Search

Although artificial neural networks are widely used in various fields, including mining and mineral processing, they can be problematic for appropriately choosing the model architecture and parameters. In this article, we describe a procedure for the optimization of the structure of a convolutional neural network model developed for the purposes of metallic ore pre-concentration. The developed model takes as an input two-band X-ray scans of ore grains, and for each scan two values corresponding to concentrations of zinc and lead are returned by the model. The whole process of sample preparation and data augmentation, optimization of the model hyperparameters and training of selected models is described. The ten best models were trained ten times each in order to select the best possible one. We were able to achieve a Pearson coefficient of R = 0.944 for the best model. The detailed results of this model are shown, and finally, its applicability and limitations in real-world scenarios are discussed.

The article presents a novel application of the most up-to-date computational approach, i.e., artificial intelligence, to the problem of the compression of closed-cell aluminium. The objective of the research was to investigate whether the phenomenon can be described by neural networks and to determine the details of the network architecture so that the assumed criteria of accuracy, ability to prognose and repeatability would be complied. The methodology consisted of the following stages: experimental compression of foam specimens, choice of machine learning parameters, implementation of an algorithm for building different structures of artificial neural networks (ANNs), a two-step verification of the quality of built models and finally the choice of the most appropriate ones. The studied ANNs were two-layer feedforward networks with varying neuron numbers in the hidden layer. The following measures of evaluation were assumed: mean square error (MSE), sum of absolute errors (SAE) and mean absolute relative error (MARE). Obtained results show that networks trained with the assumed learning parameters which had 4 to 11 neurons in the hidden layer were appropriate for modelling and prognosing the compression of closed-cell aluminium in the assumed domains; however, they fulfilled accuracy and repeatability conditions differently. The network with six neurons in the hidden layer provided the best accuracy of prognosis at MARE≤2.7% but little robustness. On the other hand, the structure with a complexity of 11 neurons gave a similar high-quality of prognosis at MARE≤3.0% but with a much better robustness indication (80%). The results also allowed the determination of the minimum threshold of the accuracy of prognosis: MARE≥1.66%. In conclusion, the research shows that the phenomenon of the compression of aluminium foam is able to be described by neural networks within the frames of made assumptions and allowed for the determination of detailed specifications of structure and learning parameters for building models with good-quality accuracy and robustness.

The knowledge on strength properties of porous metals in compression is essential in tailored application design, as well as in elaboration of general material models. In this article, the authors propose specification details of the ANN architecture for adequate modelling of the phenomenon of compressive behaviour of open-cell aluminum. In the presented research, an algorithm was used to build different structures of artificial neural networks (ANNs), which approximated stress-strain relations of an aluminum sponge subjected to compression. Next, the quality of the built approximations was appraised. The mean absolute relative error (MARE), coefficient of determination between outputs and targets R2, root mean square error (RMSE), and mean square error (MSE) were assumed as criterial measures for the assessment of the fitting quality. The studied neural networks (NNs) were two-layer feedforward networks with different numbers of neurons in the hidden layer. A set of experimental stress-strain data from quasistatic uniaxial compression tests of open-cell aluminum of various apparent densities was used as data for training of neural networks. Analysis was performed in two modes: in the first one, all samples were taken for training, and in the second case, one sample was left out during training in order to play the role of external data for testing the trained network later. The taken out samples were maximum and minimum density samples (for extrapolation) and one random from within the density interval. The results showed that good approximation on the engineering level MARE<5% was reached for teaching networks with ≥7 neurons in the hidden layer for the first studied case and with ≥8 neurons for the second. Calculations on external data proved that 8 neurons are enough to actually obtain MARE<10%. Moreover, it was shown that the quality of approximation can be significantly improved to MARE≈7% (tested on external data) if the initial region of the stress-strain relation is modelled by an additional network.

Modelling of comfort with the use of neural networks in modern times has become extremely popular. In recent years, scientists have been using these methods because of their satisfactory accuracy. The article proposes a method of modelling feedforward neural networks, thanks to which it is possible to obtain the most efficient network with one hidden layer in terms of a given quality criterion. The article also presents the methodology for modelling a PMV index, on the basis of which it can be demonstrated whether the network will work properly not only on paper but in reality as well. The objective of this work is to develop a performance model allowing the effective improvement of all electrical and mechanical devices affecting the energy efficiency and indoor environment in smart buildings. To achieve this, several attributes of indoor environment are included, namely: air leakage as a connection to the outdoor environment, but also as uncontrolled component of energy, ventilation as delivery and distribution of fresh air in the building space, individual ventilation on demand indoor air quality (IAQ) in the dwelling or as a personal IAQ control, source control of pollutants in the building, thermal comfort, temperature, air movement and humidity control (humidity modifiers, i.e., buffers different from the air conditioning radiation from cold and hot surfaces bringing forward a question about the strategy of the process control. One may either develop a series of control models to be synthesized later or one can use one over-arching characteristic and use its components for operating the control system. The paper addresses the second strategy and uses the concept of PMV for a criterion of broadly defined thermal comfort (including ventilation and air quality).

