Research On Automatic Extraction And Analysis Method Of Monitoring Data Trend

In the context of the era of big data,it is very necessary to extract valuable information from complex data through data analysis,and data trend analysis is an important part of it Through the analysis of data trend,not only can the difference between the current state of the data and the real state in the historical knowledge base be compared,but also the future change of the data can be predicted.Qualitative Trend Analysis(QTA)is a commonly used technology in data trend analysis.Aiming at the existing problems of the QTA,the following works have been carried out:1.Aiming at the problem that the poor adaptability of the window length and the error threshold are artificially set in the traditional qualitative trend analysis,which leads the inaccuracy of the extracted trend.The classification characteristic of the seven primitives in qualitative trend analysis is studied,and a method for automatically extracting local trend of data based on adaptive window is proposed.The relationship between the local and overall data is balanced by this method,so that the extracted trend is neither too detailed to trapped in some local details of the data nor too rough to lose the true information of the data.Take the temperature data of train axle as an object,compared to the extracted trend method of the traditional fixed window and variable window.Experimental simulation results prove that the method in this paper can extract and analyze the data trend accurately according to its own characteristics.2.Aiming at the problem that the inaccurate matching of trend primitives in practical applications in the traditional qualitative trend analysis.A combination of qualitative and quantitative method to expand the existing seven qualitative primitives is proposed and the existing matching matrix is fixed.The non-quantitative singularity representation in qualitative primitives is avoided and the accurate distinction of primitives in the same type is realized3.Aiming at the problem that traditional qualitative trend analysis focus on local trend analysis and the overall trend is not considered.A combination of the local trend of the data and the Mann-Kendall(M-K)trend verification algorithm is proposed,the accurate extraction and analysis of the overall trend of the data is achieved.The approximate change state of the data on the entire time axis is obtained.It helps to grasp the overall state of the data from a global perspective.Experimental simulation results show the effectiveness of this method4.Researchers are more inclined to observe the historical trend status of data in the form of graphical curves in practice.Therefore,a new idea is proposed in this paper that is to analyze the data trend by converting the data domain to the image domain.Based on scale changes of the trend extraction method,a scale factor scaled image is introduced to represent the relationship between local and overall firstly,the trend segment is extracted through the image processing process,secondly,the image domain is mapped to the data domain,and lastly,the experimental simulation results prove that the method in this paper can extract the data trend accurately and show the trend state intuitively.