We have built a Sydney Real-Time Digital Twin Platform, aiming at being the second generation of Digital Twin models for smart cities integrating multiple data sources such as: 3D city model with top layers including real-time public transport movement, transport simulation for incident scenario management in real-time, water pipes layout and IoT sensing data transmission, air quality and car parking real-time transmission, etc.

Digital Twins are powerful tools for operation planning, incident management and asset management. While most Digital Twins worldwide are static (operating at maximum level 2 of maturity), our Sydney Digital Twin model provides an uplift in its power of real-time analytics and deep learning prediction, which raises its functionality to Level 4 of maturity. To the best of our knowledge, this is the first real-time and AI-powered Digital Twin of an Australian city. Our approach represents a powerful step ahead of integrating artificial intelligence with spatial and temporal modelling to improve decision-making and optimise the cities we live in. The innovative aspect of our Sydney Digital Twin consists in its unique combination of 2D, 3D spatial modelling, real-time train movement information, water pipe and IoT sensing information, incident management simulation modelling as well as air quality real-time monitoring.

This project advanced rental estimation from static property-based modelling toward dynamic, market-aware forecasting. By combining spatio-temporal market patterns with property-level information, the work improved robustness, stability, and generalisation across regions.

The resulting framework provides a scalable analytical foundation for housing analytics, valuation modelling, and urban decision support in fast-changing property markets.

UTS and Sydney Trains have successfully applied advanced machine learning techniques to develop a timetable robustness evaluation model. The model can assess timetables and response plans to ensure that that timetables/response plans are operationally robust and resilient. The outcome of this application of the intelligent timetable evaluation technology significantly reduces delay-caused losses and increases the operation efficiency, enables the train operating system to meet performance metrics and recover from incidents.

The outcome of this application of the intelligent timetable evaluation technology significantly reduces delay-caused losses and increases the operation efficiency, enables the train operating system to meet performance metrics and recover from incidents. This work shows train operating companies that they can produce highly detailed and granular information to develop targeted timetable design and real-time scheduling strategies. This translates to improved railway network reliability and service to customers, and a reduced cost to serve. The proposed model is a generic model that can be easily applied to other traffic scenes with subtle refinements.

Smart parking solutions have become one of priorities in smart city projects. The major benefits of smart parking include decreasing traffic congestion, reducing energy consumption and reducing car emissions. With over 6.1 million of tourists to the Mornington Peninsula Shire annually, the council faced increasing challenges to provide both visitors and residents with a quality and fair parking experience. To meet those challenges, more than 700 in-ground and camera car parking bay sensors were installed in Rye and 100 in-ground sensors were installed in Mornington.

During the trial periods, we worked with the council on a data analytics project to discover parking behaviour patterns and predict the impact of parking-sensor-based parking enforcement. The occupancy and overstay patterns discovered in the project are helpful for the council to understand parking behaviours and improve the efficiency of parking enforcement in the trial areas.

Trains are an important means of transportation which serve millions of residents in mega cities. In order to keep the operation running smoothly, the performance of trains, such as punctuality and fuel efficiency, need to be reasonably predicted beforehand. The operator can then carry out timely adjustments on train operation, for example, changing timetable to prevent further delays, planning maintenance to reduce faults, and giving extra time for overcrowded stations.

The solution is built on innovative machine learning techniques including retrieval techniques, ensemble models, monitoring system, and continuous evaluation. The platform can aggregate different types of data sources in an agile manner with minor data format adjustments and provide more insights for hypotheses suggested by experts or knowledge discovered by additional data.

Trains play an increasingly important role in keeping a city moving and achieving urban sustainability. The train timetable should be evaluated regarding its punctuality, reliability and robustness before implementation. This project utilises a wealth of historical train performance, customer and incident data to evaluate a timetable and help create more robust timetables.

We solved the problem by integrating comprehensive data sources, identifying insights and patterns and developing advanced evaluation models using cutting-edge machine learning techniques including deep learning, ensemble learning and computer vision. The outcomes enable clients to evaluate a timetable effectively in several minutes.

With increasing demand for public transportation, transit agencies are challenged with a higher volume of passenger flow and longer queuing lines at stations. Railway passenger flow data can be used for planning facilities, train scheduling and other planning. It is very important to schedule train timetables properly so that they meet increasing passenger demand.

We developed real-time object detection and tracking models based on advanced computer vision techniques using installed cameras in stations. These solutions help train controllers and managers better understand passenger demand at a given station and better schedule future train timetables.

Major road incidents such as accidents, closures, or severe congestion can disrupt bus services and cause significant delays or cancellations for passengers. At present, understanding which bus routes and stops are affected often requires pulling information from multiple systems and manually checking details, which takes time during already critical situations. This makes it difficult for transport operators to quickly understand how many passengers are impacted and what actions are needed to minimise disruption.

This project delivers a data-driven tool to support faster and more informed responses to major road incidents. By combining road network data, bus route information, and historical passenger data, the tool can quickly identify affected bus routes and stops, estimate the number of impacted passengers, and highlight nearby alternative transport options. Through an interactive map-based interface, the tool brings key information together in one place, helping decision-makers respond more efficiently and reduce the overall impact on passengers during disruptions.

