Search

Process of human-machine interaction is an integral part of everyday human life in a modern world. The various interfaces are intended to facilitate this interaction and provide maximum comfort for users. The speech interface is the most universal, and it can be used in a vast majority of systems. Speech is the main way of communication between people, from this point of view speech interface looks completely natural. Nowadays, human-machine speech interfaces can consist of many complex components in order to implement an interaction with a user. However, a primary element of such interfaces is an automatic speech recognition (ASR) system. Automatic recognition of spontaneous speech in a telephone channel can be solved very successfully in the presence of significant amount of speech training data, but the collection and processing of such data is a complex, time-consuming and expensive task. Great amount of data is available only for high-resource languages such as English, Spanish, German, Chinese, French, etc. At the same time, there are many languages with no significant amount of training data for ASR systems preparation. Such languages are commonly referred to as low-resource languages. The number of native speakers for such languages can reach tens, and sometimes hundreds of millions of people. Taking this into account, building ASR systems for such languages is an important task. The main objective main objective of this thesis is the development of universal methodology for robust speech recognition of low-resource languages. To achieve it, we have solved three tasks. First, we have developed a flexible methodology in order to build an ASR system for large vocabulary conversational speech on low-resource languages. Second, we have identified a set of languages representing a wide variety of acoustic and grammatical features that influence on ASR systems. Third, we have conducted experimental studies of the developed universal methodology and compared the accuracy of the resulting ASR systems with the state-of-the-art results. We have contributed to several key components of the ASR system as part of the first task. We have proposed speaker-dependent bottleneck acoustic features trained with a multilingual approach, to improve acoustic modeling. Besides, a new approach of combining acoustic features has been proposed. In terms of language modeling, we have proposed a new method of text data augmentation with the artificial texts generated by the Char-RNN model. Finally, we have created a universal methodology that combines both existing and proposed methods. This methodology represents a complete pipeline for building ASR systems for low-resource languages. As a part of the second task, we have identified the most important language features which affect the building process of ASR systems. We have considered such factors as word order, agglutination, inflection, presence of an accent, dialects, and tone. Based on the analysis, we have created a set of low-resource languages that covers the factors described above and several additional features. This set includes eight languages: Georgian, Turkish, Kazakh, Vietnamese, Swahili, Tagalog, Zulu, and Russian. As a part of the third task, we have conducted large-scale experimental studies. We have identified the most effective basic acoustic features and the training pipeline for initial GMM-HMM models. We have verified proposed multilingual speaker-dependent bottleneck features and confirmed the effectiveness of the delayed combination for acoustic features. We have determined the usefulness of the proposed text data augmentation approach. Moreover, we have found this approach complementary to the addition of filtered web-text data. We have experimented with various neural network-based acoustic models and identified the most effective ones. Finally, we have compared the accuracy of resulting ASR systems with the state-of-the-art results for studied low-resource languages. We believe that this work will accelerate further research of speech recognition in low-resource languages and make speech recognition systems available for a wide variety of world languages in the near future., Diese Dissertation entstand im Rahmen einer Kooperation mit der Universit��t ITMO in St. Petersburg. / This dissertation was written in the context of a cooperation with ITMO University in St. Petersburg.

Energy systems are in a major process of transition from both technology and business perspective. The amount of distributed generation is increasing, load profiles are changing and new types of controllable and uncontrollable resources are being connected to the system (e.g. storage units, electric vehicles). In general, these resources are called distributed energy resources (DERs). The role of microgrids, energy communities and aggregators (virtual power plants) is being particularly emphasized in the electrical system because they create an opportunity for new kind of flexibility provided by DERs, but also introduce new challenges to energy system management. Business potential of intelligent energy solutions is enormous but there are still major barriers that block most of the novel business opportunities in the present energy system. One of the most significant technical barrier is the lack of widely accepted interoperable information exchange solution (data model and interfaces) that is easily accessible and fulfills business needs by all parties dealing with the energy system.HEILA project aims to define, implement and demonstrate an integrated business platform of DERs for information exchange between energy market participants. The first implementation of such a platform will be realized during the project to integrate smart grid demonstrations in Finland to develop, test, pilot and finally also commercialize new smart energy system functionalities consisting of interactions and impacts of multiple participants. Development of such a system begins as a national smart grid demonstration platform in Finland but will be designed to support Europe wide deployment of the system.

This research project aimed at establishing an interconnection between multiple battery storage units as well as defining and testing operation strategies for battery energy systems in different use cases. The project took full benefit of the existing battery storage infrastructure located in Helsinki (Suvilahti), Suomenniemi (storage in an LVDC microgrid) and Lappeenranta (LUT Green Campus stationary and mobile storages). Each of these storages was in active operation already before the project started. However, the storages are operated independently of each other, and their operation is not yet fully optimized for the needs of electricity markets and the power system. In the research project, a storage system has a stakeholder-specific multi-objective role, which means that the storage system has to respond to several service requests simultaneously. This may mean, for instance, operating at the same time in the frequency control in the electricity markets, trading electricity in the day-ahead, intraday, and ancillary markets, simultaneously offering various services to local network operations and several other stakeholders. This kind of multi-objective operation requires full understanding of interactions of different markets and stakeholders and risks related to the conflicting objectives of the stakeholders. One of the key outcomes of the project work was the establishment of a connection to the Suvilahti BESS unit through an IEC 104 protocol. Another outcome was constructing a simulation tool in Matlab that enables testing of numerous scenarios of a single BESS unit operation with different operating parameters and various operating strategies. The major part of the analyses was done based on the results of the simulation tool. There are two further main outcomes of the project. The first one is that it is technically possible to remotely control multiple BESS units against multiple tasks according to a pre-defined logic. The second outcome of the project is that a BESS can and should be operated against multiple tasks simultaneously. During such an operation, there may or may not emerge a conflict of objectives between the involved stakeholders. The nature of the conflicts has been investigated and the methods to mitigate the conflict have been analysed. The aggregation of BESS resources is one way to mitigate the conflict of objectives. Publishers version