- Deep Learning
- Time Series Forecasting
- Statistical Modelling
- Business Intelligence
- Deep Learning
- Time Series Forecasting
- Statistical Modelling
- Business Intelligence
Finding relevant features that can represent different types of faults under a noisy environment is the key to practical applications of intelligent fault diagnosis. However, high classification accuracy cannot be achieved with only a few simple empirical features, and advanced feature engineering and modelling necessitate extensive specialised knowledge, resulting in restricted widespread use. This paper has proposed a novel and efficient fusion method, named MD-1d-DCNN, that combines statistical features from multiple domains and adaptive features retrieved using a one-dimensional dilated convolutional neural network. Moreover, signal processing techniques are utilised to uncover statistical features and realise the general fault information. To offset the negative influence of noise in signals and achieve high accuracy of fault diagnosis in noisy settings, 1d-DCNN is adopted to extract more dispersed and intrinsic fault-associated features, while also preventing the model from overfitting. In the end, fault classification based on fusion features is accomplished by the usage of fully connected layers. Two bearing datasets containing varying amounts of noise are used to verify the effectiveness and robustness of the suggested approach. The experimental results demonstrate MD-1d-DCNN’s superior anti-noise capability. When compared to other benchmark models, the proposed method performs better at all noise levels.
In the context of a resilient energy system, accurate residential load forecasting has become a non-trivial requirement for ensuring effective management and planning strategy/policy development. Due to the highly stochastic nature of energy load profiles, it is difficult to predict accurately, and usually, predictions are error-prone. This paper explores the potential of Empirical Mode Decomposition (EMD) in simplifying the dynamics of complex demand profiles. The simplified components are then embedded within a deep learning model, specifically Convolution Neural Network (CNN) and Long Short-Term Memory (LSTM), to forecast short-term residential loads. The novel modelling framework integrates Bayesian optimisation strategy, feature decomposition technique, feature engineering phase, and percentile-based bias correction algorithm to enhance model accuracy. The model is developed using a case-study residential dwelling located in Fintry (Scotland), and the model performance is assessed over four forecast horizons. The overall efficiency of framework is also investigated for three algorithms: random forest, gradient boosting decision trees (GBDT), and an LSTM network. While EMD and feature engineering were found to greatly improve prediction accuracy, the number of IMFs used was shown to significantly impact the model’s performance and computational complexity. The model was tested on two further case studies from Fintry.
Water inflow caused by tunneling can have severe impacts on the springs’ discharge rate. If these impacts have not been predicted beforehand, technical, economic, and environmental challenges could occur. While there are a few methods for evaluating the risk of water drawdown, their shortcomings create the need to develop a new one. First, in this research, five main tunneling projects in Iran were studied for evaluating the influence of tunneling on spring’s discharge, and a comprehensive database that contains information on 111 springs located in the vicinity of these tunneling projects was developed. Then, by learning from previously developed methods’ shortcomings and using an appropriate decision analysis method (Analytic Hierarchy Process or AHP), a new model was proposed for evaluating the risk of discharge reduction in springs located in the vicinity of tunneling projects. This new model, named TIS (Tunneling Impacts on Springs), was developed based on four important parameters of a) volume of water inflow toward the tunnel, b) distance between spring and tunnel, c) hydraulic connectivity, and d) aquifer recharge potential. In the next step, using data recorded in the database, TIS values were calculated for each spring, and using suitable statistical methods, the obtained TIS values were classified based on the actual behavior of springs. For using this model in practice, all springs must be checked using a screening process. In this process, according to some limitation criteria (including distance from the tunnel, groundwater condition in tunnel, spring origin), unimportant springs are excluded from the list and only springs with possible influence from tunneling are considered for further assessments. This helps to investigate the in-risk springs more effectively.
Load forecasting necessitates a significant amount of smart meter data. Several elements in this process, including device malfunctions and signal transmission issues, produce missing data gaps. Missing values in the dataset significantly influence the learning ability of machine learning algorithms, and they must be infilled before proceeding with any statistical analysis. This paper investigates the handling of missing values in demand data, and a new approach is developed for improving the performance of demand analytics, such as energy forecasting. The proposed model uses a transformer neural network to impute the missing values at various rates in the demand profile. Our model uses a k-means algorithm to fill in the missing values with proxy values in the dataset. The model is applied to two case-study residential house located in Cornwall and Fintry, United Kingdom. The developed algorithm is assessed for it potential for infilling missing values for three widely understood missing value scenarios: missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR). The proposed model's imputed outputs are compared to the original dataset to assess model performance. The performance of the framework is compared with a selection of widely used statistical and machine learning models. The proposed transformer model shows significant improvements over the common linear method in all three scenarios (with 30% missing values), with percentage improvements ranging from approximately 49.71% to 57.52% for Cornwall dataset.
Near accurate forecasting of energy demand has become a non-trivial requirement for developing effective management and planning strategies/policies for a resilient energy system. This paper is aimed to develop a novel deep learning-based energy demand prediction model by utilising the combination of Convolutional neural networks and Long Short-term Memory units. The proposed model consists of two one dimensional convolutional layer with max pooling, two bidirectional LSTM layers and finally three fully connected dense layer. The energy consumption data available for a household based in Findhorn ecovillage located in the north of Scotland for a six-week period during the February and March of 2015 was utilised to train, validate, and test the models. The proposed model provides energy demand prediction for short-term forecasting (5 minutes). The results obtained from the model are compared against four of the classical and widely applied algorithms for time series forecasting: autoregressive integrated moving average (ARIMA), light gradient boosting machine (LightGBM), random forest (RF), and deep neural networks (DNN). The result obtained demonstrated the efficiency of the proposed architecture in outperforming all well-established models.
Missing data are an integral part of a large dataset and one of the first key challenge that needs to be resolved effectively before conducting any reliable data-driven analytics or model development. Recently, several studies focused their attention on investigating the potential of computational approaches for missing data imputation. With readily available smart meter data, there is a growing demand for an efficient missing data imputation algorithm that can effectively capture the intrinsic patterns of highly stochastic dynamics of electricity demand data, specifically in the region of large gaps. However, due to the highly stochastic nature of electricity demand data, most of the conventional approaches facilitate limited scope and applicability. This paper is aimed to investigate the potential of a simple logical algorithm (developed by authors) in parallel to the widely applied ‘mice’(R package) algorithm for infilling high-resolution (1 minute) electricity demand data simultaneously at multiple sites. To optimise the performance of the ‘mice’algorithm a hierarchical cluster analysis using absolute correlation as a distance measure is utilised. The paper investigated a block of two months of data for 121 sites (methodologically), extracted from a large database of 661 sites monitored for almost a year, for a case-study community “Auroville” in India. The performance of both algorithms is intensively assessed using key statistical indicators involving discrete percentile distribution, continuous density distribution, and pattern analysis of large gaps in missing data.