针对浮选回归模型精度和适应性较差的问题,提出了一种基于遗传算法优化反向传播神经网络(GA-BP)的尾煤水灰分视觉检测方法.对尾煤水图像进行预处理,在去除主要噪声干扰和保证一定彩色特征完整的前提下,提取不同颜色空间的彩色特征、灰度特征以及浓度特征值;以上述特征值为输入变量,以尾煤水灰分作为输出变量,建立基于遗传算法优化BP神经网络(GA-BP)的回归模型.该模型较好地实现了尾煤水灰分的在线检测,预测精度达97.3%,均方误差降低至0.23,提高了精煤产率和经济效益.
为提高蒸发镀膜机内基板温度均匀性,以蒸发镀膜机为研究对象进行建模,采用COMSOL有限元软件对真空环境下的基板加热器温度场进行有限元仿真,并耦合PID算法用于温度控制,揭示了基板温度分布规律.在此基础上对基板温度均匀性的影响因素进行研究,并借助Nelder-Mead单纯形法对加热系统进行优化.结果表明,当顶加热丝间距与侧加热丝间距为1∶1.23、加热距离为440 mm、反射距离为3.56 mm时,基板表面最大温差由优化前的16.5℃降低至9.9℃,温度均匀性得到改善.最后,结合实测数据分析显示,仿真结果与实测数据误差为6%,验证了优化的有效性.
以桑迪亚国家实验室公开的SNL-13.2 MW半潜浮式风机计算模型为研究对象,采用Openfast软件对浮式风机进行"气动—水动—控制-弹性"全耦合时域仿真模拟,研究分析变化波高与风速对13.2 MW风机运动特性产生的影响,并与同工况下10 MW风机的运动特性进行比较.研究发现:在不同湍流风不规则波作用的工况下,SNL-13.2 MW半潜浮式风机系统的纵荡、纵摇、垂荡响应幅值对波高变化不敏感,对风速变化较敏感,与同工况下10 MW风机的运动响应差异较小;在恶劣工况下,风机系统的波动幅值基本符合风机安全工作规范要求,满足风机的安全运转条件.
为研究地铁车站基坑围护结构的力学性状及周边土体变形情况,运用ABAQUS软件建立地铁深基坑开挖全过程的三维数值模型,在充分考虑材料非线性的基础上,结合特定断面分析基坑开挖全过程中支护结构和基坑的变形情况.结果表明:坑底竖向位移与开挖深度呈正相关,整体表现为"鼓肚子"的变形形式;随着开挖深度的增加,支护结构侧向位移先增大后减小,最大位移出现在墙体顶部以下0.7H～0.8H处,且端头井处侧向位移明显小于标准段;基坑开挖导致地连墙周边10 m范围内出现明显的地面倾斜,会对既有建筑物造成不利影响,距离基坑边缘30 m以外的地面倾斜则不会对地面建筑造成明显影响.
对自润滑齿轮齿面沟槽的电解加工流场进行分析及优化,选取加工过程中电解液的流道为研究对象,建立工件电化学腐蚀以及加工间隙内流体流动的物理模型,对电场及流场进行耦合,并数值模拟齿面沟槽的加工过程及对应状态下电解液的流动情况,分析了电解液流动形式及导流段、出口背压、阴极结构对流场的影响,得出如下结论:正流式流道的加工间隙内,电解液的均匀性较好;在电解液出口处增加导流段能有效提升电解液流速;在电解液出口处施加背压,流场均匀性得到显著提高;将阴极端头处的直角转换为圆角,能够提高电解液流动的稳定性,在阴极端头处圆角半径为0.3 mm时,加工间隙的流场均匀性最好.
声屏障作为高速公路的附属设施,其屏蔽效果好坏直接影响沿线居民的身心健康.本文首先对高速公路常见的声屏障整体结构形式及屏体材质进行分类与性能的对比分析,并以福建省南平市典型的桥梁式声屏障结构为例,进行力学性能的验算分析,最后从结构、景观、材质选取等方向展开设计思考.
为提高无线传感器网络节点的覆盖率,提出一种基于改进微分进化算法的网络覆盖优化方法.首先,针对无线传感器网络覆盖模型中传感节点跳出覆盖区域及传感节点易于重叠的问题,设计了改进的覆盖模型;其次,为提高微分进化算法应用于WSN覆盖问题上的收敛速度和寻优能力,提出一种改进的选择优化机制;最后,将改进的覆盖模型和改进的微分进化算法应用于WSN的覆盖优化中,发现采用改进覆盖模型以及改进微分进化算法后,无线传感器的网络覆盖率大大提高.
