Distributed control schemes have transformed frequency and voltage regulation into a local task in distributed generators (DGs) rather than by a central secondary controller. A distributed scheme is based on information shared among neighboring units; thus, the microgrid performance is affected by issues induced by the communication network. This paper presents a distributed predictive control applied to the secondary level of microgrids. The model used for prediction purposes is based on droop and power transfer equations; however, communication features, such as connectivity and latency, are also included, thus making the controller tolerant to electrical and communication failures. The proposed controller considers the frequency and voltage regulation control objectives and consensus over the real and reactive power contributions from each power unit in the microgrid. The experimental and simulation results show that the proposed scheme (i) responds properly to load variations, working within operating constraints, such as generation capacity and voltage range; (ii) maintains the control objectives when a power unit is disconnected and reconnected without any user updating in the controllers; and (iii) compensates for the effects of communication issues over the microgrid dynamics.

Distributed Predictive Control for Frequency and Voltage Regulation in Microgrids

Deep learning has allowed a paradigm shift in pattern recognition, from using hand-crafted features together with statistical classifiers to using general-purpose learning procedures for learning data-driven representations, features, and classifiers together. The application of this new paradigm has been particularly successful in computer vision, in which the development of deep learning methods for vision applications has become a hot research topic. Given that deep learning has already attracted the attention of the robot vision community, the main purpose of this survey is to address the use of deep learning in robot vision. To achieve this, a comprehensive overview of deep learning and its usage in computer vision is given, that includes a description of the most frequently used neural models and their main application areas. Then, the standard methodology and tools used for designing deep-learning based vision systems are presented. Afterwards, a review of the principal work using deep learning in robot vision is presented, as well as current and future trends related to the use of deep learning in robotics. This survey is intended to be a guide for the developers of robot vision systems.

A Survey on Deep Learning Methods for Robot Vision.

Point and interval forecasting for wind speed based on linear component extraction

Currently available uncertainty quantification (UQ) neural networks (NNs) are trained through the statistical error minimization. Therefore, NNs perform poorly for critical input patterns. Some input patterns have lower coverage probabilities than others. Such input dependent performance is evident in electricity price prediction where different input features are coming from heterogeneous monitoring sources. This paper proposes a prediction interval (PI) based UQ of the electricity price with the proposed partial adversarial training to achieve the input-domain independent performance. The proposed training consists of initial training, adversarial sample generations from critical samples and a final training with the combined datasets of critical samples and initial training samples. Critical situations are situations where prediction systems struggle to predict and make higher statistical errors. Multiple NNs are first trained with different initializations. Each uncovered sample to the NN pair in the training set generates an adversarial sample. The adversarial dataset is concatenated with the initial samples. The final NN training is performed with the combined dataset. The technique is visualized with rough sketches in both time and input domain. The feasibility and performance are examined on experimental electricity market price datasets.

Partial Adversarial Training for Neural Network-Based Uncertainty Quantification

Interval forecasting for urban water demand using PSO optimized KDE distribution and LSTM neural networks

Prediction interval modelling has been proposed in the literature to characterize uncertain phenomena and provide useful information from a decision-making point of view. In most of the reported studies, assumptions about the data distribution are made and/or the models are trained at one step ahead, which can decrease the quality of the interval in terms of the information about the uncertainty modelled for a higher prediction horizon. In this paper, a new prediction interval modelling methodology based on fuzzy numbers is proposed to solve the abovementioned drawbacks. Fuzzy and neural network prediction interval models are developed based on this proposed methodology by minimizing a novel criterion that includes the coverage probability and normalized average width. The fuzzy number concept is considered because the affine combination of fuzzy numbers generates, by definition, prediction intervals that can handle uncertainty without requiring assumptions about the data distribution. The developed models are compared with a covariance-based prediction interval method, and high-quality intervals are obtained, as determined by the narrower interval width of the proposed method. Additionally, the proposed prediction intervals are tested by forecasting up to two days ahead of the load of the Huatacondo microgrid in the north of Chile and the consumption of the residential dwellings in the town of Loughborough, UK. The results show that the proposed models are suitable alternatives to electrical consumption forecasting because they obtain the minimum interval widths that characterize the uncertainty of this type of stochastic process. Furthermore, the information provided by the obtained prediction interval could be used to develop robust energy management systems that, for example, consider the worst-case scenario.

