The Distributed PV Power Generation Detection And Capacity Estimation In Residential Load

In recent years,as environmental pollution and other issues become more and more serious,the development of renewable energy has become imminent.Due to the advantages of saving investment,compatibility with environment,flexible installation location and considerable benefits,more and more residential customers begin to install distributed photovoltaic system(DPVS).Accurate installed capacity information of DPVS is very important for load forecasting and demand response capacity estimation.Although customers will register the installed capacity in the power company when installing the DPVS,there are such phenomena as the failure of the DPVS in operation,the installation of the DPVS by customers without the approval of the power company,and the expansion of the installed DPVS by customers.As a result,the original recorded installed capacity information of DPVS is inconsistent with the actual situation.Although the installed capacity of DPVS can be checked by using investigation of interview and questionnaire,this method takes a long time and invests a large amount of money.In practice,this method is not desirable.In order to solve the above problems,this paper studies the detection and capacity estimation of DPVS in residential load based on the net load curve of residential customers.The main contents are as follows:Firstly,in order not to depend on any other data information besides the customers net load curve(such as weather state),according to the output characteristics of DPVS under different weather types,based on the output data of DPVS installed by residential customers,a method of generating generalized weather types based on clustering and voting is proposed.Secondly,the detection model of DPVS based on characteristic parameters of typical net load curve is established by using support vector classifier.The simulation results show that the detection model of DPVS based on characteristic parameters of typical net load curve and support vector classifier has good detection ability.Finally,the customer net load curve with different DPVS capacity is studied and analyzed,and three characteristic parameters reflecting the installed capacity of DPVS are extracted;Meanwhile,in order to solve the problem of unbalanced distribution of sample data,Bootstrap method is introduced to generate a large number of virtual customers load data,so that the distribution of sample data is balanced.And the DPVS capacity estimation model based on Bootstrap-SVR is proposed.The simulation results show that the Bootstrap-SVR based DPVS capacity estimation model not only achieves good estimation results,but also adapts well to unbalanced data sets.Distributed photovoltaic is an important means to solve the current energy shortage and environmental pollution.With the vigorous support and rapid development of the state,the relative policies,regulations and market mechanism should be established to avoid the uneven installation market and lead to potential risks and hazards to customers and power grids.