| Lower limb disabilities is a huge group. To monitor the rehabilitation training and evaluation is of great significance. But it's also the difficult issue still unresolved. This study tried to use multi-sensor data fusion method which has developed rapidly in recent years to extract the information of the upper limb(HRV).In order to effectively extract the information of the upper limb, this paper finishes the following work:1. Finite element analysis of walking aids. Doing the load test on the structure of the walking aids therotically, considering the actual use of the walking aids. The different forces ratia and different load points in a certain range of movement under the walking aids may get different deformation and the relative moment of the distribution of output. The results is used to make sure that the position of the multi-sensor is correct.2. Using the indirect method of measuring the upper limbs'information and therotically demonstrating its feasibility. Indirect measurement method effectively eliminate the user's psychological fear, and it can ensure the accuracy of measurements of authenticity.3. Establishing a measuring devices based on indirect measurement of HRV. Through the installation of the 12 sets of strain gauges on the walking frame, the upper limbs'information can be effectively exacted by using the redundancy- optimization of the principle in static linear calibration.4. Multi-sensor data fusion by the artificial neural network(ANN) method of extraction, processing and calibration of the upper limb of information, and with the traditional linear methods were compared. The results show that multi-sensor data fusion based on ANN is superior comparing to the traditional linear method. The max single-direction accuracy error is 7.78% smaller than traditional method(8.45%); the maximum cross-interference is 7.49% smaller than the traditional method(19.96%). ANN demarcation effectively improved the accuracy of single-direction and cross-interference. |
PDF Full Text Request