Development of an innovative non rule-based energy management strategy for a Hybrid Electric Vehicle, structured for predictive driving controls based on external information knowledge

Recently, the interest of the automotive market for hybrid vehicles has increased due to the more restrictive pollutants emissions legislation and to the necessity of decreasing the fossil fuel consumption, since such solution allows a consistent improvement of the vehicle global efficiency. The term hybridization regards the energy flow in the powertrain of a vehicle: a standard vehicle has, usually, only one energy source and one energy tank; instead, a hybrid vehicle has at least two energy sources. In most cases, the prime mover is an internal combustion engine (ICE) while the auxiliary energy source can be mechanical, electrical, pneumatic or hydraulic. It is expected from the control unit of a hybrid vehicle the use of the ICE in high efficiency working zones and to shut it down when it is more convenient, while using the EMG at partial loads and as a fast torque response during transients. However, the battery state of charge may represent a limitation for such a strategy. That’s the reason why, in most cases, energy management strategies are based on the State Of Charge, or SOC, control. Several studies have been conducted on this topic and many different approaches have been illustrated. The purpose of this dissertation is to develop an online (usable on-board) control strategy in which the operating modes are defined using an instantaneous optimization method that minimizes the equivalent fuel consumption of a hybrid electric vehicle. The equivalent fuel consumption is calculated by taking into account the total energy used by the hybrid powertrain during the propulsion phases. The first section presents the hybrid vehicles characteristics. The second chapter describes the global model, with a particular focus on the energy management strategies usable for the supervisory control of such a powertrain. The third chapter shows the performance of the implemented controller on a NEDC cycle compared with the one obtained with the original control strategy.

I sistemi di raccomandazione in ambito e-learning

I sistemi di raccomandazione per come li conosciamo nascono alla fine del XX secolo, e si sono evoluti fino ai giorni nostri approcciandosi a numerosi campi, tra i quali analizzeremo l’ingegneria del software, la medicina, la gestione delle reti aziendali e infine, come argomento focale della tesi, l’e-Learning. Dopo una rapida panoramica sullo stato dell’arte dei sistemi di raccomandazione al giorno d’oggi, discorrendo velocemente tra metodi puri e metodi ibridi ottenuti come combinazione dei primi, analizzeremo varie applicazioni pratiche per dare un’idea al lettore di quanto possano essere vari i settori di utilizzo di questi software. Tratteremo nello specifico il funzionamento di varie tecniche per la raccomandazione in ambito e-Learning, analizzando tutte le problematiche che distinguono questo settore da tutti gli altri. Nello specifico, dedicheremo un’intera sezione alla descrizione della psicologia dello studente, e su come capire il suo profilo cognitivo aiuti a suggerire al meglio la giusta risorsa da apprendere nel modo più corretto. È doveroso, infine, parlare di privacy: come vedremo nel primo capitolo, i sistemi di raccomandazione utilizzano al massimo dati sensibili degli utenti al fine di fornire un suggerimento il più accurato possibile. Ma come possiamo tutelarli contro intrusioni e quindi contro violazioni della privacy? L’obiettivo di questa tesi è quindi quello di presentare al meglio lo stato attuale dei sistemi di raccomandazione in ambito e-Learning e non solo, in modo da costituire un riferimento chiaro, semplice ma completo per chiunque si volesse affacciare a questo straordinario ed affascinante mondo della raccomandazione on line.  

Named Entity Extraction: la piattaforma Gate/Annie

In questo lavoro si introducono i concetti di base di Natural Language Processing, soffermandosi su Information Extraction e analizzandone gli ambiti applicativi, le attività principali e la differenza rispetto a Information Retrieval. Successivamente si analizza il processo di Named Entity Recognition, focalizzando l’attenzione sulle principali problematiche di annotazione di testi e sui metodi per la valutazione della qualità dell’estrazione di entità. Infine si fornisce una panoramica della piattaforma software open-source di language processing GATE/ANNIE, descrivendone l’architettura e i suoi componenti principali, con approfondimenti sugli strumenti che GATE offre per l'approccio rule-based a Named Entity Recognition.