以湖北省五峰土家族自治县湾潭镇某T梁桥的新建项目为工程背景,利用Midas FEA软件建立三维有限元模型,分析了预制T梁的温度场和应力场,研究了混凝土配合比、昼夜温差、寒潮对预制T梁温度应力的影响.结果表明:预制T梁的最高温度出现在梁端肋板的马蹄部位内部,温度应力最大值尚在混凝土抗拉强度的控制范围以内;使用外加剂、掺合料可以一定程度上降低预制T梁的温度应力,降低开裂风险;昼夜温差越大,预制T梁的温度应力超过混凝土抗拉强度的幅度以及对应的龄期范围越大;寒潮会造成预制T梁温度应力的显著升高,且寒潮出现越早预制T梁产生的开裂风险越高.
研究了水泥剂量、集料级配、含水率和压实度对水泥稳定碎石基层不同龄期无侧限抗压强度的影响,分析了影响无侧限抗压强度的关键施工参数及规律,并依托河北京德高速公路永久路面半刚性基层建立施工参数变异水平与无侧限抗压强度的响应面模型.研究结果表明:水泥剂量为水稳碎石基层各龄期无侧限抗压强度的最显著影响因素;施工参数变异水平对7 d无侧限抗压强度的影响排序为水泥剂量均值>压实度均值>水泥剂量变异系数>压实度变异系数;7 d无侧限抗压强度与施工参数及变异性的响应面回归模型的F值为203.82,P值小于0.0001,精度较高,可用于实际工程半刚性基层无侧限抗压强度的预测和控制.
图的Aα-矩阵是图的度对角矩阵和邻接矩阵的凸线性组合,是图的邻接矩阵和无符号拉普拉斯矩阵的共同推广,其最大特征值称为图的Aα-谱半径.对于α ∈[1/2,1,),本文确定了围长给定的n阶双圈图的Aα-谱半径的上界和极图,推广了已有的成果.
针对文献[1]第166-177页例4.25系统根轨迹的复数部分(不包括实轴上的部分)是一个圆周,通过严格的数学推导和证明,不仅发现二阶系统的根轨迹形状与圆周有重要联系,而且还精确绘制了所有二阶系统的根轨迹(非概略绘制),并通过一些仿真实例进行了验证.
针对真空预压法加固软黏土地基存在深层土体加固效果差的问题,提出掺砂改良软黏土压缩和渗透特性的方法.通过室内掺砂软黏土的一维固结渗透试验,探讨了掺砂率对软黏土压缩性状和渗透性状的影响规律,建立了掺砂土体的归一化压缩公式,得出了如下结论:不同掺砂率土体的压缩曲线形态相近;与初始孔隙比相比,1600 kPa荷载下的孔隙比减小幅度均在45%左右;可根据参数e/eL对压缩曲线进行归一化处理;孔隙比相同时,渗透系数随着掺砂率的增大而增大;掺砂土体的压缩系数、压缩指数和渗透指数均随掺砂率的增加呈幂函数减小规律.
针对等离子体浸没式离子注入(PIII)过程中,无法对实际注入目标基板的离子剂量进行检测和显示的问题,设计了一套基于PIII设备的离子剂量检测装置.该装置围绕注入基板放置离子检测装置,等离子体注入过程中,通过信号调理电路和LabVIEW数据采集卡对检测信号进行分析处理,以实现等离子体注入剂量的实时监测.实验结果表明,基于PIII设备的离子剂量检测装置能够有效收集等离子体进行检测,检测结果可以为腔室结构优化和注入工艺改良提供参考依据.
In order to study the anti-synchronization problem of a class of inertial memristor neural networks（IMNNs） with unbounded distributed time delays, according to the anti-synchronization error system, a suitable adaptive switching controller is designed using the non-reduced-order method, and then the appropriate Lyapunov functional is constructed. Sufficient conditions for the anti-synchronization of the system are obtained using inequality techniques. Finally, the effectiveness of the designed adaptive switching controller is verified by numerical simulation.
In order to study the characteristics and laws of hydrodynamic interference in double rotor vertical axis tidal current turbines, computational fluid dynamics methods were used to numerically simulate the hydrodynamic forces of dual rotor turbines under steady flow.By studying the energy utilization ratio, velocity field evolution, wake field evolution of double-rotor turbines with different turbine spacing and rotor rotation directions under typical speed ratio and the comparative analysis of single-rotor turbines under corresponding working conditions, it is found that the energy efficiency of double-rotor turbines is higher than that of single-rotor turbines at high speed ratio, reaching as high as 36%, and the energy enhancement effect of double-rotor turbines decreases with the increase of turbine spacing.Due to the existence of interference effects, the blade surface of a double rotor turbine produces a larger high-speed region, and the tail vortex is also more complex and chaotic than that of a single rotor turbine.The tail vortex has a tendency to deflect and merge, and decreases as the turbine spacing increases.