W artykule przedstawiono model generatora udaru kombinowanego w sieciach neuronowych dla potrzeb analizy pracy elementow i ukladow ochrony przeciwprzepieciowej. Jest to model typu feedforward o jednej warstwie ukrytej z czteroma neuronami. Struktura ta przy wykorzystania algorytmu Leveneberga-Marquardta umozliwia osiągniecie wysokiej efektywności, nawet w przypadku braku algorytmu optymalizacji struktury sieci. Opracowany model udaru napieciowego siecią neuronową daje mozliwośc zastąpienia modelu matematycznego generatora. Podejście to umozliwia efektywne wykonywanie obliczen symulacyjnych skutkow przepiec pochodzenia atmosferycznego dzialających na uklady ochrony przeciwprzepieciowych urządzen elektrycznych.

In this paper series, we will examine the relation between the smart grid and the next generation of buildings. The new objective is now to design buildings so that the energy exchange takes place in time suitable for both sides. The building takes energy from the grid and give it back at different time because in the next generation of buildings they are equipped in advanced control system that controls energy storage for several hours and the new or retrofitted building can eliminate energy peaks and valleys for itself and assist the smart grid in equalization of the load. The control systems include monitoring and modeling of energy use and indoor environment to arrive to a weather-based prediction, Therefore, in part 1 of the paper series we address collecting, processing and analysis of the measurement data that can be done by the new modular statistical software alone or in conjunction with a dedicated neural network. We want to minimize energy use and maximize the thermal comfort of the occupants. Using the modular structure of the database, data transformation technologies and other existing tools, the modular statistical software (MSS) has been created to process large amounts of data as input to the decision-making algorithm. This is to enable buildings automated control (BAC) to take over the control of heating/cooling, ventilation, illumination and other performance aspects that relate to sustainability of built environment.

This article describes the factors of presentation of the data visualization in infographics. We constructed a graph of interconnections among these factors. Using the methods of system analysis and graph theory, we developed a graphical model, which contains hierarchically ordered by priority factors presentation of the data visualization in the infographics. Experimental studies have found that the greatest influence on the quality of perception of data visualization in the infographics has the form of representation of numerical information. We also established that the scale of the structural elements of data visualization occupies a key place in the development of the presentation of infographics.

On the basis of theoretical research, we worked out a number of literary sources, which indicate the importance of infographics and the correctness of the information submission. To systematize the received information and solve the problem of recommendations for the creation of qualitative, understandable infographic design, it was proposed to use the theory of graphs and the method of pairwise comparisons. Using this technique, we carried out the study of the factors influencing the compositional design of infographics and established the importance of the technological process for the design of high-quality infographics. We developed the hierarchical model of the factors priorities of infographic composite design. In particular, we carried out the evaluation and optimization of the developed model, by using the pair comparison method and the relative importance scale at Saaty. The method of the hierarchy analysis provided an opportunity to evaluate elements of higher levels, based on the interconnection of different levels of the hierarchy, and not with the direct dependence of the elements on these levels.Three different design projects of infographics were explored. From the results obtained, the most effective in terms of perception and memorizing the information was a project that focused on textual design using graphic elements to enhance visual perception of information.These studies have found that the use of pairwise comparisons makes it possible to compare factors that cannot be qualified and quantified. The received calculations allow us to confirm the reliability of obtaining a high or low consistency of the results of expert surveys.

Currently, scientific research shows an increasing use of artificial intelligence technology to model various types of phenomena. One of such issues is the modeling of the output signals from special electric surge generator. The authors in earlier work presented the model of the output voltage signal from such a generator. This model used the technology of artificial neural networks. This article presents the model of the output current signal from the surge generator. This model also uses an artificial neural network. Both of them, with suboptimal structure, the above models have 1 hidden layer and 4 neurons. The simulation tests of these models showed their high accuracy. Having the model of the output voltage and current from the surge generator available, it is possible to simulate the influence of such signals on various elements of electrical circuits. This can be used, for example, for testing of projected surge protection systems.

The paper determines the optimal equation of dynamic error generated by the charge output accelerometer for the case of absolute criterion. The optimal formula is provided by fit function neural network algorithm. The paper presents the optimization algorithm and the process of learning of the network. There are also shown the errors generated by the network. The neural network optimization procedure results shows that obtained fit function regression compatibility is more over 99.8%.

This article is the continuation (the next stage) of works related to modeling of the time waveform of voltage signal generated by combination wave generator. In the previous stage of work, a preliminary voltage signal model was developed. This model, that uses an artificial neural network, is built from one hidden layer and four neurons. This article presents the optimal structure of the model. This new, optimal structure of this model, had one hidden layer consisting of fourty seven neurons. The authors of this article propose the replacement of typical mathematical and numerical models describing the 1.250 s voltage surge of the combination wave generator, an elaborated optimal model of artificial neural network in the simulation of the voltage surge. This optimal model authors envisage using in simulation studies on surge protection systems. Such systems are also used in the surge protection of electric motors.

