Extended object model for product configuration design

This paper presents an extended object model for case-based reasoning (CBR) in product configuration design. In the extended object model, a few methods of knowledge expression are adopted, such as constraints, rules, objects, etc. On the basis of extended object model, case representation model for CBR is applied to product configuration design system. The product configuration knowledge can be represented by the extended object. The model can support all the processes of CBR in product configuration design, such as case representation, indexing, retrieving, and case revising. The presented model is an extension of the traditional object-oriented model by including the relationship class used to express the relation between the cases, constraints class used in the product configuration knowledge representation, index class used in case retrieving, and solution class used in case revising. Therefore, the product configuration knowledge used in the product configuration design can be represented by using this model. In the end, a metering pump product configuration design system is developed on the basis of the proposed product configuration model to support customized products.     

基于线电流解耦的三相光伏逆变器控制研究

三相半桥电压源型逆变器存在三相开关控制相互干扰的问题,要实现定频滞环控制,必须对三相开关控制进行解耦。如果采用基于相电流的解耦控制方法,电流误差准确度过分依赖于逆变器交流侧的电感参数。提出基于线电流的解耦控制方法,保持某相开关状态不变,实现另外两相线电流控制的解耦,有效改善光伏逆变器输出性能;并通过采用基于积分法的定频滞环电流控制算法,提高系统控制精度。Matlab仿真结果证明,基于线电流解耦的解耦控制方法,能准确实现电流跟踪;积分滞环电流控制算法在实现定频控制的同时,能有效降低输出电流低次谐波含量,减少稳态误差。     

基于CDIO教育模式软件技术专业项目教学实践

本文探讨基于CDIO工程教育理念进行高职软件技术专业的项目教学。我们分别从课程体系构建、授课体系改革、评价方法改进等方面进行了实践探索,使学生在项目开发和工程实际环境中学习专业知识和工程实践技能,提高人际交往能力及产品、过程和系统建造能力,获得了良好的教学效果。     

项目驱动的软件人才培养方案(英文)

The education of software talent is the basis of the development of China's software industry.This article demonstrates the software training goals for undergraduate students in the School of Software Engineering of Chongqing University,and put forwards the education program of the software talent based on project-driven.The program aims to cultivate software talents of "compound,application-oriented,international" characteristics to meet the development of China's software industry through a series of integrated courses on project practice and training.     

一种记忆功率放大器自适应线性化仿真

为了更好地实现有记忆功率放大器高效率而又高线性的工作.对两种常用的描述功率放大器的记忆效应和非线性特征的模型--W-H模型和记忆多项式模型进行了自适应线性化仿真,采用AM/AM和AM/PM失真特性对由于功放特性不理想而引起的放大信号的振幅畸变及相位偏移进行描述,运用自适应预失真处理方法,进行记忆预失真的模拟,并用误差矢量值(EVM)衡量这种自适应预失真方法的效果.仿真结果表明,文中的自适应预失真方法对上述两种记忆功放模型具有较好的线性化效果.     

可伸缩视频编码的码率控制技术研究

可伸缩视频编码能生成空域、时域、质量可伸缩的视频码流,如何对其进行码率控制来生成最优的可伸缩性码流,对适应不同需求的终端十分重要.结合可伸缩视频编码码率控制的研究现状,描述了最新的研究进展,综述了码流控制的原理、经典算法,分析总结了可伸缩视频编码的关键技术及在可伸缩视频编码中的挑战.通过对增强层中各种码率控制算法的优缺点的分析,总结出增强层的时域可扩展和空域可扩展中码率控制的最优解决方案.最后对可伸缩编码中码率控制技术的发展趋势进行了展望.     

混合模型的用户兴趣漂移算法

针对个性化信息服务中的用户兴趣漂移问题,提出了一种新的正态分布密度曲线遗忘函数,该函数符合用户兴趣遗忘的规律.并且将用户模型定义为长期模型和短期模型相结合的混合模型,其中短期模型使用最近最久未使用的滑动窗口算法进行更新,长期模型采用正态渐进遗忘算法进行更新.实验表明,该方法能够较迅速地发现和准确地跟踪用户的兴趣变化,提高了个性化信息服务的效率.     

PSO并行优化LSSVR非线性黑箱模型辨识

针对非线性黑箱系统辨识中存在不确定性、高阶次,采用常规辨识方法建立其精确数学模型十分困难等问题,提出一种基于自适应粒子群算法的最小二乘支持向量机回归(PSO-LSSVR)非线性系统辨识方法.该方法采用2组自适应粒子群算法并行计算模型,分别利用自适应粒子群算法对LSSVR中的参数进行自动选取和矩阵迭代求解,既克服了传统LSSVR参数难以确定的缺点,提高了辨识精度,同时避免了复杂矩阵求逆运算,加快了计算速度.将该方法应用于船舶操纵性模型非线性系统辨识,仿真结果表明,由该方法得到的LSSVR能够有效地对系统进行建模,仿真精度高,结构简单,具有一定的理论推广意义.     

启发式搜索在数学智能解题系统中的应用研究

在人工智能领域,对问题求解的方法都需要某种对解答的搜索,即为一个搜索过程.文中论述的数学智能辅导系统采用了与或树知识表示方法,也可称为问题规约法.它把初始问题通过一系列变换最终变为一个子问题集合,而这些子问题的解可以直接得到,从而解答了初始问题.系统使用以推理深度作估价函数的启发式搜索,使得问题的求解更加有效与合理.论述了采用启发式搜索的必要性及可行性.对比了采用启发式搜索前后,系统解题合理性得到很大提高.     

Multi-goal Q-learning of cooperative teams

This paper studies a multi-goal Q-learning algorithm of cooperative teams. Member of the cooperative teams is simulated by an agent. In the virtual cooperative team, agents adapt its knowledge according to cooperative principles. The multi-goal Q-learning algorithm is approached to the multiple learning goals. In the virtual team, agents learn what knowledge to adopt and how much to learn (choosing learning radius). The learning radius is interpreted in Section 3.1. Five basic experiments are manipulated proving the validity of the multi-goal Q-learning algorithm. It is found that the learning algorithm causes agents to converge to optimal actions, based on agents’ continually updated cognitive maps of how actions influence learning goals. It is also proved that the learning algorithm is beneficial to the multiple goals. Furthermore, the paper analyzes how sensitive the learning performance is affected by the parameter values of the learning algorithm.