Near Real-Time Detection Of Pipe Bursts in Water Distribution Systems

The research work presented in the thesis describes a new methodology for the automated near real-time detection of pipe bursts in Water Distribution Systems (WDSs). The methodology analyses the pressure/flow data gathered by means of SCADA systems in order to extract useful informations that go beyond the simple and usual monitoring type activities and/or regulatory reporting , enabling the water company to proactively manage the WDSs sections. The work has an interdisciplinary nature covering AI techniques and WDSs management processes such as data collection, manipulation and analysis for event detection. Indeed, the methodology makes use of (i) Artificial Neural Network (ANN) for the short-term forecasting of future pressure/flow signal values and (ii) Rule-based Model for bursts detection at sensor and district level. The results of applying the new methodology to a District Metered Area in Emilia- Romagna’s region, Italy have also been reported in the thesis. The results gathered illustrate how the methodology is capable to detect the aforementioned failure events in fast and reliable manner. The methodology guarantees the water companies to save water, energy, money and therefore enhance them to achieve higher levels of operational efficiency, a compliance with the current regulations and, last but not least, an improvement of customer service.

Development of Optimal Energy Management Strategies for a Hybrid Boat

The present work proposes different approaches to extend the mathematical methods of supervisory energy management used in terrestrial environments to the maritime sector, that diverges in constraints, variables and disturbances. The aim is to find the optimal real-time solution that includes the minimization of a defined track time, while maintaining the classical energetic approach. Starting from analyzing and modelling the powertrain and boat dynamics, the energy economy problem formulation is done, following the mathematical principles behind the optimal control theory. Then, an adaptation aimed in finding a winning strategy for the Monaco Energy Boat Challenge endurance trial is performed via ECMS and A-ECMS control strategies, which lead to a more accurate knowledge of energy sources and boat’s behaviour. The simulations show that the algorithm accomplishes fuel economy and time optimization targets, but the latter adds huge tuning and calculation complexity. In order to assess a practical implementation on real hardware, the knowledge of the previous approaches has been translated into a rule-based algorithm, that let it be run on an embedded CPU. Finally, the algorithm has been tuned and tested in a real-world race scenario, showing promising results.

Development of a Predictive Control Function for a Plug-in Hybrid Electric Vehicle with Electrically-heated Catalyst

This master thesis work is focused on the development of a predictive EHC control function for a diesel plug-in hybrid electric vehicle equipped with a EURO 7 compliant exhaust aftertreatment system (EATS), with the purpose of showing the advantages provided by the implementation of a predictive control strategy with respect to a rule-based one. A preliminary step will be the definition of an accurate powertrain and EATS physical model, starting from already existing and validated applications. Then, a rule-based control strategy managing the torque split between the electric motor (EM) and the internal combustion engine (ICE) will be developed and calibrated, with the main target of limiting tailpipe NOx emission by taking into account EM and ICE operating conditions together with EATS conversion efficiency. The information available from vehicle connectivity will be used to reconstruct the future driving scenario, also referred to as electronic horizon (eHorizon), and in particular to predict ICE first start. Based on this knowledge, an EATS pre-heating phase can be planned to avoid low pollutant conversion efficiencies, thus preventing high NOx emission due to engine cold start. Consequently, the final NOx emission over the complete driving cycle will be strongly reduced, allowing to comply with the limits potentially set by the incoming EURO 7 regulation. Moreover, given the same NOx emission target, the gain achieved thanks to the implementation of an EHC predictive control function will allow to consider a simplified EATS layout, thus reducing the related manufacturing cost. The promising results achieved in terms of NOx emission reduction show the effectiveness of the application of a predictive control strategy focused on EATS thermal management and highlight the potential of a complete integration and parallel development of involved vehicle physical systems, control software and connectivity data management.

New markov chain based methods for single and cross-domain sentiment classification

Nowadays communication is switching from a centralized scenario, where communication media like newspapers, radio, TV programs produce information and people are just consumers, to a completely different decentralized scenario, where everyone is potentially an information producer through the use of social networks, blogs, forums that allow a real-time worldwide information exchange. These new instruments, as a result of their widespread diffusion, have started playing an important socio-economic role. They are the most used communication media and, as a consequence, they constitute the main source of information enterprises, political parties and other organizations can rely on. Analyzing data stored in servers all over the world is feasible by means of Text Mining techniques like Sentiment Analysis, which aims to extract opinions from huge amount of unstructured texts. This could lead to determine, for instance, the user satisfaction degree about products, services, politicians and so on. In this context, this dissertation presents new Document Sentiment Classification methods based on the mathematical theory of Markov Chains. All these approaches bank on a Markov Chain based model, which is language independent and whose killing features are simplicity and generality, which make it interesting with respect to previous sophisticated techniques. Every discussed technique has been tested in both Single-Domain and Cross-Domain Sentiment Classification areas, comparing performance with those of other two previous works. The performed analysis shows that some of the examined algorithms produce results comparable with the best methods in literature, with reference to both single-domain and cross-domain tasks, in $2$-classes (i.e. positive and negative) Document Sentiment Classification. However, there is still room for improvement, because this work also shows the way to walk in order to enhance performance, that is, a good novel feature selection process would be enough to outperform the state of the art. Furthermore, since some of the proposed approaches show promising results in $2$-classes Single-Domain Sentiment Classification, another future work will regard validating these results also in tasks with more than $2$ classes.

