Search Results

Összefoglalás. A mély mesterséges neuronhálók elterjedése az ipari alkalmazásokban évekkel azok megbízhatóságával, értelmezhetőségével, és biztonságával kapcsolatos szakterületek fejlődését megelőzően történt. Az egyik, gyakorlatban is jelentős területen, a képfelismerésben például a megvalósult megoldások szinte már emberi teljesítményre képesek, de ezzel együtt célzott zajjal ezek a rendszerek félrevezethetők, megzavarhatók. Jelen kéziratban ismertetünk néhány tipikus biztonsági problémát, valamint rámutatunk arra, hogy a hagyományos szoftverfejlesztés területén alkalmazott minőségbiztosítási módszerekkel rokon megoldásokra szükség van az MI-re épülő rendszerek fejlesztésében, akár a mesterséges neuronhálók biztonságát, akár az MI rendszerek hagyományos komponenseinek fejlesztését tartjuk szem előtt.

Summary. Research on the trustworthiness, interpretability and security of deep neural networks lags behind the widespread application of the technology in industrial applications. For example, in image recognition, modern solutions are capable of nearly human performance. However, with targeted adversarial noise, these systems can be arbitrarily manipulated. Here, we discuss some of the security problems and point out that quality assurance methods used in traditional software development should also be adapted when developing AI-based systems, whether in the security of artificial neural networks or traditional components of AI systems. One of the main concerns about neural networks today that – to the best of our knowledge – affects all deep neural networks is the existence of adversarial examples. These examples are relatively easy to find and according to a recent experiment, a well-chosen input can attack more networks at the same time. In this paper we also present a wider perspective of security of neural architectures borrowed from the traditional software engineering discipline. While in traditional development several methods are widely applied for software testing and fault localization, there is a lack of similar well-established methods in the neural network context. In case of deep neural networks, systematic testing tools and methods are in the early stage, and a methodology to test and verify the proper behavior of the neural networks is highly desirable. Robustness testing of machine learning algorithms is a further issue. This requires the generation of large random input data using fuzz testing methods. The adaptation of automatic fault localization techniques has already started by defining notions like code coverage to neural networks. Lastly, we argue that the effective development of high quality AI-based systems need well suited frameworks that can facilitate the daily work of scientists and software developers – like the Deep-Water framework, presented in the closing part of the paper.

Összefoglalás.

Az energiaigény és a megújuló energiaforrásokból származó kínálat között fennálló szezonális eltérés áthidalható a hidrogéngáz bevezetésével az energiaellátásba. A nagy léptékű energiatárolás hidrogén formájában a felszín alatti térben lehetséges. Azonban a kőzet pórusterében az injektált hidrogén hatására végbemenő reakciók nemcsak a kitermelendő hidrogén mennyiségét és minőségét csökkentik, de a kőzet hosszabb távú állékonyságát is ronthatják. A Kárpát-Pannon régióban jelentős mennyiségben találhatók porózus kőzetek, amelyek hidrogéntárolásra is alkalmasak lehetnek. Ugyanakkor, ezek a kőzetek változatos ásványos összetételük miatt reakcióba léphetnek a hidrogénnel. Vizsgálatunk célja, hogy megismerjük a kőzetalkotó ásványok viselkedését pórusvíz és hidrogén jelenlétében, amely elengedhetetlen a rezervoár tárolási potenciáljának felméréséhez.

Summary.

One of the key substances in the modern-day energy transition is hydrogen, which can be utilized as an energy storage chemical substance. To store hydrogen on the scales required for global hydrogen economy, porous geological formations should be considered. However, geochemical challenges associated with hydrogen storage in sedimentary formations are still not well understood. Mineral dissolution and precipitation, as a result of hydrogen injection into the rocks not only can decrease the quality and the quantity of the stored hydrogen but may have an impact on the rock integrity as well. The Carpathian Pannonian region is rich in porous rocks, which could serve as hydrogen storage sites. However, many of them show various mineralogical compositions, which could behave differently under high hydrogen partial pressure. The main objective of our study is to predict geochemical reactions among rock-forming minerals, pore water and hydrogen. For this purpose, we apply analytical techniques and geochemical modeling.

The subject of this research is the Late Miocene Szolnok Sandstone Formation located in the Pannonian Basin, Carpathian-Pannonian Region. In the future this Formation can play a significant role in hydrogen storage, due to its favorable reservoir geological and petrophysical characteristics.

X-ray diffraction analyses were carried out, polished and thin sections were prepared for petrographic and geochemical analyses. The collected data were used in the PHREEQC modeling environment. In the first stage, equilibrium batch models were made to assess the potential long-term impacts of hydrogen on the reservoir rock and the effect of the geological environment. The modeling results of the project showed that hydrogen almost does not react with silicates (e.g., quartz). Possible hydrogen loss can occur due to redox reactions. Pyrite (FeS₂) can react with hydrogen producing hydrogen sulfide (H₂S) and since petrography has revealed that the studied sandstones have pyrite as accessory mineral.

Abstract

The anaerobic digestion is a well-known method in waste management of biodegradable wastes to transform waste to energy. Proper digestion requires optimal fermentation conditions to improve the quality and yield of biogas.

