基于数据驱动和故障模型的光伏阵列故障诊断方法研究

The performance of solar power stations depends on the working condition of solar panels，but solar panel system builts on normally used series-parallel structure complicates its fault diagnosis and localization. The most directions of current researches focus on the offline and regional fault detection and efficient online detection method used for the common faults of PV array is seldom discussed or implemented in public. In order to solve the problem of online detection and positioning of solar panel system, the proposed research based on three related national patents, which are obtained by the applicant, will be conducting on system connection structure optimization, information extraction, fault modeling, and fault diagnosis methods. Firstly, in order to achieving the requirement of getting essential criterion with the minimum number of sensors, photovoltaic panel array structure and the optimized strategy of sensor access mode will be investigated; Secondly, on the basis of analyzing the fault mechanism of photovoltaic battery, photovoltaic battery modeling of nonlinear faults knowledge discovery and the optimization research of parameters matching will be carried out; Finally, using data driven and information fusion theory to research on-line fault diagnosis and localization on PV array, studying on multi-source information basis knowledge discovery and troubleshooting algorithms to achieve measurement - modeling - diagnosis of the three aspects of critical theory and technology breakthroughs, which will establish theoretical foundation for the large scale use of solar power and realize solar energy from complementary energy to main energy shift in the period of "The twelfth Five-Year Plan"

太阳能电站性能的优劣取决于太阳能电池板工作状态，而太阳能电池板系统常用的串并联结构使其故障诊断及定位变得复杂。目前该方向的研究多注重离线和区域的故障检测。为了解决太阳能电池板系统在线故障检测及定位问题，本项目在申请人三个国家专利的基础上，对太阳能电池板系统连接结构的优化、信息提取、故障建模和故障诊断方法开展研究。首先，为了达到以最少的传感器数量得到必需的判别依据，拟开展光伏电池板阵列结构及传感器接入方式优化策略的研究；其次，分析光伏电池的故障机理，进行光伏电池非线性故障模型建模和参数匹配优化的研究；最后，基于数据驱动和信息融合理论，利用故障模型复杂非线性约束优化问题的可行域求解理念，对光伏电池板阵列在线故障诊断和定位问题进行复合算法研究，实现测量-建模-诊断三个环节关键理论和技术问题的突破，为实现"十二五"期间太阳能从补充能源到主要能源转变的目标奠定理论基础。

太阳能电站是太阳能有效且重要的应用，其性能的优劣取决于光伏电池板工作状态。如何对光伏电池板阵列进行在线监控及故障诊断已经成为光伏发电大规模应用一个亟待解决的问题。要实现光伏阵列的故障诊断，首先要解决如何在通用的光伏阵列结构上以最少的传感器将必须的信息提取出来，而太阳能电池板系统常用的串并联结构使其故障诊断及定位变得复杂。目前该方向的研究多注重离线和区域的故障检测。.为了解决太阳能电池板系统在线故障检测及定位问题，本项目对太阳能电池板系统连接结构的优化、信息提取、故障建模和故障诊断方法开展研究。为了达到以最少的传感器数量得到必需的判别依据，开展了光伏电池板阵列结构及传感器接入方式优化策略的研究；其次，分析光伏电池的故障机理，进行光伏电池非线性故障模型建模和参数匹配优化的研究；最后，基于数据驱动和信息融合理论，利用故障模型复杂非线性约束优化问题的可行域求解理念，对光伏电池板阵列在线故障诊断和定位问题进行复合算法研究，实现测量-建模-诊断三个环节关键理论和技术问题的突破。.本项目的重要成果有：提出了一种光伏电池老化故障模型的建模方法及电池模型非线性参数的求解方法，初步解决了光伏电池模型在不同环境下参数值的求解问题；首次提出了有发明专利支撑的光伏阵列故障检测连接结构，可实现在不影响光伏发电输出功率情况下对故障点的快速诊断；研究了一种在故障诊断中相关的问题，将预测日天气参数转变为光伏电池模型参数预测算法；提出了有发明专利支撑的锂电池状态估计及二阶模型参数在线辨识方法，提高了锂电池SOC、SOH和SOP估计的精度，为锂电池在太阳能发电的能源存储的应用问题奠定了基础。