Dimensionado de la instalación fotovoltaica

The apparatus and data acquisition tools used in this study diered concerning the objective being carried out. The main experiments included EMG acquisition using surface electrodes, nor- mal gait analysis using cameras and reective markers, orientation and anatomical angles analysis using IMU sensors and signal analysis. However, a generalised description of the equipment is presented in this section. Detailed descriptions are given in the next sections. The experiments were carried out at the Human Motion Analysis Unit, Central Analytical Facilities, Stellenbosch University, South Africa. Validation tests were carried out at Neuromechanics Unit, Stellenbosch. The raw signals sEMG were acquired using the Noraxon active electrode system, which has an onsite amplier. The system has a 10 GΩ impedance with a common mode rejection ratio of 115 dB at 50 Hz. The signals were sampled at 1500 Hz as required by the SENIAM and ISEK standards. The measurement function accuracy (MFA) of the sensors was ±2 µVRMS. The base-

line noise was less than 1 µVRMS. The electronic gain was set at 200 with an overall gain of 500.

The 50 Hz notch lter was not applied during data acquisition. Wireless communication was used between the electrodes and the data acquisition unit to minimise cable movement artefact. The experimental protocol involved the acquisition of maximum voluntary contraction (MVC) level from the participants by holding a movement contraction for at least 5 s, since the recom- mended processing time for the targeted processor is 300 ms. As a result, the 5 s duration pro- vided enough data without the participants experiencing fatigue. The participants were tasked to perform ve motions of dorsiexion and plantarexion movements with a resting pause in- between (dorsiexion-rest-plantarexion). The second set of ve motions included performing dorsiexion and plantarexion continuously without a resting pause (dorsiexion-plantarexion). All these initial experiments were conducted while the participants were seated. Another set of data was carried out when the participants were walking along a 10 m walking envelope. The reason for removing the resting pause in the second set of data was to facilitate an evaluation to determine whether there is muscle crosstalk and also testing the ability of the developed system to classify the movements in the event of fast reactions. The signals were recorded using the Noraxon MyoMotion System, also known as MyoResearch3 (MR3), USA. Ten data sets were collected for each motion on normal gait activity. The activities were carried out during design parameter considerations and in evaluating the performance of the developed design. The raw data sets were stored for further processing. The post-processing of the signals was done using MR3 Noraxon MyoMotion System. In the MR3 platform, ltering using digital lters was im- plemented. The results were exported to Matlab 2017a for further analysis.

The signal analysis was carried out based on the method proposed by [72] and [180] and supported by [98]. However, much guidance for signal processing was based on ISEK standards since it is more focused on sEMG signals. The raw signal was ltered using a nite impulse response band pass Butterworth lter between 15 to 50 Hz with the aid of a Hamming window which allows for 50% overlap. A 60 db/oct was utilised as a method of minimising overshoot reduction of settling time. Further amplication was achieved with a 60 dB gain. The removal of the 50 Hz interference from the power line was achieved using a bidirectional innite impulse response lter during post-processing in MR3 software. Rectication was then applied to allow calculations related to time domain, frequency domain and time-frequency domain features of the signals. Smoothing of the signal was done using a three-stage algorithm which involves the mean (moving average), mean absolute (the mean average with combined rectication) and root mean square techniques.

The processed results from the MR3 Noraxon Software were then exported to Matlab for further analysis. The power spectral density analysis was based on the Welch method, hence Fast Fourier Transform was applied for the power spectral analysis. The recording of the posi- tional data enabled the labelling of the events during post-processing of the data for analysis. Furthermore, individual modules could easily be integrated with the whole system or work freely. The Noraxon IMU MyoMotion system consists of 16 individual wireless sensors. The system was able to acquire orientation and anatomical data in real-time. During post-processing, the user can export data as quaternions or even raw accelerometer, gyroscope and magnetometer values. In dynamic mode, the system has an accuracy of ±1.20_{. The IMU sensors had an internal}

sampling frequency of 400 Hz for both gyroscopes and accelerometers. However, when integrated to the data logger, the data sampling frequency is reduced to 200 Hz. The gyroscopes had a full scale of ±2000 deg/s and the accelerometers had a full scale of ±1.7 g. The Nexus Motion System from Vicon, USA was used for the anatomical and orientation angle analysis. The system consists of motion tracking cameras and an algorithm used to develop a 3D skeletal model of the participant. Reective markers were installed on selected anatomical landmarks such as the foot, tibia, thigh and pelvis of the participants. These reective markers were then detected by the high-resolution cameras during gait analysis, provided that the participant was within the calibrated working envelop. In this study, the working envelop was a 10 m by 15 m platform with force plates at the centre. It was a complimentary technique for gait analysis. The modules within the MR3 software system include motion, EMG, force and video [191], as illustrated in Figure 3.1.

Working in Nexus platform was made easy due to the availability of the Biomechanics Work- ow feature, which enabled the development of a custom-made data collection protocol that was developed to best suit the study of amputees. This feature has been used successfully in other previous studies [192]. Other features which were explored were Optimum Common Shape Technique (OCST), Symmetrical Center of Rotation Estimation (SCoRE) and Symmetrical Axis of Rotation Analysis (SARA), as they enabled easy data acquisition and analysis within Nexus Software.

Figure 3.1: MyoResearch3 recording Noraxon System modules

The Noraxon MyoMotion System and the Vicon Nexus Motion System were both calibrated before the recordings. This enabled the validation of the recorded results using normative data already pre-loaded within the system. The name and activity number (Label, e.g. Partic- ipant1.Activity1.) was used for every recording. The participants were instructed to stand straight, with arms exed along the hips. In the software, the chosen body segments were com- pared to the known starting position of all body joints on the skeletal avatar hence compared to the known starting position of all body joint. The calibration position was maintained for a period of 1 s or more after triggering the calibration module within the software. After initial calibration, recalibration was carried out after every ve sets of activities. That was done to continuously develop a reliable measuring system. An audible sound was heard, notifying the user of the beginning of the calibration process and another sound was heard during the end of the calibration process.