Based On The Research Of Abnormal Heart Rhythms Of ECG Signal Recognition Algorithm

In recent years, the prevalence of cardiovascular disease is increasing, cardiovascular disease patients often appear the phenomenon of arrhythmia at the beginning of the illness, It is great significance to prevent the prevalence of cardiovascular diseases that if the types of patients' arrhythmia can be detected and be found out in real time, then make the warning and carry out targeted therapy.ECG recorded the change of the weak current in the heart beat which has an important means to detect the type of arrhythmia. At present, for the detection of arrhythmia is mainly through the professional doctor of medical knowledge and experience of ECG signal to recognize, however, The large data of ECG signal is more complex and every day the doctor need to deal with a lot of heart waves to identify, it is easy to cause a miscarriage of justice. Therefore, it has become a hot spot to recognize and deal with the ECG signal in recent years.The phenomenon of abnormal ECG signals in the electrocardiogram showed: it has abnormal time and shape of the heart beat- abnormal heart beat. Therefore,it is an important step in the classification of the ECG signal that how to identify abnormal heart beat. Although lots of researches on Intelligent ECG signal processing, but due to the different ECG signal is complicated and have noise interference, classification of ECG arrhythmia types did not reach desired real-time recognition and accurate recognition, also failed to meet the clinical requirements. This paper from the ECG signal every stroke of the heart beat type recognition problem in-depth research, and designs a kind of improved genetic SVM classifier to recognize the heart beat automatic identification system, to realize the automatic recognition of six categories of conventional ECG heart beat and further as a physician's assistant means to accurately determine the type of arrhythmia.Especially for the real-time identification of the dynamic ECG signal, it has important practical significance to find out the reason of the arrhythmia and realize the real-time alarm.Firstly, according to the generation mechanism and characteristics of the ECG signal, the noise type was analyzed.On the basis of the wavelet theory a kind of adaptive filter of wavelet threshold denoising was designed, the algorithm is better to filter noise signal interference;Secondly, in the detection of ECG signal waveform feature points, a recognition algorithm was designed that based on wavelet multi-scale, according to the relationship of the signal singularity and the wavelet transform modulus maxima, this algorithm realize the detection of the characteristic points; Third, in order to better extract ECG signal feature information, time domain ? frequency domain and time-frequency domain angle, more comprehensive extracted the feature vector of the signal was extracted from the mathematical statistics and waveform mentality. But if the feature vector too much, there will be the redundancy of feature information. Therefore, this paper designs the maximum likelihood estimation to reduce the dimensionality-the PCA algorithm, it is better to improve the operation speed of the data and to optimize feature information; The fourth aspect, two kinds of classifier of abnormal ECG rhythm beat classification was put forward: BP neural network and improved genetic SVM. And two kinds of classifier performance are evaluated By MIT/BIH database data, the results show that the two algorithms can accurately identify the heart beat of ECG signals,but the improved genetic SVM classification algorithm has higher accuracy, and a set of test data can also reaches real-time recognition. Finally, a automatic recognition system of ECG abnormal heart beat was designed according to the abnormal ECG rhythm recognition algorithm process.