论文部分内容阅读
For the first time, the diagnosis idea based on a correlation integral is proposed, which regards the correlation integral as a feature set. The correlation dimension is contained in the double-log curve of the correlation integral to scale, so extracting features directly from the correlation integral can avoid the bottleneck problem of determining the range of non-scale length. Several features extracted from the correlation integral are better than the single feature of the correlation dimension when describing the signal. It is obvious that this method utilizes more information of the signal than does the correlation dimension. The diagnosis examples verify that this method is more accurate and more effective.
For the first time, the diagnosis idea based on a correlation integral is is, which regards the correlation integral as a feature set. The correlation dimension is contained in the double-log curve of the correlation integral to scale, so extracting features directly from the correlation integral can avoid the bottleneck problem of determining the range of non-scale length. Several features extracted from the correlation integral are better than the single feature of the correlation dimension when describing the signal. It is obvious that this method utilizes more information of the signal than does the correlation dimension. The diagnosis examples verify that this method is more accurate and more effective.