Finding High Risk Hazards
As the province’s population continues to grow, maximizing the safety oversight we can offer to contractors and clients is of critical importance. However, we can’t be everywhere at once. As a result, we continue to evolve our business practices to maximize our resources to help us identify sites that pose the highest risk.
For many years, we have used a proprietary linear algorithm known as the Resource Allocation Program (RAP) to help prioritize sites that will be visited by a safety officer.
To help us better adapt to the ever-changing safety environment and leverage up and coming opportunities presented by artificial intelligence, we added machine-learning algorithms that predict where hazards are most likely to exist in British Columbia. As a result of this change, experiments we undertook in 2018 showed the algorithms improved the ability to predict high hazard electrical and gas installation work by 77% and 61% respectively.
increase in # of high hazards found with electrical installation permits
increase in # of high hazards found with gas installation permits
sample plans produced and tested
Our work in 2018 showed that in areas where we are implementing machine learning, our systems have become more predictive for what we prioritize as high risk hazards. This means that every time our systems identify a potential technical safety hazard, there is a higher chance of our safety officers actually finding a hazard once they conduct an inspection. At the same time, by improving efficiency of the software to discover high risk hazards, machine learning is helping us increase the number of inspections driven by safety officers’ discretion and other priorities identified by the organization as critical – for example, ammonia refrigeration units.
With more than 235,000 installation and operation permits issued in 2018, the scope of the safety system regulated by Technical Safety BC is significant. While our safety officers cannot be everywhere in the province at once, our focus on risk and innovation allows them to be in the places that matter the most – those that provide the highest risks. In 2019, we’ll continue to implement machine learning in other areas beyond electrical and gas installation to help us further increase the effectiveness of our safety hazard assessments.