSYS Fluent was used to perform the Computational Fluid Dynamics (CFD), and further, the standard k-ϵ turbulence modeling has been adopted to solve the steady-state, 3D Reynolds Averaged Navier-Stokes (RANS) equations. Results of near and far wake regions are analyzed. In fact, due to the squeezing effect of the wind turbine generators on the surrounding air, it is found that there is an acceleration area nearby the wind turbine rotors. The arrangement distance of wind turbine generators in a row is analyzed. The results show that as the arrangement distance of the wind turbine generators increases, the average wind speed and the wind power density in the acceleration area decrease. This conclusion has guiding significance for the arrangement of wind turbines in wind farms.

Correction for ‘The synthesis of polyurethane with mechanical properties that are responsive to water retention states’ by Xuemin Li et al., Polym. Chem., 2021, 12, 1014–1022, DOI: 10.1039/D0PY01559H.

汽轮机正常运行工况的数据比较容易采集,而故障数据的采集存在一定的困难.针对这种情况,采用自编码神经网络对汽轮机高压缸内三个级组的运行数据进行建模,用正常运行工况下的数据训练模型之后对未知工况数据进行重构分析.通过分析,重构前后压降、温差以及效率的变化情况与常见通流故障结垢、磨损、等发生时相关参数的表现,最终分析得出高压缸的一段抽汽到高压排汽这一级组发生了磨损故障的结论.

通过对三种加装不同形状叶尖小翼的等直径轴流风机进行数值计算,研究了不同形状叶尖小翼对轴流风机性能的影响.数值模拟表明,光滑过渡的叶尖小翼更有利于提高风机的效率,但是会不同程度地降低风机的风量.垂直的叶尖小翼会提高轴流风机的风量,同时提高效率,但是提高效率的幅度低于光滑过渡的小翼提高效率的幅度.

基于风洞实验和数值模拟方法,研究了简化车辆模型在超车过程中气动特性的变化规律.整个过程视作准静态,两车横向间距固定且为模型宽的35％,测得两车在不同纵向相对位置时被超车的气动力及流场分布.结果表明:主超车从被超车的后部运动至前部时,被超车受到的气动力和横摆力矩会发生显著变化,当主超车刚到达被超车尾部时,后者所受阻力和侧力均取得最大值.

BACKGROUND: Hemolysis is an important problem in the design and development of artificial heart pumps. Hemolysis is associated with the shear force of blood cells in the pump and the movement time in the pump. OBJECTIVE: To analyze the relationship between the operating parameters of blood pump (flow rate of blood pump), the structure of blood pump (outlet width of blade) and the shear stress of blood cells in artificial heart pump. METHODS: The basic structure of the centrifugal blood pump was designed according to the velocity coefficient method. The three-dimensional model of the blood pump, mesh division and flow field simulation were established by using CFturbo, ANSYS ICEM and FLUENT software respectively to study the best mesh division method. The results of the numerical simulation corrected the diameter of the impeller outlet to finally obtain an artificial heart pump meeting the needs of the human heart. In order to reduce the shear stress and movement time of blood cells in the heart pump, the optimal design and analysis of the impeller inlet flow rate Q (2, 3, 4, 5, 6, 7 L/min) and the operating parameter blade outlet width b2 (2.0, 2.1, 2.2, 2.3, 2.4, 2.5 mm) of the centrifugal artificial heart pump were designed to reduce the probability of hemolysis. RESULTS AND CONCLUSION: (1) As the flow rate increased, the shear stress of the blood cells in the blood pump gradually increased, and the movement time gradually decreased. As the flow rate increased, the standard hemolysis value in the blood pump gradually decreased, and the damage to the blood cells in the blood pump alleviated. When the flow rate was 7 L/min, the standard hemolysis value in the blood pump was the lowest. (2) With the increase of blade outlet width, the shear stress of the blood in the pump decreased, and the movement time of the blood in the pump increased. When the blade outlet width was 2.0-2.5 mm, the hemolysis value in the blood pump decreased with the increase of the blade outlet width. When the blade outlet width was about 2.5 mm, the hemolysis value in the blood pump was the lowest.

为研究水中的非线性声波传播问题,主要开展了基于弱可压缩方程的理论分析及水中声波动及声流模拟工作.首先,采用正压流体的密度与压力函数,结合黏性流体动力学方程,对方程组采用特征变量的波动分析,证明了该方程能描述水中的涡旋和声波传递.然后,结合时空守恒元解元算法求解该方程组;并对一些典型气动声学问题进行了计算和对比研究.通过计算分析了二维顶盖半圆腔驱动流,计算结果与其他研究结论进行对比分析,证实了该方法可以准确模拟出流体的黏性;通过对平面声波在不同黏性流体中传播进行了数值模拟,声波衰减趋势与理论解完全一致.最后,通过对Rayleigh声流问题进行非定常数值模拟,得到的模拟结果与其他研究学者结论完全一致.这说明,采用二维弱可压缩方程组及时空守恒元解元方法,能够模拟水中的声波传播及耗散过程.

