-
A generalised multi-factor deep learning electricity load forecasting model for wildfire-prone areas
Authors:
Weijia Yang,
Sarah N. Sparrow,
David C. H. Wallom
Abstract:
This paper proposes a generalised and robust multi-factor Gated Recurrent Unit (GRU) based Deep Learning (DL) model to forecast electricity load in distribution networks during wildfire seasons. The flexible modelling methods consider data input structure, calendar effects and correlation-based leading temperature conditions. Compared to the regular use of instantaneous temperature, the Mean Absol…
▽ More
This paper proposes a generalised and robust multi-factor Gated Recurrent Unit (GRU) based Deep Learning (DL) model to forecast electricity load in distribution networks during wildfire seasons. The flexible modelling methods consider data input structure, calendar effects and correlation-based leading temperature conditions. Compared to the regular use of instantaneous temperature, the Mean Absolute Percentage Error (MAPE) is decreased by 30.73% by using the proposed input feature selection and leading temperature relationships. Our model is generalised and applied to eight real distribution networks in Victoria, Australia, during the wildfire seasons of 2015-2020. We demonstrate that the GRU-based model consistently outperforms another DL model, Long Short-Term Memory (LSTM), at every step, giving average improvements in Mean Squared Error (MSE) and MAPE of 10.06% and 12.86%, respectively. The sensitivity to large-scale climate variability in training data sets, e.g. El Niño or La Niña years, is considered to understand the possible consequences for load forecasting performance stability, showing minimal impact. Other factors such as regional poverty rate and large-scale off-peak electricity use are potential factors to further improve forecast performance. The proposed method achieves an average forecast MAPE of around 3%, giving a potential annual energy saving of AU\$80.46 million for the state of Victoria.
△ Less
Submitted 20 April, 2023;
originally announced April 2023.
-
Spatial-Temporal Anomaly Detection for Sensor Attacks in Autonomous Vehicles
Authors:
Martin Higgins,
Devki Jha,
David Wallom
Abstract:
Time-of-flight (ToF) distance measurement devices such as ultrasonics, LiDAR and radar are widely used in autonomous vehicles for environmental perception, navigation and assisted braking control. Despite their relative importance in making safer driving decisions, these devices are vulnerable to multiple attack types including spoofing, triggering and false data injection. When these attacks are…
▽ More
Time-of-flight (ToF) distance measurement devices such as ultrasonics, LiDAR and radar are widely used in autonomous vehicles for environmental perception, navigation and assisted braking control. Despite their relative importance in making safer driving decisions, these devices are vulnerable to multiple attack types including spoofing, triggering and false data injection. When these attacks are successful they can compromise the security of autonomous vehicles leading to severe consequences for the driver, nearby vehicles and pedestrians. To handle these attacks and protect the measurement devices, we propose a spatial-temporal anomaly detection model \textit{STAnDS} which incorporates a residual error spatial detector, with a time-based expected change detection. This approach is evaluated using a simulated quantitative environment and the results show that \textit{STAnDS} is effective at detecting multiple attack types.
△ Less
Submitted 15 December, 2022;
originally announced December 2022.
-
Anomaly Detection for Industrial Big Data
Authors:
Neil Caithness,
David Wallom
Abstract:
As the Industrial Internet of Things (IIoT) grows, systems are increasingly being monitored by arrays of sensors returning time-series data at ever-increasing 'volume, velocity and variety' (i.e. Industrial Big Data). An obvious use for these data is real-time systems condition monitoring and prognostic time to failure analysis (remaining useful life, RUL). (e.g. See white papers by Senseye.io, an…
▽ More
As the Industrial Internet of Things (IIoT) grows, systems are increasingly being monitored by arrays of sensors returning time-series data at ever-increasing 'volume, velocity and variety' (i.e. Industrial Big Data). An obvious use for these data is real-time systems condition monitoring and prognostic time to failure analysis (remaining useful life, RUL). (e.g. See white papers by Senseye.io, and output of the NASA Prognostics Center of Excellence (PCoE).) However, as noted by Agrawal and Choudhary 'Our ability to collect "big data" has greatly surpassed our capability to analyze it, underscoring the emergence of the fourth paradigm of science, which is data-driven discovery.' In order to fully utilize the potential of Industrial Big Data we need data-driven techniques that operate at scales that process models cannot. Here we present a prototype technique for data-driven anomaly detection to operate at industrial scale. The method generalizes to application with almost any multivariate dataset based on independent ordinations of repeated (bootstrapped) partitions of the dataset and inspection of the joint distribution of ordinal distances.
△ Less
Submitted 9 April, 2018;
originally announced April 2018.