Maintaining water quality throughout the distribution system to the customers tap stands as one of the most formidable challenges faced by water utilities. This project seeks to identify and implement opportunities to optimize process control and monitoring feedback, resulting in greater consistency in total chlorine residual across the water distribution network.

The solution encompasses network topology modelling, estimation of water travel time, water quality modelling, and total chlorine prediction. By integrating these components, a comprehensive prediction model for water quality has been developed.

In the realm of water management, the silent challenge of leaks and breaks echoes across global water networks, disrupting communities and causing immense water loss. Sydney Water, in collaboration with the UTS Data Science Institute and UTS Robotics Institute, embarked on a transformative journey to tackle this challenge.

Over the past two years, Sydney Water and DSI demonstrated world-leading capabilities by adapting acoustic sensing to target non-surfacing leaks. The collaboration spans 270 semi-permanent and a variety of lift-and-shift acoustic sensors deployed across CBD areas and pressure zones.

In an era of escalating climate uncertainties, extreme weather events wreak havoc on water quality in catchments and reservoirs. The urgent need was clear: a solution to quantify the impact of climate change-induced extreme events on water quality in real-time.

The collaborative initiative leveraged advanced machine learning methods to build predictive water catchment models for NSW and Victoria assets, incorporating weather and upstream effects to predict raw water colour, natural organic matter, and turbidity.

Sydney Water, managing over 790km of sewer pressure mains, faced a critical challenge: predicting and preventing failures in this aging infrastructure. In response, they partnered with DSI to develop an innovative, data-driven solution using XGBoost to forecast potential failures across the network.

This work integrates data-driven models with forensic engineering and robotic sensing, creating a proactive sewer management system with implications for minimizing service disruptions, reducing repair costs, and mitigating environmental hazards.

Sydney Water sought a revolutionary solution to combat the widespread and costly issue of corrosion in its extensive network of gravity concrete sewers spanning approximately 823 kilometers.

A collaboration between Sydney Water and UTS developed a state-of-the-art system to predict and address corrosion hotspots using a combination of machine learning models and robotic systems.

Water utilities often manage water supply and demand for millions of properties. In this project, we developed new machine learning models for water demand forecasting, water quality modelling and dosing optimisation, which effectively predict water demand in the near future and define optimal strategies for chlorination.

New machine learning models are also being built to optimise energy usage, by balancing pumping regimes and costs while meeting water demand. All these models work together to optimise the water supply network.

Sewer corrosion is a serious issue in wastewater systems for water utilities. In order to improve the dosing effect and save chemicals, chemical dosing profiles should be optimised according to dynamic factors such as H2S level and flow volume.

This project developed a smart toolkit consisting of sewer dosing analysis, H2S estimation and sewer corrosion prediction. It enhances the capability of Sydney Water teams by improving performance and investment in controlling H2S, managing corrosion and odour.

Concrete sewer pipes are essential to modern cities, but many are ageing and slowly deteriorating due to harsh conditions inside the sewer environment. A major cause of this damage is microbial-induced corrosion, where gases and moisture allow bacteria to grow on pipe surfaces and gradually break down the concrete. Because this process happens out of sight and over long periods, utilities often rely on inspections or respond only once damage is already significant. This reactive approach can lead to unexpected failures, safety risks, and costly repairs or early pipe replacements.

This project aimed to provide a better way to understand and predict how long sewer pipes will last. Working with industry partners, the research team developed new sensors that can survive inside sewers and continuously measure conditions linked to corrosion, such as humidity, moisture, temperature, and gas levels. These measurements were combined with historical inspection data and advanced analytics in a digital twin platform, which creates a virtual representation of sewer assets. By linking real-world conditions with data-driven models, the project enables utilities to estimate remaining service life more accurately and plan maintenance proactively, helping extend asset life, reduce costs, and improve long-term infrastructure management.

Original Equipment Manufactures of electronic products often offer service warranty to customers. To maintain customer satisfaction, spare parts are required to be stocked at a sufficient level, but suppliers may stop producing certain models. It is therefore important to order one last quantity of spare parts, known as last time buy.

We built a visualisation tool that closely monitors engineering KPIs based on install base and failure data, and are developing predictive models that account for part failure, install base, stock and warranty data for forecasting last time buy quantity.

Understanding customer expectations is the key for broadband service providers. Predictions of activation demand and utility rate are challenging due to multiple interacting factors such as service types and local demographics.

Our team designed an innovative machine learning model for customer demand forecasting in fine-grained spatial and temporal dimensions. The techniques developed in this project can be applied to other customer-centered services and support better service rollout decisions.

Routine inspection of rail network is essential for the safety and efficient operation of trains. Train operators usually capture tens of millions of images of tracks and surrounding environment, which are then manually reviewed by inspectors to find defects.

Our team developed deep learning models to detect regions of interests for suspicious defects. The models detect missing clips and bolts from inspection images with high accuracy and improve inspection efficiency while maintaining high standards.

This project developed a unified multi-task framework for anomaly detection and intelligent monitoring in large-scale industrial and infrastructure systems.

By consolidating multiple detection and decision-support tasks into a single scalable architecture, the project improved efficiency, reduced redundancy, and strengthened the technical foundation for intelligent operations and asset management.