An inter-comparison between a VIS/IR and a MW satellite-based methods for cloud detection and classification

This thesis is aimed to assess similarities and mismatches between the outputs from two independent methods for the cloud cover quantification and classification based on quite different physical basis. One of them is the SAFNWC software package designed to process radiance data acquired by the SEVIRI sensor in the VIS/IR. The other is the MWCC algorithm, which uses the brightness temperatures acquired by the AMSU-B and MHS sensors in their channels centered in the MW water vapour absorption band. At a first stage their cloud detection capability has been tested, by comparing the Cloud Masks they produced. These showed a good agreement between two methods, although some critical situations stand out. The MWCC, in effect, fails to reveal clouds which according to SAFNWC are fractional, cirrus, very low and high opaque clouds. In the second stage of the inter-comparison the pixels classified as cloudy according to both softwares have been. The overall observed tendency of the MWCC method, is an overestimation of the lower cloud classes. Viceversa, the more the cloud top height grows up, the more the MWCC not reveal a certain cloud portion, rather detected by means of the SAFNWC tool. This is what also emerges from a series of tests carried out by using the cloud top height information in order to evaluate the height ranges in which each MWCC category is defined. Therefore, although the involved methods intend to provide the same kind of information, in reality they return quite different details on the same atmospheric column. The SAFNWC retrieval being very sensitive to the top temperature of a cloud, brings the actual level reached by this. The MWCC, by exploiting the capability of the microwaves, is able to give an information about the levels that are located more deeply within the atmospheric column.

Lightweight Internet Traffic Classification - A Subject Based Solution with Word Embeddings

Internet traffic classification is a relevant and mature research field, anyway of growing importance and with still open technical challenges, also due to the pervasive presence of Internet-connected devices into everyday life. We claim the need for innovative traffic classification solutions capable of being lightweight, of adopting a domain-based approach, of not only concentrating on application-level protocol categorization but also classifying Internet traffic by subject. To this purpose, this paper originally proposes a classification solution that leverages domain name information extracted from IPFIX summaries, DNS logs, and DHCP leases, with the possibility to be applied to any kind of traffic. Our proposed solution is based on an extension of Word2vec unsupervised learning techniques running on a specialized Apache Spark cluster. In particular, learning techniques are leveraged to generate word-embeddings from a mixed dataset composed by domain names and natural language corpuses in a lightweight way and with general applicability. The paper also reports lessons learnt from our implementation and deployment experience that demonstrates that our solution can process 5500 IPFIX summaries per second on an Apache Spark cluster with 1 slave instance in Amazon EC2 at a cost of $ 3860 year. Reported experimental results about Precision, Recall, F-Measure, Accuracy, and Cohen's Kappa show the feasibility and effectiveness of the proposal. The experiments prove that words contained in domain names do have a relation with the kind of traffic directed towards them, therefore using specifically trained word embeddings we are able to classify them in customizable categories. We also show that training word embeddings on larger natural language corpuses leads improvements in terms of precision up to 180%.

SynBA: A contextualized Synonim-Based adversarial Attack for text classification

With the advent of high-performance computing devices, deep neural networks have gained a lot of popularity in solving many Natural Language Processing tasks. However, they are also vulnerable to adversarial attacks, which are able to modify the input text in order to mislead the target model. Adversarial attacks are a serious threat to the security of deep neural networks, and they can be used to craft adversarial examples that steer the model towards a wrong decision. In this dissertation, we propose SynBA, a novel contextualized synonym-based adversarial attack for text classification. SynBA is based on the idea of replacing words in the input text with their synonyms, which are selected according to the context of the sentence. We show that SynBA successfully generates adversarial examples that are able to fool the target model with a high success rate. We demonstrate three advantages of this proposed approach: (1) effective - it outperforms state-of-the-art attacks by semantic similarity and perturbation rate, (2) utility-preserving - it preserves semantic content, grammaticality, and correct types classified by humans, and (3) efficient - it performs attacks faster than other methods.