This paper presents a novel particle filter based dynamic state estimation scheme for power systems where the states of all the generators are estimated. The proposed estimation scheme is decentralized in that each estimation module is independent from others and only uses local measurements. The particle filter implementation makes the proposed scheme numerically simple to implement. What makes this method superior to the previous methods which are mainly based on the Kalman filtering technique is that the estimation can still remain smooth and accurate in the presence of noise with unknown changes in covariance values. Moreover, this scheme can be applied to dynamic systems and noise with both Gaussian and non-Gaussian distributions.

Particle filter approach to dynamic state estimation of generators in power systems

Robust framework based on hybrid deep learning approach for short term load forecasting of building electricity demand

An Improved Hybrid Model for Short Term Power Load Prediction

Due to the influence of netizens’ behaviors, social activities, media and other factors, the trend of online public opinion shows the characteristics of nonlinear and seasonal fluctuation, but most researchers ignored it. In order to accurately predict the hot-degree of online public opinion, this paper proposes an improved seasonal grey decomposition and ensemble model. The STL decomposition algorithm is used to decompose original public opinion data. And the grey modified exponential model is proposed based on the grey difference information. Then the dynamic seasonal factors and Bernoulli equation are introduced to establish the seasonal modified exponential grey Bernoulli model. The SMEGBM model is used to predict the seasonal sequence and trend sequence, and the ARIMA model is used to predict the remainder sequence. In order to validate the prediction effect of the new model, the hot-degree predication of “Lin Shengbin” and “Tangshan beating” online events are implemented for empirical analysis. Compared with other models, the model proposed in this paper shows higher prediction accuracy. The results show that it is necessary to take the periodicity into account in the establishment of network public opinion model. And the hybrid model. can provide theoretical supports for relevant departments to monitor and give early warning of sudden online public opinion events. • A novel hybrid grey seasonal model is proposed to predict online public opinion. • Based on the grey differential information principle, MEGM(1,1) model is proposed. • The dynamic seasonal factors that extracted from seasonal sequence are proposed. • The Bernoulli equation is introduced to the establishment of SMEGBM model. • Comparative studies illustrate the effectiveness of the SMEGBM-ARIMA model. 

Online public opinion prediction based on a novel seasonal grey decomposition and ensemble model

The complication of existing electrical power systems is gradually increasing. This is stimulating researchers and developers to suggest new answers capable of addressing several challenges, particularly those related to power system operation. Massive penetration of asynchronously connected renewable energy generation, the generation connected over inverters is significantly changing the dynamics of modern power systems. Lately, the wide area monitoring system (WAMS) helps the power system operators to continuously analyze all the features of a large power network in real time. Utilizing phasor measurement units (PMUs), information can be recorded and monitored. Wide Area Monitoring Protection and Control (WAMPAC) enables power systems to detect disturbances and improve knowledge of network behavior under dynamic conditions, enabling system operators to maximize power flow and network stability. This chapter will give a basic overview of wide area monitoring and describe the components associated with it and its applications.

Introduction to Condition Monitoring of Wide Area Monitoring System

A novel seasonal segmentation approach for day-ahead load forecasting

State estimation is an important area required for enabling proper operation and control of the power system and its real-time monitoring. Correct knowledge of the states (like rotor angle and speed deviations) enables us to develop control schemes to enhance the system stability and reliability. Earlier the power system loads were almost constant and it was a static system as the load models became complex and the power system changed to quasi-static or sometimes dynamic it was not computationally possible to use older methods to achieve real-time monitoring. Dynamic State Estimation (DSE) techniques use the Dynamic model of the system in conjunction with the measurement model which can be used to accurately predict the states in advance. In this paper, various algorithms of traditional state estimation are presented, based on the available literature

A Review and Performance Comparison of Power System State Estimation Techniques

This paper presents a Long short-term memory (LSTM) based framework for a block and an hour ahead forecasting. LSTMs are the most suitable neural network architecture for sequence learning as it allows learning long-term dependencies without vanishing the gradients. The lagged variables have been used as separate features for load series, determined by studying the original series's correlogram. The lags with the highest autocorrelation have been chosen as the features. The proposed framework has been tested on real data of different zones of the Madhya Pradesh State Load Dispatch Centre (SLDC) in India. The results show that the proposed methodology outperforms the listed rival algorithms by a considerable margin.

