Research On Time-frequency Analysis Methods For Building Electrical Loads

In order to break through the energy dilemma,China has continuously optimized its energy structure by vigorously developing renewable energy and adjusting the form of energy supply.However,the random and intermittent nature of renewable energy sources and the large number of grid connections pose a challenge to the power grid,and demand-side management is a good solution to achieve local consumption and improve energy efficiency by adapting renewable energy sources to demand-side loads.With the renewal of electrical equipment in recent years,this has made demand-side loads,mainly building electrical loads,more complex and variable.Therefore,in order to achieve better renewable energy consumption and demand-side load management,the changing characteristics of building electrical loads need to be clearly understood.At the same time,the rapid development of smart grids also provides data support for accurate analysis of load characteristics.Traditionally,indicators such as peak load,peak-to-valley ratio and load factor,or load curves,have been used to analyze the characteristics of building loads.However,this method is based on time domain analysis,which makes it difficult to fully characterize the cyclical and fluctuating characteristics of the load.This paper explored the application of frequency domain analysis methods in extracting load characteristics,expanding the single time domain analysis into a combined time and frequency domain analysis,so as to grasp the changing characteristics of the load as a whole.This paper researched and collected 15-minute level electrical load data for three commercial buildings in Beijing for one year,and based on this,explored the application of Fourier transform and wavelet transform to electrical loads,analyzed the fluctuating characteristics of the researched building loads,and compared them with the loads calculated using energy consumption simulation software,with the following main conclusions:(1)The physical significance of the frequency component characterization of building loads is analyzed and summarized,and a discrete Fourier transform-based frequency domain analysis method for electrical loads is proposed,which can clearly quantify the periodic characteristics of loads but loses time domain information.Three types of buildings are analyzed in conjunction with the time domain: all three types of building electrical loads are dominated by 24 h periodic fluctuations,but significant differences still exist.The office building load in a single-peak pattern,with a "polarization" between weekdays and nonweekdays where the frequency and amplitude are inversely proportional;the hotel load in a double-peak pattern,with a time difference of about 7 h between the two peaks with an equal amplitude between the 12 h and 8 h periods in the frequency domain;the commercial complex load in a double-peak pattern,with a trend of decreasing,then increasing and then decreasing amplitude in the frequency domain,with the higher amplitude of the 8 h period indicating the double-peak time difference of the load.(2)An analysis method based on the discrete wavelet transform is proposed,which can provide both time domain and frequency domain information of load fluctuations,reducing the demand for data volume and improving the analysis efficiency.During the air-conditioning season,the three types of buildings show different characteristics,with the load fluctuations in office buildings being stronger on non-working days than on working days,while the opposite is true for hotel buildings,and the load fluctuations in commercial complexes being stronger on working days than on non-working days at high frequencies,while at low frequencies the intensity of the two is essentially the same.In the non-air-conditioning season,the intensity of fluctuations in the high-frequency band of office buildings is basically the same;while in the low-frequency band,it is different,as the load trend of non-working days’ lags behind working days by 1 h.The load fluctuations of working days in hotel buildings are stronger than those of non-working days,and not only does the high load last for a longer period of time,but the peak load is 6 h earlier,and the fluctuations of working days and non-working days in commercial complexes are not very different.(3)The building loads calculated using the energy consumption simulation software are more dispersed than the actual building spectrum energy from the research,with peaks appearing at several period components,while the intensity of fluctuations in each frequency band was much greater than the fluctuations in the same frequency band in the research buildings,and the limitations of the software resulted in the loss of high frequency information,and the dynamic settings of the model required further changes.Therefore,the dual path analysis method based on the time-frequency domain can correctly identify and extract the electric load fluctuation characteristics of the building.It is of great significance for improving load forecasting accuracy,building load modelling,and demand-side management.