Research On Real-time Detection System For Heart Rate And Falling Models

Now people pay more attention to their safety and health while staying in physical activity status, and especially focus on falling which happens on the elderly. And with the development of microelectronics and weak signals detection technique, the research based on integrated systems of heart rete and falling models has already become an important topic as a research direction of biomedical engineering. At the same time, wireless body sensor network technology is being developed as a research hotspot of health monitoring while movements with such the characteristics as energy saving, ad hoc, safety, integration and micromation. The main contents include:The thesis first introduces the current research status of integration system based on heart rate and falling. After analyzing the virtue and shortcoming of current falling recognition theories, the paper summarizes the heart rate detection while falling and places an emphasis on the most efficient method by detection of R-R periods in two-electrode ECG signals. Chapter 2 introduces the structure and contents of system platform considering the micromation, low power, wearable, high reliability and safety. And then the principle, design methods and corresponding signals extraction technology of each module of the system. Considering the artifical singals such as skin dithering and EMG, the wavelet transform algorithm is employed to extract the QRS waves from the polluted ECG singal, and voting algorithm based on the smallest circle is also brought forward to remove the noises. Several experiments were conducted to give evidence of the robustness and accuracy of the proposed algorithms. For falling phenomenons of the elderly, Chapter 4 discussed the recoginiton technique of plenty of falling models and employed Support Vector Machine (SVM) to binary-classify the models of different activities and recognize falling events. As the shortcoming of complexity and huge memory requirement brought by characteristic vetctors, we used Primary Component Analysis (PCA) to reduce dimensions of vectors space. In the experiment, seven falling patterns were implemented to evaluate the method and obtained satisfying results.For real-time health monitoring, wireless technology was employed to transfer biomedical data. So all the physiological phenomenons can be synchronized and tracked by wireless body sensor networks. Chapter 5 introduced one network topology suitable for body area, and enhanced the communication protocols including Secuirity and transport layers. Considering a wireless sensor network where the nodes have limited energy, we propose a novel model Energy*Delay based on ant algorithms (E&D ANTS) to minimize time delay in transferring a fixed number of data packets in an energy-constrained manner in one round. However, because of the tradeoff of energy and delay in wireless network systems, the Reinforcement Learning (RL) algorithm is introduced to train the model. The simulation results show that our method performs about seven times better than AntNet and than AntChain by about 150 percent in terms of energy cost and delay per round.Finally, the thesis discussed some applications about falling monitoring. Many experimental researches of heart rate and falling recongnition are finished, And the ninety percent early-warning ratio are successfully obtained. At the same time, we can track and monitor realtime heart-rate fluctuation curves while subjects move. The primary trial proved successful and effective.