Sensor data classification method based on phase space reconstruction

Abstract

The invention discloses a sensor data classification method based on phase space reconstruction, and the method comprises the steps: 1), mapping a one-dimensional sensor time sequence to an m-dimensional phase space through the phase space reconstruction based on coordinate delay so that sensor data exposes implicit key information to acquire a phase space track matrix; 2) inputting the extracteddata of each sensor at the moment i into a long-short-term memory network module, and setting the time step length input by the long-short-term memory network to be the embedded dimension m of phase space reconstruction; and 3) inputting the RGB matrix of the state information analogy image output by each long-term and short-term memory network into a two-dimensional convolutional neural network module, and inputting features extracted by the two-dimensional convolutional neural network into a full connection layer to obtain a classification result. The problems of limitation of an original one-dimensional sensor data structure and limitation of sensor data classification accuracy are solved, and the sensor data at the time point can be accurately classified.

Description

technical field [0001] The invention relates to the technical field of sensor data identification, in particular to a sensor data classification method based on phase space reconstruction. Background technique [0002] Early recognition based on sensor data mainly used traditional machine learning methods, including decision trees, support vector machines, hidden Markov and other models. These methods achieve state classification by manually extracting time-domain features, but the ability to manually extract features is very limited. , and need to provide a lot of prior experience to make up for the lack of data mining. In more complex problems, it is no longer realistic to rely on manually extracted features, so it is necessary to reduce the model's dependence on manual features. [0003] Sensor data recognition has its particularity compared to image data recognition. How to solve the classification of sensor data at a time is a technical problem, that is, it is necessar...

AI Technical Summary

Problems solved by technology

[0005] In view of this, the purpose of the present invention is to provide a sensor data classification method based on phase space reconstruction, to solve the traditional machine learning method in the recognition effect on one-dimensional sensor data rely heavily on manual extraction of features, sensor data structure limitations The limitation of the feature extraction model and the limitation of the sensor data structure make the calculation problem of the classification accuracy limited, and solve the technical problem of the difficulty of accurately classifying the sensor data at the time point

Images