A monocular vision-based industrial robot grasping system is designed to address the problem that the industrial robot cannot grasp workpieces with deviations in position when loading in the teaching mode. The system uses monocular vision to collect the workpiece image, and uses nine-point calibration method to convert the centroid coordinates of the workpiece in the image into the coordinates in the robot coordinate system. The rotation angle of the workpiece is calculated by image processing,which guides the robot to move to the target position and cooperates with the gripper to grasp the workpiece. Finally, the grasping system is built with PCB as the object. After several grasping experiments, it is found that the grasping system can meet the requirements of adaptive grasping workpiece positioning with the position error within 2 mm and attitude error within 1.5°.
Aiming at the static deviation problem existing in the traditional input current sharing droop control strategy of IPOSDAB converter, an improved input current sharing droop control strategy is proposed. According to the reason for the static deviation of the traditional input current sharing droop control strategy, in the traditional input current sharing droop link, the module input current average value is introduced to reduce the output voltage drop, so as to realize the voltage/current sharing of each module while reducing the static deviation of the system. Finally, the simulation verification is carried out through MATLAB/Simulink. The results show that the improved input current sharing droop control strategy can make the system operate normally and stably with excellent voltage equalization under the conditions of sudden changes in input voltage, sudden changes in output load and different inductance parameters.
Taking the anti inclined slope on the north side of Louxiang Highway as an example, a numerical model is established by using the finite difference software FLAC3D, and the deformation and failure characteristics of the reverse slope under different anchoring parameters and the reinforcement mechanism of anchoring force are analyzed, revealing the response relationship between each parameter and the sliding deformation characteristics of the anti inclined slope. The results show that the bolt reinforcement at the bottom of the slope plays a major control role in the deformation of the shear slip type slope, and the slope has the highest anti-sliding capacity. The longer the bolt, the smaller the deformation of the slope. The anchoring force of the bolt increases with the increase of the bolt length, and the stability of the slope is improved. With the bolt length and anchor position unchanged,the increase of anchor angle can increase the anchor force of slope and improve the stability of slope,but the impact is small.
The Aiming at the problem that k value needs to be set manually and the initial clustering center needs to be randomly selected in the roller fault detection of Kmeans algorithm, a fault detection method of I-Canopy-Kmeans algorithm is proposed to optimize it.This algorithm uses the principle of "farthest closest" in the random selection of initial clustering centers, that is, when obtaining n Canopies, the distance between any two Canopy center points should be as far as possible, and the nth Canopy center point should be the smallest of the farthest distances between other data points and the previous n-1 center points.In terms of threshold selection, Euclidean distance is used to calculate the mean point of all data points, and then the distance from the mean point to all data points is calculated. L 1 and L 2 are used to represent the farthest and closest distances respectively. Then,（L 1 +L 2 ）/2 is assigned to threshold T 1 , and（L 1 +L 2 ）/3 is assigned to threshold T 2 . The experimental results show that, compared with the traditional Kmeans algorithm, the I-Canopy-Kmeans algorithm has improved in all evaluation indexes, among which IAR has improved the most, reaching 40.01%.
In order to study the inertial lateral migration of a single particle in a channel laminar flow field, a numerical simulation of the inertial lateral migration of a neutral single particle in laminar flow field was carried out by taking circular and square tubes as examples. The particles are placed in several positions on the entrance cross sections of round and square tubes and released.The discrete particle model was used to calculate the migration trajectories of the three kinds of spherical particles with particle sizes d=0.8, 1.0 and 1.2 mm in the flow field and the equilibrium positions reached after lateral migration under the condition of Reynolds number Re=100, 125 and 150. The results show that for the circular tube, when the particles are released at the initial radius of 0.7R and 0.8R（R is the radius of the circular tube）, the equilibrium positions of inertial migration are all around 0.6R.With the increase of particle size, the equilibrium position of particle migration will shift towards the pipeline axis. With the increase of Re, the equilibrium position of particle migration shifted towards the tube wall. For the square tube, the initial release position of particles is set as z0=0.22L and 0.33L（L is the side length of the square tube section）. When the release position of particles is closer to the channel axis, the lateral migration distance of particles is shorter. With the increase of Re, the equilibrium position of particles will be more inclined to the center of channel. The equilibrium position of small particles will be closer to the wall than that of large particles.
