Anomaly Detection in Power Generation Plants with Generative Adversarial Networks

Anomaly detection is a critical task that involves the identification of data points that deviate from a predefined pattern, useful for fraud detection and related activities. Various techniques are employed for anomaly detection, but recent research indicates that deep learning methods, with their ability to discern intricate data patterns, are well-suited for this task. This study explores the use of Generative Adversarial Networks (GANs) for anomaly detection in power generation plants. The dataset used in this investigation comprises fuel consumption records obtained from power generation plants operated by a telecommunications company. The data was initially collected to identify irregularities in the fuel consumption patterns of the generating sets located at the company's base stations. The dataset was subsequently divided into two categories: anomalous and normal data points, based on specific variables. Approximately 64.88% of the data was classified as normal, while 35.12% was classified as anomalous. An examination of feature importance, using the random forest classifier, revealed that the variable “Running Time Per Day” exhibited the highest relative importance. To enhance performance, a Generative Adversarial Network (GANs) model was trained and fine-tuned, both with and without data augmentation, in order to increase the dataset's size. The generator model consisted of five dense layers with the tanh activation function, while the discriminator included six dense layers, each incorporating a dropout layer to prevent overfitting. Following data augmentation, the model achieved an accuracy rate of 98.99%, in contrast to 66.45% without augmentation. This demonstrates that the model almost perfectly classified data points into normal and anomalous categories, with the augmented data significantly improving the GANs' performance in anomaly detection.