8 resultados para Modelica
Resumo:
为了提高现有OpenModelica软件中对DAE系统的预处理模块中,求强连通分量与拓扑排序部分的性能,提出了基于Kosaraju算法实现的策略。首先阐述Modelica软件的实现原理,叙述拓扑排序相关算法在其中的重要性,再分析现有Modelica软件中使用的求强连通分量与拓扑排序部分的算法,然后比较Tarjan算法的实现方案与Kosaraju算法实现方案。通过实现两种方案,并对实验结果进行比较和分析,验证了Kosaraju算法方案的可行性和有效性。
Resumo:
随着产品设计的复杂化,应用领域中的数学建模和仿真越来越重要,传统建模方法基于赋值语句,主要考虑单一系统,工程人员需要对程序设计语言和算法求解有相当程度的熟悉,这导致了传统建模方法很难满足复杂产品设计的需要。欧洲仿真界的学者在总结现有建模方法的基础上于1997年推出了一种面向对象的、陈述式的基于方程的语言——Modelica,Modelica语言支持多领域建模,Modelica语言得到了仿真领域众多厂商的支持,成为统一建模领域的事实标准,实现Modelica仿真环境具有重大意义。 编译器是Modelica仿真环境中的核心组件,Modelica编译器会产生高指标DAE系统,除了部分特殊结构的高指标DAE,现有求解器一般不能对通用高指标DAE直接求解,现在一般做法是先对高指标DAE进行指标约简,将其转换成等价的低指标DAE,然后进行求解。这就要求Modelica编译器具有指标约简的功能,以便利用现有求解器进行求解。 本文介绍了几种现有的指标约简算法,给出了各个算法指标约简原理和约简步骤,并对这些算法进行了分析和比较,得出各自的优缺点。最后,本文设计并实现了一个指标约简系统,采用三个Modelica模型产生的方程对系统进行了测试,并验证了实验结果的正确性。
Resumo:
Modelica语言仿真建模在科研工作中已经得到了广泛应用。它能方便地对包含机械、电子、液压、控制、热流等领域的复合物理系统进行基于组件的仿真。现有基于Modelica语言的仿真建模软件支持图形化建模和文本建模两种方式,集成了面向对象、陈述式描述、统一建模、组件重用的优势,给科研工作带来了巨大的便利。 Modelica软件仿真的过程可归结为微分代数方程(differential algebraic equation,DAE)系统的求解。在求解DAE系统时,需要对DAE系统进行约简,直到庞大的DAE系统约简为目前自动求解方法成熟的ODE系统,或约简为方程个数不多的、指标较低的DAE系统,才能使Modelica建模仿真软件具有工业上的应用价值。在约简DAE系统之前,需要对之进行预处理,根据方程之间的数据依赖关系进行拓扑排序,确定求解顺序。排序的过程对应着将DAE系统结构关联矩阵进行块状下三角(block lower triangle,BLT)变换。寻找强连通分量和拓扑排序是对DAE系统进行预处理的重要组成部分。 本文剖析了Modelica软件在仿真时的运行机制,使用几个实例来详细描述在仿真过程中,Modelica软件完成的工作。在寻找强连通分量和拓扑排序这一步,本文提出了使用Kosaraju算法的策略,对由模型得到的有向图直接使用Kosaraju算法,得出DAE系统的求解顺序。文章叙述了强连通分量的含义,并阐述了在求强连通分量时的理论依据,由此引出了Tarjan算法和Kosaraju算法,再分析和比较Tarjan算法和Kosaraju算法,对比了两种策略的优劣,并进行了实验。同时,本文分析了OpenModelica软件包的结构,修改了软件包在寻找强连通分量及拓扑排序相关模块的代码。
Resumo:
Modelica建模软件会产生高指标的微分代数方程(DAE),因为现有数值求解器一般不能直接对高指标DAE进行求解,所以Modelica建模软件必须对高指标DAE进行指标约简。为提高Modelica建模软件性能,针对现有指标约简算法,展开了理论分析和实验比较,着重对负权二部图算法进行了详细分析,并给出了负权二部图算法的时间复杂度。理论分析和实验结果表明,负权二部图算法相对Pantelides算法有较好的性能,因此可以考虑在Modelica建模软件中使用负权二部图算法来提高性能。
Resumo:
The software package Dymola, which implements the new, vendor-independent standard modelling language Modelica, exemplifies the emerging generation of object-oriented modelling and simulation tools. This paper shows how, in addition to its simulation capabilities, it may be used as an embodiment design tool, to size automatically a design assembled from a library of generic parametric components. The example used is a miniature model aircraft diesel engine. To this end, the component classes contain extra algebraic equations calculating the overload factor (or its reciprocal, the safety factor) for all the different modes of failure, such as buckling or tensile yield. Thus the simulation results contain the maximum overload or minimum safety factor for each failure mode along with the critical instant and the device state at which it occurs. The Dymola "Initial Conditions Calculation" function, controlled by a simple software script, may then be used to perform automatic component sizing. Each component is minimised in mass, subject to a chosen safety factor against failure, over a given operating cycle. Whilst the example is in the realm of mechanical design, it must be emphasised that the approach is equally applicable to the electrical or mechatronic domains, indeed to any design problem requiring numerical constraint satisfaction.