LSTM-based short-term electrical load forecasting framework with improved input feature space

Accurate estimation of travel times is an important step in smart transportation and smart building systems. Poor estimation of travel times results in both frustrated users and wasted resources. Current methods that estimate travel times usually only return point estimates, losing important distributional information necessary for accurate decision-making. We propose using neural network-based mixture distributions to predict a user’s travel times given their origin and destination coordinates. We show that our method correctly estimates the travel time distribution, maximizes utility in a downstream elevator scheduling task, and is easy to retrain—making it a versatile and an inexpensive-to-maintain module when deployed in smart crowd management systems. International Workshop on Statistical Methods and Artificial Intelligence 2023 c © 2023 MERL. This work may not be copied or reproduced in whole or in part for any commercial purpose. Permission to copy in whole or in part without payment of fee is granted for nonprofit educational and research purposes provided that all such whole or partial copies include the following: a notice that such copying is by permission of Mitsubishi Electric Research Laboratories, Inc.; an acknowledgment of the authors and individual contributions to the work; and all applicable portions of the copyright notice. Copying, reproduction, or republishing for any other purpose shall require a license with payment of fee to Mitsubishi Electric Research Laboratories, Inc. All rights reserved. Mitsubishi Electric Research Laboratories, Inc. 201 Broadway, Cambridge, Massachusetts 02139 Available online at www.sciencedirect.com Procedia Computer Science 00 (2019) 000–000 www.elsevier.com/locate/procedia The 4th International Workshop on Statistical Methods and Artificial Intelligence (IWSMAI) March 15-17, 2023, Leuven, Belgium Travel-time prediction using neural-network-based mixture models Abhishek Sharma*a, Jing Zhangb, Daniel Nikovskib, Finale Doshi-Veleza aSEAS, Harvard University, Allston, MA, USA bMitsubishi Electric Research Labs, Cambridge, MA, USA Abstract Accurate estimation of travel times is an important step in smart transportation and smart building systems. Poor estimation of travel times results in both frustrated users and wasted resources. Current methods that estimate travel times usually only return point estimates, losing important distributional information necessary for accurate decision-making. We propose using neural network-based mixture distributions to predict a user’s travel times given their origin and destination coordinates. We show that our method correctly estimates the travel time distribution, maximizes utility in a downstream elevator scheduling task, and is easy to retrain—making it a versatile and an inexpensive-to-maintain module when deployed in smart crowd management systems.Accurate estimation of travel times is an important step in smart transportation and smart building systems. Poor estimation of travel times results in both frustrated users and wasted resources. Current methods that estimate travel times usually only return point estimates, losing important distributional information necessary for accurate decision-making. We propose using neural network-based mixture distributions to predict a user’s travel times given their origin and destination coordinates. We show that our method correctly estimates the travel time distribution, maximizes utility in a downstream elevator scheduling task, and is easy to retrain—making it a versatile and an inexpensive-to-maintain module when deployed in smart crowd management systems. © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the Conference Program Chairs.

Travel-time prediction using neural-network-based mixture models /Author=Sharma, Abhishek; Zhang, Jing; Nikovski, Daniel; Doshi-Velez, Finale /CreationDate=March 17, 2023 /Subject=Data Analytics, Machine Learning

Long-term load forecasting is required for power system planning, long-term investments, and expansion policies. In this paper, an ensemble of Long Short Term Memory (LSTM) and Convolutional Neural Network (CNN) is proposed that minimizes the forecasting error for the long-term. The features are extracted from the decomposition of load series using Variational Mode Decomposition (VMD) into various timescale subseries' and are combined with the exogenous variables like temperature and rain. This data is then used to train a Fully Connected Neural Network (FCNN) that produces the final output. This ensemble is expected to provide an accurate forecast as local variations in the series are captured by CNN and long-term dependencies by LSTM, while the VMD decomposes the raw data into low and high-frequency subsequences. The efficacy of the methodology is assessed on the actual load data of Madhya Pradesh (MP) State, India. It is observed that the proposed method has higher forecasting accuracy.

Abhishek Sharma

Papers

A novel seasonal segmentation approach for day-ahead load forecasting

A Review and Performance Comparison of Power System State Estimation Techniques

LSTM-based short-term electrical load forecasting framework with improved input feature space

Travel-time prediction using neural-network-based mixture models /Author=Sharma, Abhishek; Zhang, Jing; Nikovski, Daniel; Doshi-Velez, Finale /CreationDate=March 17, 2023 /Subject=Data Analytics, Machine Learning

VMD based Modified Multi-View Neural Network for Long-Term Load Forecasting