Vision-based optimization of the generalized predictive active disturbance rejection controller

The batching system of the integrated mixing and spreading equipment for MOH material is a nonlinear system with large uncertainty. It is difficult for conventional control strategies to meet the requirements for system performance. This research combines generalized predictive control and active disturbance rejection technique to propose a new generalized predictive active disturbance rejection controller (GPADRC) used in the batching system of MOH material. For the nonlinearity and uncertainty of the batching system, the extended state observer in the active disturbance rejection technique is used for estimation and compensation. The batching system model is converted into an integrator form, based on which the use of generalized predictive control can greatly reduce the impact of nonlinear models and uncertainties on the controller. Aiming at the problem that the parameters of the proposed new controller are numerous and difficult to tune, the adaptive genetic algorithm is used to realize the automatic tuning of the parameters. The simulation experiment shows that the designed GPADRC can well adapt to the working conditions of the batching system and can meet the requirements for various control indicators. At the same time, the adaptive genetic algorithm can realize the rapid tuning of the controller parameters, which reduce the difficulty and time consumption of the tuning process, and improve the applicability and achievability of the designed controller.     

基于改进SIFT特征提取的车标识别

为减少SIFT车标识别算法中检测极值点的冗余以及各种图像变化因素的不利影响,提出了基于边缘约束和全局结构化的改进SIFT算法。利用图像不变矩理论及图像边缘检测算法只对目标图像的边缘区域检测,剔除与车标识别区域无关的极值点;同时将特征点邻域划分为圆形并计算出同心圆内像素点最大曲率来构建全局SIFT组合特征向量,使SIFT描述子具有全局描述特性;并结合SVM模型作为车标图像特征向量的分类器进行特征分类、识别。仿真实验结果表明:改进的SIFT算法可以减少冗余极值点约25%~45%,提高了检测极值点的有效性;使车标平均识别率达到97%以上,改善了识别实时性。改进SIFT的车标识别方法与几种常用的图像特征提取算子相比较具有识别率高、识别速度快的优点。     

The structural evolution and mechanical properties of a Zr-based bulk metallic glass during superplastic gas pressure forming

The structural evolution and mechanical properties of a Zr-based bulk metallic glass during superplastic gas pressure forming were investigated by differential scanning calorimetry (DSC), transmission electron microscopy (TEM) and nanoindentation measurements at the optimum temperature of 676 K under different stress states. It was found that the crystallized volume fraction of the deformed specimens with die aspect ratios of 1:1 and 3:2 are estimated to be 15% and 25%, respectively. The high resolution TEM results show that more particles at the nanometer level exist in the deformed specimen with the aspect ratio of 3:2. The nanoindentation tests further illustrate that the Young’s modulus and hardness of the deformed specimens have increased, and the deformed sample with the aspect of 1:1 shows slightly smaller modulus and hardness. The nano-scale crystalline structure induced by gas pressure forming process is considered to be the main reason for the change of the mechanical properties, and in addition to thermal and strain energy. In addition, the structural evolution is also affected by the stress state in the process of gas pressure forming.     

基于反热传导方程与Sobel算子的图像锐化方法

目的应用偏微分方程理论与经典的图像锐化算予,研究图像锐化的有效方法．方法根据热传导方程的物理意义,把图像的灰度值视为平面物体的温度,推导出反热传导方程,在满足模糊图像是原图像被热传导滤波器作用的结果和参数t足够小的情况下,建立图像锐化模型,利用Sobel算子求解偏微分方程,实现图像锐化．结果数值实验显示应用笔者所提方法对图像进行锐化,得到了相对清晰的图像,锐化后的图像没有信息偏移现象,而且不会出现噪声．同时对图像的边缘和细节特征保持较好．结论笔者所提算法具有长时间计算的稳定性。锐化图像不会随着迭代次数的增加而变得模糊,是一种有效的图像锐化方法．