The paper presents a methodology of calculating the maximal possible acceleration of the rail vehicle limited by the adhesion condition. The aim of devising this methodology was to propose to the traction vehicle's purchaser an accessible in computational complexity mathematical apparatus for verification of compatibility of the minimal vehicle's acceleration in a specified velocity section with the conditions of the order. The proposed method was devised in such a way so that the purchaser can theoretically verify the compatibility of the above mentioned acceleration on the basis of the data which constitute the so-called confidential know-how of the vehicle's seller. The equation for the maximal possible acceleration of the rail vehicle limited by the adhesion condition was presented along with its derivation on purpose for the better understanding of the phenomenon. All equations necessary for calculating the discussed acceleration in the method were presented in a table. These equations were showed in the order of calculating the values of the particular coefficients in the method. The devised method enables theoretical verification of the above mentioned condition of the minimal acceleration of the vehicle for diverse types of railroad surfaces (for a sprinkled with sand or a dry railroad, for average conditions, for wet rails). Thanks to this fact, the method may become a condition of general verification of compatibility of the minimal acceleration of the vehicle for a specified velocity section with the conditions of the order. The paper presents also an example of calculation of such acceleration for the FLIRT ED 160 vehicle produced by Stadler. This calculation was performed for wet rails' surface, in the velocity section from 0 to 120 km/h.

The paper presents an analysis of the upper bound of the dynamic error generated by voltage output accelerometer by using an artificial neural network. The integral-square error criterion was assumed as objective function for the above analysis. The relations for determining the above error were presented and also it was shown that after reaching the steady-state of impulse response of the accelerometer, the relationship between the integral-square error and time is linear. Relationship between the linear equation coefficients on two selected parameters of the mathematical model of accelerometer were determined for the range of this sensor work which is mentioned above. These relations were obtained by polynomial regression which optimal order was determined by means of artificial neural network.

The paper constitutes a continuation of research on load variability of rectifier units. The research are made for the selected tram substation. The performed analysis uses the actual measurements. This time the analysis focuses on relation between the maximum loads and 60 minutes overloads currents. The second part of the paper shows the effectiveness of use of the feedforward type artificial neural network. The effectiveness of the analyze was calculated for 250 times, for 50 cases. The results shown in the paper were obtained for optimal structure of the artificial neural network. The results presented in this publication prove to be the best results among the results known by the authors of the work.

This paper presents the construction the time-voltage signal output model of the combination wave generator using neural networks. The discussed model of surge was developed using feedforward neural networks with one hidden layer with four neurons. The results of the comparative tests show the high efficiency of Leveneberg-Marquardt algorithm using in the developed model, despite the lack of optimization of the structure. Presented at work model of the combination wave generator described in neural network, you can replace the traditional mathematical model of such a generator. Discussed above model can be used, inter alia, in the analysis of the operating of components and systems for surge protection of electrical equipment. This approach allows a particularly efficient execution of simulation calculations of the effects of atmospheric origin overvoltage protection systems operating on the surge electrical equipment.

A novel double-emitter ultrasonic system for distance measurements based on the correlation method is presented. The proposed distance measurement method may be particularly useful in difficult conditions, e.g. for media parameters undergoing fast changes or in cases when obstacles and mechanical interference produce false reflections. The system is a development of a previously studied single-head idea. The present article covers a comparison of the two systems in terms of efficiency and precision. Experimental research described in this paper indicated that adding the second head improved the measurement exactness – standard deviation decreased by 40%. The correlation method is also described in detail, also giving the criterion for the quality of the measurement signal.

This paper presents the continuation of studies on the evaluation of load variation of selected railway traction substations. Later in this article a reference will be made to the adequacy of the adopted model and its parameters in the light of approximation of the mean substation load with high degree polynomials. A need has been identified to continue the research in order to find a more appropriate method of approximation of the characteristics of the analyzed traction loads.

The paper presents a mathematical model of brushless DC machine induced by permanent magnets. Its construction uses the classical model of permanent magnet synchronous machine and induced model of power inverter using the serraphil form. The results of the computer simulation were presented for such states: startup, work under active constant load and the behavior of the machine in terms of exponential and stepping change of the power inverter's control angle.

While modelling a power supply system for an electric railway traction, knowing equivalent circuits of locomotives supplied this way is an essential issue. In alternating current traction it is important to diagnose, inter alia, processes taking place in transformers installed on electric vehicles. In the article there were presented specific phenomena occurring during the work of mono-phase, multi-winding, multisystem (systems AC: 50Hz, 16.7Hz) laboratory traction transformer. In the paper there were also presented difficulties encountered during the process of identifying multi-port equivalent scheme's elements of the described device, in which a construction defect occurs.