Cristina Nichiforov

Dr. Cristina Nichiforov

University Politehnica of Bucharest, Bucharest, Bucureşti, Romania

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 October 2017
Verified
Conference Paper
Closed access
CNCristina Nichiforov
Cristina Nichiforov
ISIulia Stamatescu
Iulia Stamatescu
IFIoana Fagarasan
Ioana Fagarasan
Grigore StamatescuGrigore Stamatescu
Grigore Stamatescu
Energy forecast is essential for a good planning of the electricity consumption as well as for the implementation of decision support systems which can lead the decision making process of energy system. Energy consumption time series prediction problems represent a difficult type of predictive modelling problem due to the existence of complex linear and non-linear patterns. This paper presents two approaches for energy consumption forecast: an autoregressive integrated moving average (ARIMA) model and a non-linear autoregressive neural network (NAR) model. The two models are deeply described and finally compared in order to evaluate their performance.

1 October 2018
Verified
Conference Paper
Closed access
CNCristina Nichiforov
Cristina Nichiforov
Grigore StamatescuGrigore Stamatescu
Grigore Stamatescu
ISIulia Stamatescu
Iulia Stamatescu
VCVasile Calofir
Vasile Calofir
IFIoana Fagarasan
Ioana Fagarasan
SISergiu Stelian Iliescu
Sergiu Stelian Iliescu
As large scale energy management strategies have gradually shifted the focus from the producer to the consumer side, buildings are starting to play a critical role in the efficient management of the electrical grid. Moreover some buildings have become prosumers by integrating local generation capabilities from renewable sources thus inducing additional complexity into the operation of the energy systems. As alternative to conventional energy consumption modelling techniques, a blackbox input-output approach has the ability to capture underlying consumption patterns and trends while making use of the large quantities of data being generated and recorded through dense instrumentation of the buildings. The paper discusses and illustrates an approach to apply deep learning techniques, namely Recurrent Neural Networks implemented by means of Long Short-Term Memory layers, for load forecasting. We focus on large commercial buildings which can be better managed by central operators and where better models can result in significant energy savings and broad economic and social impact. The case study is illustrated on two university buildings from temperate climates over one year of operation using a reference benchmarking dataset for replicable results. The obtained results show promise and can be further used in reliable load management algorithms with limited overhead for periodic adjustments and model retraining.

1 June 2019 in Information
Verified
Journal Article
Open access
PDF file
CNCristina Nichiforov
Cristina Nichiforov
Grigore StamatescuGrigore Stamatescu
Grigore Stamatescu
Iulia StamatescuIulia Stamatescu
Iulia Stamatescu
Ioana FagarasanIoana Fagarasan
Ioana Fagarasan
Buildings play a critical role in the stability and resilience of modern smart grids, leading to a refocusing of large-scale energy-management strategies from the supply side to the consumer side. When buildings integrate local renewable-energy generation in the form of renewable-energy resources, they become prosumers, and this adds more complexity to the operation of interconnected complex energy systems. A class of methods of modelling the energy-consumption patterns of the building have recently emerged as black-box input–output approaches with the ability to capture underlying consumption trends. These make use and require large quantities of quality data produced by nondeterministic processes underlying energy consumption. We present an application of a class of neural networks, namely, deep-learning techniques for time-series sequence modelling, with the goal of accurate and reliable building energy-load forecasting. Recurrent Neural Network implementation uses Long Short-Term Memory layers in increasing density of nodes to quantify prediction accuracy. The case study is illustrated on four university buildings from temperate climates over one year of operation using a reference benchmarking dataset that allows replicable results. The obtained results are discussed in terms of accuracy metrics and computational and network architecture aspects, and are considered suitable for further use in future in situ energy management at the building and neighborhood levels.

1 June 2018
Verified
Conference Paper
Closed access
CNCristina Nichiforov
Cristina Nichiforov
Grigore StamatescuGrigore Stamatescu
Grigore Stamatescu
Iulia StamatescuIulia Stamatescu
Iulia Stamatescu
IFIoana Fagarasan
Ioana Fagarasan
SISergiu Stelian Iliescu
Sergiu Stelian Iliescu
As the built environment currently plays a determining role as driver of energy consumption in the developed world, with economic, social and environmental impact, better understanding of the underlying consumption models can lead to more reliable energy efficiency applications and programs. Within these frameworks, smart buildings generate large amounts of data from networks of embedded sensors and metering devices which have to be efficiently exploited for real-time decision and control. Lately, increasing attention is being given to black-box models which serve as a suitable alternative to more conventional system identification methods. In this context data-driven methods to derive reliable predictive models are needed on benchmark datasets that enable implementation and validation. We present the application and evaluation of feed-forward auto-regressive neural networks to intelligent load forecasting of smart buildings. By using public benchmarking datasets, replicable research is enabled as well as direct performance evaluation in both error metrics complexity. Experimental results are discussed with conclusion and outlook for implementation in energy management.

18 October 2021
Verified
Conference Paper
Closed access
CNCristina Nichiforov
Cristina Nichiforov
Miltiadis AlamaniotisMiltiadis Alamaniotis
Miltiadis Alamaniotis
The deployment and operation of advanced monitoring infrastructure in non-residential buildings have accommodated the acquisition of datasets in the form of long multivariate time series pertained to energy use. These datasets may be used to extract information in order to characterize and improve the building's operation by establishing energy consumption patterns, identifying abnormal operating modes, and detecting and anticipating faults ahead of time. The current research focuses on a two-fold data analysis method applied to academic building energy data. Initially, Matrix Profile (MP) is used for feature extraction - i.e., MP is a time-series data mining technique - to extract and build feature datasets of anomaly patterns from academic building energy traces. Second, the extracted feature datasets are fed to two supervised machine learning classification models aiming at discriminating the type of building load pattern. The case study is carried out on a set of public real-world energy load consumption patterns of four different types of academic buildings. The method can prove useful for exploiting complementary load patterns in a control infrastructure attaining efficient building energy management and reliable operation.

9 November 2021 in Energies
Verified
Journal Article
Open access
PDF file
CNCristina Nichiforov
Cristina Nichiforov
Antonio Martinez-MolinaAntonio Martinez-Molina
Antonio Martinez-Molina
Miltiadis AlamaniotisMiltiadis Alamaniotis
Miltiadis Alamaniotis
Building type identification is an important task that may be used in confirming and verifying its legitimate operation. One of the main sources of information over the operation of a building is its energy consumption, with the analysis of electricity patterns being at the spotlight of a non-intrusive identification approach. However, electricity patterns are the only source of information, and therefore, their analysis imposes several restrictions. In this work, we introduce a new approach in energy-driven identification by adding one more source of information beyond the electricity pattern that may be utilized, namely the gas consumption pattern. In particular, we propose a new intelligent approach that jointly analyzes the electricity–gas patterns to provide the type of building at hand. Our approach exploits the synergism of the matrix profile data analysis technique with a feed-forward artificial neural network. This approach has applicability in the energy waste elimination through the implementation of different energy efficiency solutions, as well as the optimization of the demand-side process management, safer and reliable operation through fault detection, and the identification and validation of the real operation of the building. The obtained results demonstrate the improvement in identifying the type of the building by employing the proposed approach for joint electricity–gas patterns as compared to only using the electricity patterns.