基于ADPSS的直驱风机半实物仿真研究

高比例新能源设备给电网带来电源不确定性增高和转动惯量变弱等问题,通过含新能源设备精细化模型的大电网仿真可拟合设备运行控制特性,校验控制策略,提前预防电网潜在风险,有效提高电网安全性。新能源设备模型的建模精度影响大电网仿真精度,为解决新能源设备模型的精度问题,利用贴近实际设备的半实物仿真平台开展高低电压穿越等工况仿真。通过使用ADPSS(advanced digital power system simulator,ADPSS)技术,利用相关基础元件建立直流卸荷回路,结合相关两电平换流器控制策略,搭建详细的单电机双功率单元两电平直驱风机半实物仿真电磁模型,并通过直接频率控制方法优化模型,从而增强了半实物仿真的稳定性,结果比对显示,直接频率控制方法优化模型降低了仿真的波动性,可有效提升仿真的准确性。

Simulation of the Interaction between Driver and Seat

Test is one of methods to acquire human-seat pressure distribution in driving, with the deficiency of being uneasy to obtain the stress information of soft tissue inside human body and the sheer force of interface between human and seat, which can be obtained by simulation. But current simulation method focuses mainly on calculation itself other than combining it with posture prediction and cab packaging parameters, which cause it difficult to acquire accurate pressure calculation results without accurate posture of human body, and make it almost meaningless to design optimization. Therefore, a human body geometric model with posture change capability is built and linked up with Cascade Prediction Model（CPM）, which takes cab packaging parameters as inputs. A detailed finite element model of driver human body is constructed and used to conduct the driver-seat interaction simulation between human body and seat. Good accordance of pressure distribution is observed between simulation and test, which validates the simulation. In addition to the distribution pattern, curves on key sections are used to analyze the pressure and shear stress on the seat surface, as well as soft tissue stress inside human body. The simulation shows that the maximum stress of buttocks locates under the ischial tuberosity, and the maximum stress of trunk occurs near the scapula posterior and the lower waist. These are the places where fatigue usually occurs. The maximum pressure of seat appears at the driver-seat contact area corresponding to the driver＇s maximum skin tissue stress. In order to guide the seat design and cab packaging and study the influence of posture to pressure distribution, finite element models for different levels of cab packaging parameters are created by using CPM. The pressure distributions are calculated and their tendencies varying with cab packaging parameters are obtained. The method presented provides a new way to accurately simulate the interaction between driver human body and seat, and to guide the seat design and cab packaging so as to improve seating comfort.

基于多层人工神经网络的电力系统恢复过电压预测

在电力系统恢复的初始阶段,系统元件(变压器、输电线路和并联电抗器)充电时,会出现较高的过电压,可能造成设备损坏,推迟系统的恢复进度,进而威胁系统的安全稳定运行。针对这一问题,提出了一种基于多层人工神经网络的过电压预测方法。采用前馈多层感知器结构,应用Levenberg-Marquardt算法来训练人工神经网络模型,通过MATLAB/Simulink中的电力系统分析工具包PSB来取得训练所需的样本数据。针对变压器和并联电抗器充电的情况,按照合闸角和剩磁最坏的情况来训练人工神经网络,从而减少训练所需的时域仿真计算次数。另外,为了增强所构造的人工神经网络模型的通用性,将电力系统的网络等值参数作为模型的输入数据。采用新英格兰39节点系统的部分网架作为算例,结果表明所提方法可以对电力系统恢复过程中的过电压峰值和持续时间进行有效的预测,并且具有很高的精度。