针对CAD-CFD-CAE融合的创新设计理念在能源动力学生创新能力培养中的可行性进行了分析。首先分析了CAD-CFD-CAE融合的创新设计理念对培养能源动力学生的创新能力的重要性;接着从教学内容、实践教学、教学平台以及激励机制等方面分析了进行教学改革的必要性;最后从课程设置、教学团队以及硬件条件等方面分析了实施教学改革的可行性。

Acoustic streaming generated by standing waves between two-dimensional flat plates is analyzed based on multiple relaxation time lattice Boltzmann method (MRT-LBM).Simulation results are in good agreement with approximate analytical solutions of Rayleigh streaming.Effects of viscosity and width between two plates on acoustic streaming in standing waves are studied.Dimensionless velocity distributions of x component with different viscosity at x =L/4 and dimensionless velocity distributions of y component with different viscosity at x =L/2,and effects of widths on boundary layer vortex thickness and formation progress of acoustic streaming in standing waves are revealed.It demonstrates availability of MRT-LBM model in simulating acoustic streaming in standing waves.

风力机叶片在旋转过程中受重力和离心力作用,产生动力刚化导致固有频率增加.文章以NRELPha-seⅥ风力机叶片为对象,在其内部分别添加圆形腹板、单腹板和双腹板,建立3种不同截面的叶片三维模型,并结合复合材料对叶片铺层进行动力学分析.结果表明,叶片采用的铺层方案能有效避免共振,并且3种叶片模型的重量均接近叶片的真实值.在额定转速下,3种腹板叶片的一阶频率增量随腹板的厚度增加而增加,但在两倍额定转速时,单腹板和圆形腹板的一阶频率增量随腹板厚度增加而减少;同时,腹板中的双轴向玻璃布材料以±45°铺设时,一阶固有频率最大,而由动力刚化引起的一阶频率增量较其他角度小.

利用时空守恒元解元(CE/SE)算法求解二维无量纲非稳态可压缩Navier-Stokes方程,将CE/SE算法应用于二维平板内驻波声流的计算,对求解出的流场与理论近似解作对比分析,在此基础上通过对比自然对流换热来研究驻波声流对二维平板内对流换热的影响.分析结果,CE/SE算法能很好的计算出声流流场,计算得到的声流流场与理论近似解相吻合,同时得到了声流流场下的温度场.结果表明了驻波声流能明显的加快对流换热,有效的降低流场中的热量.

对塔影效应简化模型进行了试验研究，将上游叶轮简化为二维平板，下游塔架简化为二维圆柱，借助压力传感器及粒子图像测速技术，测量了8种平板位置下圆柱表面压力分布及干涉区域瞬态速度场数据，分析了圆柱表面压力、升力因数、阻力因数、干涉区域时均速度场及其相干特性。研究结果表明：靠近平板的圆柱一侧受平板影响较为明显，且随平板远离圆柱，圆柱表面压力因数与无平板时压力因数的相关系数自－0．37增至0．97，圆柱所受合力因数在原值（1．12）的72％与115％之间。

文章基于非对称翼型NACA4415,以功率系数为依据,以CFD仿真为手段,研究了在不同尖速比下叶片数与叶片弦长对升力型垂直轴风力机气动性能的影响,以及不同尖速比和叶轮实度不同时,垂直轴风力机功率系数的变化.研究结果显示,该类升力型垂直轴风力机叶轮实度取0.25～0.45,尖速比λ取2.5 ～3.4时,具有较高的功率系数.流场分析表明,当叶片弦长与叶片数的变化对流场的扰动能力小于垂直轴风力机从流场中获取风能的能力时,叶片弦长与叶片数的变化会增加垂直轴风力机的功率系数;反之,垂直轴风力机的功率系数降低.该研究为此类20 kW垂直轴风力机的设计与选型提供依据.

MaPFlow-Actuator line模型是一种通过引入体积力代替叶片的致动线技术与三维Navier-Stokes方程相结合来对风力机周围流场进行数值模拟的方法.采用MaPFlow-Actuator line模型对NREL phase Ⅵ风力机叶轮进行数值模拟,将叶片载荷分布与实验数据进行对比,结果表明该模型较为准确.

In this paper,3D unsteady numerical simulation of NREL Phase VI wind turbine was performed.Computations are presented on an isolated rotor and the upwind configuration of the turbine which includes rotor and tower.Different incoming wind velocities (7,13,and 20 m/s) are selected for unsteady simulation with ICEM Hexa mesh.The results show that an abrupt drop oftorque can be observed around 180° azimuthal angle,but the influence is small comparing with the rotor torque output.A hysteresis of blade-tower interaction is also observed with the increasing of wind speed.The variation of section loads and flow field on isosurface provide a reference for the optimization of blade tower interacton.