The objective of this study was to characterize the biodegradation process of synthetic and organic wastewater. Microalgae (Chlorella vulgaris) were utilized as a bioindicator for anaerobic digestion and monitoring of the fermentation process. Besides bioindication, the viability of the microalgae and the chlorophyll concentration were also assessed in such fermentation processes, since microalgae can be a potential source for biofuel production and a plant nutrient.

The biodegradation process was studied for a month in an anaerobic tank reactor. The fermentation processes and lengths of the fermentation stages were successfully monitored and separately identified based on the pH and gas development. Furthermore, the amount and dynamics of the biogas yield also revealed that the fermentation process was about 510 hours in both cases. In contrast, increased temperature in thermophilic range (45°C) accelerates the degradation processes and resulted in shorter hydrolysis (60 hours), acetogenesis (24 hours) and longer methanogenesis (81 hours) stages, where higher biogas yield was also achieved (59.3%). During the process, the concentration of nutrients showed logarithmical tendencies and COD showed power tendency in time. The extent and the direction of the changes were in correspondence with microalgae activity. In thermophilic circumstances, living microalgae biomass dropped significantly without recovery therefore such an environment is not a viable option for microalgae growth. Moreover, dead microalgae biomass seems to act as a substrate for fermentation slightly increasing the concentration of some nutrients in the wastewater.

Összefoglalás.

Jelen tanulmányunkban két Duna menti parti szűrésű vízbázis vizsgálatát tűztük ki célul. A Tiszta Ivóvíz: a biztonságos ellátás multidiszciplináris értékelése a forrástól a fogyasztóig című Nemzeti Kiválósági Program keretében két olyan kiemelt fontosságú parti szűrésű vízbázist vizsgáltunk, amelyek meghatározó szerepet játszanak a Fővárosi Vízművek ellátási területén, Budapesten. Választásunk a Szentendrei-szigeten található surányi vízbázisra, valamint a Csepel-szigeten található ráckevei vízbázisra esett. A két vízbázis vizsgálatának célja az volt, hogy a területen végzett terepi mérések és mintavételezések, valamint a modellezési szimulációk segítségével átfogó képet kapjunk ezen két terület hidrogeológiai viszonyairól, hidraulikai viselkedéséről. Ennek érdekében számos terepi mérést végeztünk, amelyek fontos adatokat szolgáltattak a két vízbázis hidrodinamikai modelljének felépítéséhez és kalibrációjához.

Summary.

In this study, we dealt with two significant riverbank filtration systems along the Danube: the Surány water base located on Szentendre Island and the Ráckeve water base located on Csepel Island. These are two important drinking water bases of the Budapest Water Works, which provide a significant proportion of the capital’s water needs and are of great importance from the point of view of water security. At these riverbank filtered water bases, water is produced with horizontal collector wells, which, thanks to their design, are capable of extracting large amounts of raw water. The aim of our work was to get to know the geological and hydrogeological conditions of the area using the tools of field measurements and geophysical measurements. To this end, we measured the following parameters in different temporal resolutions: water level, temperature, specific electrical conductivity, pH. In this study, we present the results that have more significant information. Geophysical tests were also carried out at the two water bases, which provided important information regarding the geological inhomogeneity. Using the field measurements, two hydrodynamic model was also built for each study area. Our field measurements were also a great help for the calibration of our model. In this study the calibration of these models was shown. The hydrodynamic models were performed in the modeling environment of the Groundwater Modeling System, using the MODFLOW code. The characteristic of the model is that the tentacle wells were defined using MODFLOW’s Revised Multi-Node Well package. With the help of these models we would like to determine the arrival times in the future, as well as examine different scenarios that can be linked to extreme weather conditions. Based on the field measurements, we can make the following main conclusions regarding the examined water bases:

• During the entire period of the year, the hydraulic gradient from the Danube side is greater than from the background direction. Based on this, we can say that the Danube is the dominant supply water source in both cases.

• The hydraulic gradient changes over time, which is constantly influenced by changes in the water level of the Danube and changes in the produced yields.

• As the hydraulic gradient changes, the speed of the water flow also changes, so the access time changes from time to time, even from day to day.

• The specific electrical conductivity measurements showed that the Danube’s flood causes a large decrease in the value of the specific conductivity in the northern area. In addition to the decrease in conductivity, the measurement results also show a time shift, which, does not necessarily show the value of the access time.

• Based on geophysical measurements, it can be said that the Ráckeve water base has greater inhomogeneity than the Surány water base. As a result, the zonal calibration of the hydrodynamic model of the Ráckeve water base became necessary.

As a means of assisting the selection of promising soil classification systems, a set of criteria were presented and tested. Inside the studied slightly saline plot World Reference Base (WRB) and Hungarian soil classification (HU) were compared at all four levels in terms of class separability, correlation to biomass, parsimony and homogeneity of classes. WRB surpassed HU in terms of the very important homogeneity of classes only, but HU performed better in terms of class separability, correlation to biomass and parsimony of classes. With many possible classification units WRB categorized the soil into a large number of classes, but 67% and 78% of them were single-profile classes at levels 3 and 4, respectively inside the ca 0.9 km² area.