Resumo:
为了应对设计复杂产品的需求,研究者们提出了一些面向对象、基于组件的建模语言,比如说Modelica 语言。Modelica语言建模中采用了分治的思想,将大型的系统划分成一系列小型组件,极大程度地提高了模型复用性,简化了建模的过程。 Modelica语言构建的模型往往会被转化成一个DAE方程组。在一些情况中,这个DAE是指标数大于1的高指标问题。由于目前对高指标问题不存在通用的求解器,为了求解这类问题,通常的方法是先对问题进行指标约简。 指标约简有两类主流的方法,一类基于微分指标,如Gear算法;另一类基于结构指标,如Pantelides和哑导方法。Gear方法是解决一般高指标问题最为经典的方法,对它的研究和深入分析意义重大。在本文的第一部分,作者提出了一种Gear方法的修正实现。实验结果表明,对于一类特殊结构的DAE,修正实现比起经典实现做了更少的微分,最终得到了规模更小的方程组。 本文另一部分工作集中在对结构指标修正的方面。相比Gear算法,基于结构指标的指标约简算法是一类快速算法,但是并非百分之百有效。在少数情况下,当微分指标与结构指标不一致时,这类方法会失效。为了提高结构指标的适用性,必须首先消除这种不一致性。因此,作者深入分析了处理这一问题的组合松弛型算法,并做出了相应的实现。利用组合松弛型算法,作者进一步对几个Modelica模型导出的DAE方程组进行了处理。实验结果表明,作者所实现的算法解决了微分指标与结构指标不一致的问题。
Resumo:
Photovoltaic (PV) solar panels generally produce electricity in the 6% to 16% efficiency range, the rest being dissipated in thermal losses. To recover this amount, hybrid photovoltaic thermal systems (PVT) have been devised. These are devices that simultaneously convert solar energy into electricity and heat. It is thus interesting to study the PVT system globally from different point of views in order to evaluate advantages and disadvantages of this technology and its possible uses. In particular in Chapter II, the development of the PVT absorber numerical optimization by a genetic algorithm has been carried out analyzing different internal channel profiles in order to find a right compromise between performance and technical and economical feasibility. Therefore in Chapter III ,thanks to a mobile structure built into the university lab, it has been compared experimentally electrical and thermal output power from PVT panels with separated photovoltaic and solar thermal productions. Collecting a lot of experimental data based on different seasonal conditions (ambient temperature,irradiation, wind...),the aim of this mobile structure has been to evaluate average both thermal and electrical increasing and decreasing efficiency values obtained respect to separate productions through the year. In Chapter IV , new PVT and solar thermal equation based models in steady state conditions have been developed by software Dymola that uses Modelica language. This permits ,in a simplified way respect to previous system modelling softwares, to model and evaluate different concepts about PVT panel regarding its structure before prototyping and measuring it. Chapter V concerns instead the definition of PVT boundary conditions into a HVAC system . This was made trough year simulations by software Polysun in order to finally assess the best solar assisted integrated structure thanks to F_save(solar saving energy)factor. Finally, Chapter VI presents the conclusion and the perspectives of this PhD work.