光伏并网逆变器软故障诊断方法研究

发布时间：2018-03-13 13:07

本文选题：SOM神经网络　切入点：并网逆变器　出处：《安徽理工大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着快速发展的生产社会,致使能源需求不断增长,在全球范围内的能源短缺的危机变得日益严峻。化石能源在地球中是一种非再生资源,终究会致使其耗尽。化石资源也会随着资源数量的不断减少,出现价格会相应的提高,严重制约着人民生活水平和社会的生产发展的提高。可再生能源在全球范围内的能源需求有着相当重要的作用,在我们国家这样的人口大国就更加显得尤为重要。太阳能能源在我们国家非常富足,具有相当好的发展前景。将太阳能光伏阵列发出的直流电通过光伏逆变器变换成能够满足市电使用条件的交流电然后直接并入公共电网称之为光伏并网发电系统。当阳光充足时则不需使用市网电,可选择采用太阳能发出的电；当太阳能发电达不到使用量标准或阳光不充裕时,其系统可通过其控制部分切换到市网电给予供电。这种系统的最重要部分是光伏并网逆变器,其中光伏并网逆变器的可靠运行决定着其系统运行的安全性。由若干开关器件及电感、电容组合而成的并网逆变器主电路,在逆变电路运行过程的故障可划分成硬故障(结构性故障)和软故障(参数性故障),其中软故障对电路的可靠性具有很大的影响,其中较为突出的是滤波电容的软故障。由电容的引出电极长时间经受大电流的充、放电冲击造成电容出现不同程度的氧化现象,电容储电能力也会随着减小,也就是相当于电容容量大幅衰减。由于电容容量的变小,通常体现为带负载能力减弱,当负载增加时则会发生直流回路欠压的故障。从而导致更大程度上损坏电容,也有能够导致直流回路产生电压振荡,严重损毁逆变模块。本文主要分析逆变器中的电解电容在不同容差下的软故障特征,根据光伏并网逆变器特点,提出了一种光伏并网逆变器主电路拓扑结构中电解电容软故障诊断方法,并利用MATLAB搭建了光伏并网逆变器的仿真模型。因为逆变器的电解电容随着运行时间的推移出现不同程度的老化现象,致使电容值减小,导致电网中输出的线电压偏离正常工作时的电压,故本文采用逆变器三相输出电压信号作为特征量,并采用小波变换的方法对其进行处理。再利用BP神经网络和SOM神经网络分别对逆变器进行软故障诊断,通过仿真实验得出诊断结果,并将这两种诊断方法进行比较,得到SOM神经网络在逆变器软故障诊断方面更具有一定优势。
[Abstract]:With the rapid development of productive society, energy demand is increasing, and the global crisis of energy shortage is becoming more and more serious. Fossil energy is a kind of non-renewable resource in the earth. It will eventually be depleted. The fossil resources will also increase as the amount of resources continues to decrease, and the prices will rise accordingly. It seriously restricts the improvement of people's living standards and the productive development of society. Renewable energy sources play a very important role in the global energy demand, This is all the more important in a populous country like ours. Solar energy is very rich in our country. The direct current generated by the solar photovoltaic array is converted by the photovoltaic inverter into alternating current which can meet the requirements of the municipal electricity supply and is directly incorporated into the public grid as the photovoltaic grid-connected power generation system. When there is plenty of sunshine, there is no need for market electricity, The option is to use electricity from solar energy; when solar power does not meet usage standards or when there is not enough sunlight, its system can be switched to market electricity through its control portion. The most important part of such a system is a photovoltaic grid-connected inverter. The reliable operation of photovoltaic grid-connected inverter determines the safety of the system. The main circuit of grid-connected inverter is composed of several switching devices, inductors and capacitors. During the operation of the inverter circuit, the faults can be divided into hard faults (structural faults) and soft faults (parametric faults), in which soft faults have a great impact on the reliability of the circuit. Among them, the soft fault of the filter capacitor is prominent. The discharge shock causes the capacitance to be oxidized in different degree, and the capacitance storage capacity will decrease with the discharge shock. Because the capacitance capacity becomes smaller, the capacity of the capacitor is usually reflected in the weakening of the capacity of the load. When the load increases, the failure of the DC circuit will occur. This will result in more damage to the capacitance. It can also cause voltage oscillation in DC circuit and seriously damage inverter module. This paper mainly analyzes the characteristics of soft fault of electrolytic capacitance in inverter under different tolerances, according to the characteristics of photovoltaic grid-connected inverter, A soft fault diagnosis method for electrolytic capacitance in the topology of the main circuit of photovoltaic grid-connected inverter is presented. The simulation model of grid-connected photovoltaic inverter is built by using MATLAB. Because the electrolytic capacitance of the inverter is aging in varying degrees with the running time, the capacitance value decreases. The output line voltage in the power network deviates from the normal working voltage. Therefore, the inverter three-phase output voltage signal is used as the characteristic value in this paper. The soft fault diagnosis of inverter is carried out by using BP neural network and SOM neural network, and the results are obtained by simulation experiment, and the two diagnosis methods are compared. SOM neural network has some advantages in soft fault diagnosis of inverter.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM464

【参考文献】