Lacrimioara Grama

Supercapacitor modelling using experimental measurements

Supercapacitors are components for energy storage, dedicated for applications where both energy and power density are needed. Even if their energy density is ten times lower than the energy density of batteries, supercapacitors offer new alternatives for applications where energy storage is needed. We have done an experimental bench and we have measured the voltage across a supercapacitor in a constant current charge/discharge cycle. Using these measurements and a mathematical algorithm we have proposed an electrical model for the supercapacitor, and we have determined the parameters of the model also.

Spectrograms, sparsograms and spectral signatures for wildlife intruder detection

In this paper some properties of spectrograms and sparsograms are reviewed. The framework addressed is acoustic based wildlife intruder detection. The spectral signatures are also recalled within this framework. The averaged binary sparsogram is introduced and it is shown that it can be considered an effective tool for classifying the possible intruders sounds into different classes.

About quantization of audio signals for wildlife intruder detection systems

In this work we study the quantization of audio signals for a zero-crossing method recently used to detect intruders in wildlife areas. This method implements two descriptors: D (represents the number of samples between two real zeros) and S (represents the number of points of local minima/maxima between two consecutive real zeros). We show using experimental results that in the proposed audio based wildlife intruder detection framework, the number of D/S pairs are almost constant till the number of bits used for quantization is less than six.

SPICE Simulation of Analog Filters: A Method for Designing Digital Filters

Traditionally IIR digital filters are designed by using analog filters described in time or transform domain, then by converting the analog filters to digital filters using appropriate transformation from s-domain to z-domain. For many engineers analog filters mean certain circuits or a netlist of components, and digital filters are a set of statements in certain software. In this work, we show how to obtain a digital filter from a given netlist of an analog filter by skipping the transfer function description.

Alarming events detection based on audio signals recognition

Many zones, especially wild areas need surveillance systems in order to protect them against potential destroying actions. An attractive solution for automatic surveillance systems consists in audio based systems, which present some advantages compared with video or mixt surveillance systems. This paper proposes an alarming events detection system with two detection levels. On the first level the system detects only if the event is a dangerous one (alarm on) or a normal one (alarm off). On the second level, if it is the case, the system identifies exactly the nature of events in four classes: chainsaw, gunshot, human voice or tractors. The system uses a set of features of the audio signals associated with the events, as input data for two artificial neural networks that act as pattern recognition components. The experimental results prove that our system is a very reliable one, presenting the maximum possible correct recognition rate (100%) on the first level and very high correct recognition rates on the second level: 99.50% across all data set (training, validation and testing) and 95.0% in the independent testing data subset.

Experimental determination of Equivalent Series Resistance of a supercapacitor

The synergy between batteries and capacitors has been growing, to the point where supercapacitos may soon be almost as indispensable to portable electricity as batteries are now. One very important parameter of a supercapacitor is the Equivalent Series Resistance (ESR). The energy that can be stored in a supercapacitor depends on ESR, so we intend to measure that parameter. Our tests were made on a 350F supercapacitor, 14V, produced by Econd Ltd. We have two methods: the first consist in evaluating ESR by constant current discharge, and the second method consists in a dynamic discharge cycle at a constant current. After measurements and some mathematical evaluations a real value of ESR is proposed.

Acoustic classification using linear predictive coding for wildlife detection systems

In this work we compare different classification algorithms applied on different number of features (linear predictive coding coefficients) in order to detect audio signals from wildlife areas. The final goal is to find the appropriate number of linear predictive coding coefficients to provide the desired accuracy for a certain framework. The experimental results prove that the best classifier is Logistic Model Trees regardless the number of features, having a constant classification accuracy greater than 95%. In the case of a reduced number of features, both Random Forest and Lazy IBk have good results; the classification accuracy is greater than 98%.

On the optimization of SVM kernel parameters for improving audio classification accuracy

This paper presents a grid search approach to optimize the kernels parameters for the support vector machines classifier. The most encountered three kernels are considered: linear, radial basis, and sigmoid. We show that the optimization of parameters improves the recognition performance for audio signals classification, especially in the case of sigmoid kernel. The behavior of the model is very sensitive to kernels parameters, which, in turn are sensitive to data and selected features. We consider the problem of multiclass classification with imbalanced datasets. We compare the accuracies obtain with and without kernel parameters optimization. As features we use Mel frequency cepstral coefficients.

A SPICE method for designing digital filters from analog filters

Usually IIR digital filters are designed by using analog filters described in time or transform domain, then by converting the analog filters to digital filters using appropriate transformation from s-domain to z-domain. For many engineers analog filters mean certain circuits or a netlist of components, and digital filters are a set of statements in certain software. In this work, we show how to obtain a digital filter from a given netlist of an analog filter by skipping the transfer function description.

Phase approximation by divide-and-conquer piecewise linear fitting of gain

Having a piecewise linear fitting of gain, one can obtain a good approximation of phase. There are few ways to determine the slopes of a broken line approximation, but the most popular approaches need the breakpoints. In this paper we determine the breakpoints using a divide-and-conquer approach. Results of piecewise linear approximation of gain and of the resulting phase approximation are provided for different cases.