ADASA uses Artificial Intelligence to improve dam monitoring and warning systems in New South Wales, Australia

2021-12-02

  • ADASA’s advanced alarm management system uses machine learning techniques and AI to provide efficient data analysis to generate reliable alarms and reduce possible dam failure risks
  • The technology comprises multivariate regression behavioural models, neural networks, and Random Forest algorithms to predict parameters like head pressure and seepage
Dam360

ADASA has developed a new methodology for analysing dam behaviour, using machine learning to support new alarm systems. Artificial Intelligence (AI) provides efficient data analysis that automatically creates analytical models to predict dam anomalies and generate reliable alarms, helping reduce the risk of potential dam failures.  

Our client, WaterNSW of Australia, had previously researched one of the largest dams in its New South Wales catchment area. We took this study and incorporated it into a new system comprising multivariate regression behavioural models, neural networks, and Random Forest algorithms to predict parameters such as head pressure and seepage.

We apply several unsupervised classification techniques to find anomalies at various dam locations:  

  • One-Class Support Vector Machine
  • Isolation Forest (anomaly detection)
  • Long Short-Term Memory (LSTM) or Large Short-Term Memory networks.

Our system now provides dam safety engineers with a new line of defence alarms, the ability to investigate anomalous situations more thoroughly, plus understand a dam’s long-term general tendencies and future health.

At the base of this advanced alarm management system is data generated by DamGuard, ADASA’s dam safety monitoring system, installed at 42 large dams across New South Wales (Australia) in recent years. DamGuard has allowed WaterNSW to collate all its historical, inspection, and dam safety data in a single — and real-time accessible — cloud-based database. This powerful information tool facilitates immediate anomaly detection and alarm generation. According to a study presented by WaterNSW at the last Australian National Committee on Large Dams (ANCOLD), DamGuard’s installation has led to an estimated 15% reduction in safety risks.

ADASA’s pioneering work has strengthened its position as a leading provider of dam safety and management solutions. It also reinforces ADASA’s leadership in the development and implementation of dam safety monitoring systems in Australia.

WaterNSW, based in Sydney, is responsible for the supply of bulk water and the management of 42 dams across the entire state of New South Wales.