Research And Implementation Of Intelligent Appliance System Based On Wearable Devices

In the past few years, with the improvement of people’s living quality, control and research of the intelligent appliance system attracts more and more attention, it become one of the most active research directions. Since the gesture as an essential part of human daily life, it seems to be a way of extensive exchanges among people, which gradually become the emerging interactive way in human computer interaction. Based on the existing theory and research, we hava achieved a kind of gesture recognition which based on the HMM (Hidden Markov Model)after analysis and summary. Specific gestures are designed and then these gestures are recognized by computer analysis, thereupon then achieve the purpose of controlling household appliances.In this paper, we utilize gesture recognition as the research object and analyzes the relative theory and method deeply. From the actual demand of users, this paper completed the design of intelligent appliance system which based on wearable devices.Three topics are proposed as following:(1) Feature Extraction.Feature extraction mainly include the time domain features and frequency domain features. Since the time domain features seem to be more intuitive, and easy to extract, and the cost effective of extracting time domain features are higher than extracting frequency domain features, so this thesis chooses the time domain feature extraction. Finally, the time domain features which we choose to extract including:Mean Value, Synthetic Acceleration, Variance, Amplitude, the Maximum Acceleration Axial, Peak Number, Peak Distance, Root Mean Square and Signal Magnitude Are.(2) gesture recognition process based on HMM.This paper first collect the experimental data of selected features, which divided into six groups, including upper, lower, left, right, front and rear, and then obtain Hidden Markov Model by train. With this Hidden Markov Model, I utilize sequence backward feature selection methods to conduct the consequent feature selection, then select feature which we need, and establish a perfect HMM model. The features of the final result we selected as following:Mean Value. Amplitude, the Maximum Acceleration Axial, Peak Number and Root Mean Square, finally, we utilize this Hidden Markov Model to test the data set which reserved by us, then get the results. Furture, we can use this HMM to recognize gestures.In this paper, random data are generated to perform experimental simulation that verify the effectiveness of the model. This model can provide some strong theoretical support for the control of the intelligent appliance system.it can provide certain theoretical support for the future development of intelligent household electrical appliance industry. The research of intelligent appliance system let the interaction of people and appliances more humanization. intelligent and comfortable, let appliance understand users, understand the needs of users, and understand the user’s body language.