Research On Power Load Pattern Recognition Algorithm Based On Characteristic Index Dimension Reduction And Improved Entropy Weight Method

With the development of human intelligence,the continuous innovation of science and technology,and the renewal of new and old technologies,the field of communication technology has also been improved to a certain extent,which enables people to obtain various types of information through different channels in daily life,and these information is also transformed into various digital forms through different means to present and transmit.Similarly,with the continuous rise of technology level,the development of power services and the rise of power market,all kinds of data of relevant power enterprises are becoming more and more digital and informatized.Smart grid increases the amount of data collected by the power industry from TB to PB.Terminals collect data more frequently,which makes the form of data more extensive.In addition,the industry is developing towards the trend of intelligent and lean development,heterogeneous data integration,and the growth of data volume bring the urgent demand for its fast and efficient processing.The challenges of high scalability and efficient,accurate load pattern analysis and processing have also become the forefront of data mining research.Therefore,give full play to the role of artificial intelligence and data mining algorithm in machine learning,can process massive power data in time,and get valuable information from it.Accurate and efficient load pattern recognition of power load big data is an indispensable basic work to support the safe,reliable and economic operation of power grid.At present,it meets so many difficulties to handle high dimensional eigenvalues of tremendous amount of original collected data.To fulfill the demand of accurate classification of load pattern recognition,this paper reaches out a solution which based on dimensionality reduction of characteristic index and improved entropy weight method for power load pattern recognition.First of all,the purpose andsignificance of research on power load pattern recognition are introduced,and the researches of domestic and foreign scholars in related aspects are summarized.The flow of power load pattern recognition model is given.In the current era of big data,power load data in the context of smart grids can be obtained through different channels,which are multi-dimensional and heterogeneous.It emphasizes the importance of data processing for massive load data and introduces several data processing methods.At the same time,after data processing,dimension reduction processing is required to reduce storage space and improve algorithm efficiency.The classification of dimensionality reduction technology is described,and its principle and characteristics are further explained and explained.Secondly,the pattern recognition of power load requires classification labels of load data,and this process requires clustering algorithms to implement.The idea and classification of the clustering algorithm are introduced,and several common clustering algorithms are analyzed and explained.The K-means clustering algorithm is selected for subsequent calculation examples at the end.In order to determine the optimal number of classifications,the quality of the clusters needs to be evaluated.Several common clustering evaluation indexes are given,and the evaluation functions of the pedigree clustering and K-means clustering are further explained according to the selected algorithm.Subsequently,the classification algorithm is used to complete the recognition of the power load pattern.The principle and classification of the classification algorithm are introduced,several common classification algorithms are given,and their thoughts and characteristics are summarized.The K-nearest neighbor algorithm is selected for the analysis of examples at the end.At last,combined with the research in the previous sections,a case study of the proposed algorithm is performed.The simulated load data is used to verify the K-means load pattern extraction algorithm.The results show that the K-means clustering algorithm can effectively extract the electric load pattern and obtain the classification label.On this basis,it is verified that the K-nearest neighbor algorithm can identify the load pattern,thus proving the effectiveness of the proposed algorithm.Considering the impact of high-dimensional data on the algorithm,further study the importance of load dimensionality reduction.The two methods of feature selection and feature extraction are used to reduce the dimension of the measured load data,and the reduced dimension data is used as the input for the K-means clusteringalgorithm.Compared with the traditional clustering algorithm,the results show that the clustering quality of the feature selection method of dimension reduction is the best and the algorithm efficiency is the highest.Based on the analysis of the first three calculation examples,the proposed algorithm is verified.Firstly,six characteristic indexes such as daily load rate,maximum utilization hour rate,daily peak-to-valley difference rate,peak period load rate,flat load rate and valley load rate are extracted and taken as input to replace the original load data.Secondly,the improved entropy weight method is introduced to configure the weight coefficient of each characteristic index adaptively.Then,the elbow method is used to determine the optimal number of clusters,and the clustering method of weighted Euclidean distance K-means is used to get classification labels for sample data.Finally,the K-nearest neighbor algorithm is used to identify the labels and the six characteristic indexes.The method based on confusion matrix is used to evaluate the proposed model.The results of the calculation example show that the improved algorithm combining dimensionality reduction of characteristic index and improved entropy weight method is better than the traditional K-nearest neighbor algorithm model,and the efficiency and accuracy are improved.