Self-adaptive Method for sEMG Movement Classification Based on Continuous Optimal Electrode Assortment

Surface electromyography (sEMG) analysis is becoming increasingly popular in a broad variety of applications. Despite satisfactory classification rates are frequently obtained through the use of Computational Intelligence (CI) algorithms, there are some issues mostly related to the data acquisition which are not properly addressed in current studies. In this paper we aim to present a method capable of mitigate the noise in the sEMG acquisition caused mainly by loose or misplaced non-invasive electrodes. To address this issue we propose an auto-adaptive artificial neural network (AAANN) capable of identify this two anomalies in the signal and retrain itself discarding the causative channels. Once the AAANN is retrained it is possible to retrieve information of only the most significant channels which increase the accuracy rate of the CI method. The method was tested on a database containing five able-bodied subjects and four amputee subjects of both sex. The average classification accuracy for the adaptive input selection method was 83,96 ± 6,5% for the able-bodied subjects and 61,15 ± 7,7% for the amputees subjects against 72,06 ± 8,0% in able-bodied subjects and 39,77 ± 10,6% for the amputees subjects considering the non-adaptive approach. Both systems make use of the AAANN to classify 9 distinguish upper-limb movements with different degrees of freedom.

Optimization of Features to Classify Upper-Limb Movements Through sEMG Signal Processing

This paper aims to present the development of a computational intelligence method based on Regularized Logistic Regression able to classify 17 distinguish upper-limb movements through the sEMG signal processing. The choose of the tuning parameters of the regularization and the generation of the different classification methods are presented. For the different models were used variations involving 12 sEMG channels and the RMS, Variance and Medium Frequency features with which we proposed to achieve a most proper combination of parameter to perform the movements classification. The tests involved 50 subjects, including 10 amputees, using the NinaPro database and also a database currently on development by the authors. The global mean accuracy rate considering all the subjects and the channel and features variations was 70,2% prior the definition of the best case scenario. Once we defined the most proper channel and features combination, we were able to improve the accuracy rate to 87,1%, raising the rates of all movements performed for all databases.

Optimal Estimation of Mass Partition in a Centrifugal Separator by Metallurgical Balance

Mineral processing plants make use of several types of equipments for size separation (size classification) to segregate ore particles by size. A particular type of such equipments is the centrifugal separator, which is intended to receive an input stream of ore slurry to be partitioned into two output streams: a coarse and a fines one. The coarse stream contains most of the coarse solids particles of the slurry, whereas the fines stream contains most of the fine particles. Although a centrifugal separator intends to perform a physical segregation of the solids particles by their size, a chemical segregation also results, in such a way that the chemical content of the coarse and the fines streams are normally different from the content of the input stream. When evaluating the performance of the separation process, three fundamental aspects should be analyzed: 1) the size distribution of the solids particles in each stream; 2) the amount of solids mass from the input stream that goes to the coarse stream and to the fines stream, that is, the mass partition; and 3) the chemical content of each stream. This work presents the application of the Least Squares method of optimization to calculate the mass partition, based on the measured chemical content of the streams, and on the metallurgical balance equations of the separation process.

Neural Network to Failure Classification in Robotic Systems

A robotic system is a reconfigurable element, and inits programming, an algorithm can be implemented in order todetect and classify failures. This is an important step to ensurethat errors in actions do not cause damage or bring risks.Considering this, a Neural Network Multi Layer Perceptron(MLP) was used, in order to classify a set of failures in robotactuators, present in a database. This purpose is to analyze ifrobotic failures could be classified by MLP. The raw data aredivided in a temporal progression manner and torque in x, y andz axes. In total, five MLP neural networks were implemented foreach type of failure classification, using two different topologies.The number of neurons in the hidden layer is in accord with thecriteria of Kolmogorov and Weka, being the latter the besttopology for such application. In comparison to an algorithm(SKIL) using the same set of data, the MLP obtained the bestperformance in any topology of classification, with hit rates in80 to 90%.

Measurement of Overflow Density in Spiral Classifiers Using a Vibrating Fork Densitometer with Accuracy Evaluation

This work concerns the application of a vibrating fork densitometer to the measurement of overflow density in spiral classifiers. A spiral classifier is a mineral processing equipment which receives an ore slurry input and performs a gravity separation process between the solids particles of ore and the water. The classifier has two outputs: the “underflow”, formed by sedimented coarse solids; and the “overflow”, in the form of an ore slurry with fine suspended solids particles. For proper performance of a spiral classifier, the density of its overflow needs to be controlled by a feeding of dilution water at the input of the classifier. Even in present days, this control is still performed manually from manual samples of the overflow density, due to the lack of a standard instrumentation solution for this application. In this context, this work describes the application of a vibrating fork densitometer for overflow density measurement in spiral classifiers. The instrument performance was evaluated in two steps: a bench testing and a field testing. In both cases, its measurement accuracy was statistically investigated. The results obtained indicated the feasibility of the instrument for the intended application.

Cavitation Aggressiveness Estimation in Hydro Turbines Based on Cyclostationary Modeling

This work proposes a new methodology of detection and identification of erosive cavitation in hydro turbines, which is based on cyclostationary modeling of the cavitation induced vibrational signals. Different cavitation types cause damage to different turbine parts and induce different vibrational signatures, which can be employed to identify and locate the cavitation. Additionally, the cavitation aggressiveness can be estimated using the measured power of vibrational signal. High frequency accelerometers picked up the signals from two real turbines under normal operation. The methodology was implemented in software and a specific hardware was developed to run the software locally. Signals were synthesized in accord with the cyclostationary modeling and employed to validate the proposed methodology. Results obtained from real signals were similar to the ones obtained from synthetic signals, and corroborate the feasibility of this methodology in cavitation monitoring systems.