Data Acquisition And Simulation System Of Semg Based On ARM

In recent years, because of the influence of factors such as traffic accidents andenvironmental degradation, the number of the disabled in China have increased, andoccupying a greater proportion of disabled people is upper limb disability, it brings greatdifficulties to their daily life, so developing a cheap and easy to carry bionic arm play animportant role for our country on the solution of the problem.At present, a signal for controlling a bionic arm mainly extracted from EEG, EMG andvoice, and muscle signals are easy to be extracted, relatively stable, and can be widelyapplied to reflect the different motion state. Now most studies are based on a very high costof EMG signal acquisition instrument to collect sEMG signal, then the signal is input to thePC to simulate, although the collected signal accurately, but the features of expensive and noteasy to carry make it can’t be widely used. In order to solve the problem, this paper designs aSEMG collection system, and the signal collected by the system is input to the developmentboard with a ARM11processor into pattern recognition, and achieved the easy to carry, lowprice objective. This paper mainly includes the following work:1. This paper analyzed the generation of surface EMG signal and its characteristics,intrduced several kinds of commonly used mathematical model and suitable environment forthese mathematical models, study on the characteristics of common embedded operatingsystem, and described the selected Linux operating system and its architecture and kernelarchitecture and the characteristics ofARM processor and the structure in detail.2. We analyzed the Interference source on surface EMG signal acquisition system. TheInterference mainly from the system noise, the human body bioelectricity, motion stimulusartifact, the high frequency interference and the50Hz Power frequency interference. Amongthem, the50Hz power frequency is the biggest interference.3. In the collection of surface EMG signal, through the comparison of the needleelectrode and the patch electrode, we used silver/silver chloride patch electrode collected theEMG on the human body.4. In the amplification of surface EMG signal, because of the weakness of the surfaceEMG signals, when we design the hardware circuit, we consider amplify EMG signals, andthe amplifying circuit should restrain noise as soon as possible, therefore this paper adoptedthree-stage amplification. The first level is the use of differential amplification, effectivelysuppresses the interference caused by the common mode signal, AD620double inputamplifier as the second amplification, choose LF353as the third level. After three grade,EMG signal is magnified1000times or so, reached the requirements of the signal processing.5. In the aspect of surface EMG signal filtering, because the characteristics of the lowfrequency characteristics of EMG signals, we adopt high pass, low pass filter, the energymainly preserved in the useful signal from50Hz to350Hz. Because the50Hz powerfrequency is the biggest interference, so we adopted the signals collected circuit with MAX050notch filters, to eliminate the interference to the influence of signal acquisition.6. A/D conversion. After the collected signal change into digital signals through A/Dconversion, we can deal with the signal. In order to verify the correctness of the acquisitionsystem, in the PC terminal, we use the Matlab to simulate the map of the surface EMGsignals of different muscles, demonstrating the effectiveness of the signal acquisition system.7. In software design, mainly in the OK6410development board with a ARM11processor, we design the serial data receiving process, and simulate and depict the differentaction of SEMG according the received data.8. The motion pattern recognition experiments were carried out on the OK6410development board using double channel surface EMG signal data acquisition, data featureextraction, the value after feature extraction constitute vectors, those vectors are input into theneural network classifier to recognize different actions. Experiments show that the accuracyof signal collected by double channel acquisition system in this paper is higher than that ofsEMG signals collected from expensive acquisition instrument, and can be realized in thedevelopment board with ARM11processor, and achieved the goal of the portable and lowcost. At the same time, this paper proved that the design of surface EMG signal acquisitionand simulation system can meet the subsequent design of bionic arm control system needs.