罗建, 孙越,江丽娟.基于改进粒子群算法的光伏逆变器控制参数辨识[J].河南理工大学学报（自然科学版）,2025,44(1):124-133.

LUO J, SUN Y, JIANG L J. Parameter identification of photovoltaic inverter based on improved particle swarm optimization algorithm[J]. Journal of Henan Polytechnic University(Natural Science) , 2025, 44(1): 124-133.

基于改进粒子群算法的光伏逆变器控制参数辨识

罗建1, 孙越2,江丽娟1

1.重庆大学 输配电装备及系统安全与新技术国家重点实验室，重庆  400044;2.国网安徽省电力有限公司 培训中心，安徽 合肥  230022

摘要:精准的光伏并网逆变器模型是研究大规模光伏接入下电力系统故障特性的重要工具。 目的 为了解决现有光伏逆变器仿真模型与实际工作中的光伏逆变器特性相差较大的问题， 方法 提出采用参数辨识的方法构建逆变器的辨识模型。以重庆云阳某1 MW光伏电站为实际参照模型，首先根据实际工作情况将逆变器的工作区间划分为3个阶段，利用数学扰动法分别对3个阶段中的待辨识参数划分灵敏度高低等级，并由此提出不同阶段不同灵敏度参数分步辨识策略；其次，分阶段采集实际光伏电站工作数据，对该数据进行分析处理，获得各待辨识参数的初始取值范围，设计同步辨识参数实验作为参照；最后提出改进的混沌遗传粒子群优化算法（chaos genetic algorithm of particle swarm optimization， CGAPSO）作为辨识算法，分步分工作阶段辨识相关参数，通过对比参数的同步辨识结果，验证所提方法的优越性，并将辨识结果代入仿真模型。 结果 结果表明，低灵敏度参数的同步辨识结果误差远超过可接受范围，而CGAPSO分步辨识出的相关参数误差皆在1.1%以下，精度远高于同步辨识结果。 结论 基于改进粒子群算法构建的辨识模型输出数据与实际逆变器工作数据契合度高，可准确反映逆变器实际工作特性。

关键词:光伏并网逆变器;逆变器控制策略;参数辨识;数学扰动法;改进粒子群优化算法

doi: 10.16186/j.cnki.1673-9787.2023010028

基金项目:国家自然科学基金资助项目（52077017）；国网重庆市电力公司重点科技项目（2021渝电科技10#）

收稿日期:2023/01/18

修回日期:2023/08/06

出版日期:2025/01/02

Parameter identification of photovoltaic inverter based on improved particle swarm optimization

algorithm

LUO Jian,1 SUN Yue2,JIANG Lijuan1

1. State Key Laboratory of Power Transmission Equipment & System Security and New Technology， Chongqing University， Chongqing  400044， China; 2. Training Center， State Grid Anhui Electric Power Co.， Ltd.， Hefei  2300,Anhui， China

Abstract: Accurate grid-connected photovoltaic（PV） inverter model is an important tool to study the fault characteristics of power system under large-scale PV access.  Objectives In order to solve the problem that the characteristics of the existing PV inverter simulation model are quite different from those of the actual PV inverter，  Methods this paper presented the method of parameter identification to construct the identification model of inverter. Taking a 1MW photovoltaic power station in Yunyang， Chongqing as the actual reference model， the working interval of the inverter was divided into three stages according to the actual working conditions， and the high and low sensitivity levels of the parameters to be identified in the three stages were divided by mathematical disturbance method. Then， the actual working data of photovoltaic power station was collected in stages， and the initial value range of each parameter to be identified was obtained after the data was analyzed and processed， and the synchronous identification parameter experiment was designed as a reference. Finally， an improved chaos genetic algorithm of particle swarm optimization（CGAPSO） was proposed as an identification algorithm to identify the relevant parameters step by step. The advantages of the proposed method can be obtained by comparing the synchronous identification results of the parameters， and the identification results were substituted into the simulation model. Results The error of synchronous identification results of low sensitivity parameters was far beyond the acceptable range， and the error of relevant parameters identified by CGAPSO step by step was less than 1.1%， which was much higher than the accuracy of synchronous identification results. Conclusions The output data of the identification model based on the improved particle swarm optimization algorithm had a high agreement with the actual inverter working data，which could accurately reflect the actual working characteristics of the inverter.

Key words: photovoltaic grid-connected inverter; inverter control strategy; identification of parameter; mathe-matical perturbation method; improved particle swarm optimization