| Summary and Challenges |
- This project aims to utilize open weather data (online and historical) provided by the Hong Kong Observatory (HKO), available data of Mass Transit Railway Corporation Limited (MTR), and/or other data source to develop a trial AI model/solution for estimating the maximum allowable time for train operations on the MTR Corporation Limited railway network (MTR Network) upon hoisting of typhoon signal T9. One of the crucial factors that affects train operations under typhoon conditions is the wind speed.
- Project Challenges:
- Limited Online and Historical Data: The availability of weather data of online and historical from HKO/MTRCL/Market may be limited, especially for specific weather conditions such as typhoon T9 signal or above. The challenge is to ensure that the available data is representative enough to train a reliable AI estimation model.
- Data Quality and Completeness: The quality and completeness of the online and historical data may pose a challenge. Incomplete data can affect the performance and accuracy of the AI estimation model. Ensuring data integrity and addressing any data gaps or inconsistencies is crucial.
- Data Generalization: The online and historical data may not fully capture the variability and complexity of different weather scenarios. It is essential to develop the AI model/solution that can generalize well to unseen weather conditions and provide the best estimate on the maximum allowable time for train operations.
- Incorporating Relevant Factors: While wind speed is a one of the crucial factors affecting train operations under typhoon condition, there may be other factors that also play a significant role. Identifying and incorporating these additional factors into the AI model/solution can be challenging, particularly when the data is limited.
- Model Complexity and Interpretability: Developing the AI model/solution that accurately captures the relationship between wind speed and the maximum allowable time for train operations can be complex. Balancing model complexity with interpretability is essential to ensure actionable.
- Validation and Evaluation: Assessing the accuracy and reliability of the AI model/solution with limited data presents a challenge. Robust validation and evaluation processes need to be established to ensure that the model performs well and meets the required standards.
- Remarks: The solution provider should provide the I&T solution proposal to EMSD on or before the closing date. It should be included the preliminary design proposal with methodology & approach, data modelling, performance metrics, validation & testing, and budgetary ballpark cost estimate. EMSD may invite the proposer to conduct a presentation for introducing the proposal.
|