https://repository.tudelft.nl/islandora/object/uuid%3A4aaea97d-fd74-4ae0-b150-974d2592a21f/datastream/OBJ/download

Prediction interval methodology based on fuzzy numbers and its extension to fuzzy systems and neural networks

The main goal of this paper is to analyze the general problem of using Convolutional Neural Networks (CNNs) in robots with limited computational capabilities, and to propose general design guidelines for their use. In addition, two different CNN based NAO robot detectors that are able to run in real-time while playing soccer are proposed. One of the detectors is based on the XNOR-Net and the other on the SqueezeNet. Each detector is able to process a robot object-proposal in ~1ms, with an average number of 1.5 proposals per frame obtained by the upper camera of the NAO. The obtained detection rate is ~97%.

Using Convolutional Neural Networks in Robots with Limited Computational Resources: Detecting NAO Robots while Playing Soccer

This paper presents an approach for prediction interval generation by training a LSTM neural network with a joint supervision Loss Function. The prediction interval model provides the expected value and the upper and lower bounds of the interval given a desired coverage probability. The prediction interval models based on LSTM networks are compared with the classical recurrent neural network approach and are tested using two case studies. The first case corresponds to the forecasting up to one day ahead of the demand profile of 20 dwellings from a town in the UK, and the second case corresponds to the net power from an energy community made up 30 dwellings with a 50% level of photovoltaic power penetration. By using LSTM networks as the backbone of the proposed architecture, high-quality intervals are obtained with a narrower interval width compared with the classical recurrent neural network approach. Furthermore, the information provided by the prediction interval based on the LSTM network could be used to develop robust energy management systems that, for example, consider the worst-case scenario.

Prediction Intervals With LSTM Networks Trained By Joint Supervision

The main goal of this paper is to analyze the general problem of using Convolutional Neural Networks (CNNs) in robots with limited computational capabilities, and to propose general design guidelines for their use. In addition, two different CNN based NAO robot detectors that are able to run in real-time while playing soccer are proposed. One of the detectors is based on the XNOR-Net and the other on the SqueezeNet. Each detector is able to process a robot object-proposal in ~1 ms, with an average number of 1.5 proposals per frame obtained by the upper camera of the NAO. The obtained detection rate is ~97%.

Using Convolutional Neural Networks in Robots with Limited Computational Resources: Detecting NAO Robots While Playing Soccer

In this paper, a new prediction interval model based on a joint supervision loss function for capturing the uncertainties associated with the modeled phenomenon is described. This model provides the upper and lower bounds of the predicted values in accordance with the desired coverage probability, as well as their expected values. A benchmark problem is used to evaluate the proposed method, and a comparison with the neural network covariance method is performed. Additionally, the proposed method was applied to forecast the residential demand from a town in UK, considering the prediction interval performance for one-day ahead. The results show that the method is able to generate an interval with narrower width than the covariance method, and maintains the coverage probability. The information provided by the prediction interval could be used in the design of microgrid energy management systems.

Nicolás Cruz

Papers

Prediction interval methodology based on fuzzy numbers and its extension to fuzzy systems and neural networks

Using Convolutional Neural Networks in Robots with Limited Computational Resources: Detecting NAO Robots while Playing Soccer

Prediction Intervals With LSTM Networks Trained By Joint Supervision

Using Convolutional Neural Networks in Robots with Limited Computational Resources: Detecting NAO Robots While Playing Soccer

Neural Network Prediction Interval Based on Joint Supervision