| |
|
rojects can fail in diverse ways and failure due to incidents (all unexpected events that cause or have the potential to cause losses; accidents are incidents that result in physical harm to people), especially those involving fatalities, are perhaps the most devastating ones, due to |
lives lost and the tarnished reputation of engineering companies involved. Some recent examples are the recent Nicoll Highway accident and the space shuttle Challenger accident. Consequently, safety should never be compromised and should always be a top priority for project managers.
To improve safety performance, it is widely accepted that the industry needs to implement efficient and effective safety planning prior to commencement of project activities. More importantly, if a feedback loop in the safety management system is implemented, safety planning in future projects can be improved. The challenge here is to adequately represent the safety knowledge in the industry, retrieve the relevant safety information from a knowledge base, and put the knowledge to use in the safety planning of new projects. Safety planning relies heavily on the experience and competence of the safety planning team. The process of identifying hazards, assigning appropriate level of risk and selecting the most efficient control, requires extensive field knowledge and experience. A valuable source of such experience resides in the investigation of past incidents and if it can be adequately exploited, it can significantly augment the experience of the safety planning team. Another rich source of knowledge lies in the safety plans of past projects. Each safety plan contains possible hazards and proposed risk control measures that would have been carefully considered. If they can be stored in the safety knowledge base, they can be useful for planning future projects as well.
|
|
|
Through our research in the Department of Civil Engineering, we have developed a novel case-based reasoning (CBR) approach to risk assessment and incorporated it into a Safety Knowledge Management System (SKMS). CBR is a relatively new branch of artificial intelligence, but in recent years there has been increasing use of CBR concepts. CBR has its roots in psychological theory of human reasoning, and is based on the intuitive paradigm that humans solve new problems by recalling past experiences or cases. A CBR system has three key processes: (1) case representation and indexing, (2) retrieval of cases, and (3) case utilisation and adaptation. These have been articulated earlier as the challenge for the implementation the feedback process.
Using the CBR methodology, SKMS first retrieves stored safety knowledge that resembles the present situation, using similarity scoring based on situational variables that describe the context of the knowledge. Innovative semantic nets have been developed to describe the nearness of the situational variables. The retrieved similar cases are then adapted to finally derive a suitable risk assessment tree that fits the present situation. The strategy utilizes a novel approach for removal and insertion of sub-trees based on a set of necessary conditions associated with breakdown events (see Figure 1).
A prototype SKMS was implemented in a realistic case study. This validated the CBR methodology to help companies produce risk assessments by incorporating past project knowledge, and facilitating continual improvement in their safety management system. The system can be conveniently extended to include safety control measures. Moreover, with further knowledge codification, it can be applied to design related errors so that accidents such as the Nicoll Highway collapse can be avoided.
|
|
