Single-phase Series Motors In The Production Process Of Online Research About Quality Detection Methods

For single-phase series motors’ vibration testing online, its purpose is to detect the quality, proposed test implementation and signal processing methods. Consider the asymmetry of actual sample motors’ qualified and unqualified, the characteristics of the sample fail to predict, and the actual production line requires fast evaluation of the product. So the problem can be expressed as a class of learning for single-phase series motor to determine whether or not. By extracting samples of the actual site and real-time analysis of the data dimensionality reduction after the octave as a feature vector class learning algorithm can be implemented single-phase series motor line quality testing. The main work of this paper is:1. Summarized the single-phase series motor. Analysis on the principle,the overall structure,and the causes of motor vibration. To prepare for the later series motors for vibration testing production lines.2. Testing method in the production line of online motor vibration. The acquisition equipment is the first step of tested,signal processing and pattern classification. And it is also a basic link and an important link. Only the first step to ensure that no mistakes motor monitoring, the data processing and feature classification, can it makes sense.3. Signal Processing and Feature Extraction. Fault diagnosis of rotating machinery, the vibration is an important factor between safe and stable, the motor can be analyzed through motor vibrations. In line testing, the frequency of vibration signal analysis always using octave analysis. Motors for vibration data were collected octave analysis, extract an effective fault feature, and finally by PCA methods, making fault dimensionality reduction.4. Detection method for a class of motor learning. Structural Risk SVM functions perform well on statistical learning theory minimization principle. It is a good promotional vibration motor failure in small samples. And solve the problem of high dimension and the extreme conditions. By a large number of motor vibration data as training samples, feature vector extraction and dimensionality reduction, and SVM correct classification. Adjust the value of the radial basis kernel functions, the motor can be missed diagnosis accuracy rate with a reasonable choice.