The use of construction images in a safety assessment system
|dc.contributor.supervisor||Prof. David Scott|
This thesis sets out research carried out to investigate the usefulness of a descriptive database of construction methods for safety assessment. In addition, it investigates the possibility of utilising construction images as sources of safety related information.The construction industry has been recognized as a hazardous work environment with a high accident rate for years, hence, site safety is a problem. Nowadays, the use of construction images in the form of photographs is commonplace and they are used as sources of information. The literature review reveals that they have never been used as sources of information concerning construction safety practice.A preliminary investigation is conducted to study the possibility of utilizing construction images as sources of safety related information. The findings revealed that it is possible to use construction images as sources of information for construction safety assessment however, there were problems related to image interpretation and dissimilar safety judgment. It was assumed that those problems were arising from lack of safety knowledge and experiences, also no safety assessment method existed that could be used when using images as sources of information.To overcome problem related to the existence of safety assessment method, the research developed a method to assess safety by using information observed from images. As a similar safety judgment would be obtained from a same guideline therefore, a safety guideline was established, including safety checklist and safety scores. To give meaning of sets of safety score, two methods of conditional probability approach from Artificial Intelligence that quantitatively deal with uncertainty, the Bayes’ Theorem and the Fuzzy Logic Theory, were employed. The Bayes’ Theorem formula was adopted for calculating a likelihood of a hypothesis being true based on evidence or P(H/E). The hypothesis used in this research that a safe construction practice being used. The evidence used to test this hypothesis is information collected from construction images. This method allows construction practices shown in the images to be defined as having a high level of safety or low level of safety.The construction practices with low level of safety do not need to analysed further. Fuzzy logic theory can then be used for further classifying those images identified as having a high level of safety into one of three classifications: “most likely safe”, “fairly safe” or “most likely unsafe”.To overcome problem related to lack of safety knowledge and safety experience, one method of reasoning based on reuse past experience was employed, called the Case- Based Reasoning (CBR). The CBR method will allow safety information stored in database to be reused for reasoning process to give safety scores. As CBR works based on stored information from database therefore an image database has to be developed.Following works (or researches) have been done to overcome problems revealed from preliminary investigation therefore those works have to organize in a structured and systematic system. The research was developed a safety assessment system called SAFE AS.The safety assessment system worked in two processes, manual calculation and information storage into database. Manual calculation worked as follows: First, a construction practice judgment is given based on image data, safety checklist and using safety scores provided. Secondly, a construction practice is defined into one of two definitions provided: a high-level and a low-level of safety based on Bayes’ Theorem formula and given safety scores. Third, a high-level of safety of construction practice is classified into one of three classifications: most likely safe, fairly safe and most likely unsafe, which are developed, based on fuzzy sets formula. Following manual calculation process, the result from the process then become an input for the second process: information storage. All information of images and their safety practices are stored in an image database. These two processes are done separately and manually.Problem is arising from manual safety assessment system, that the processes are time-consuming. To overcome this problem, even to make a safety assessment system practically more benefit, the system is developed in a Web-based system, which allows safety assessment process and information storage process done comprehensively and automatically. All users can share their safety knowledge and experiences, and reuse stored experience as a basis of reasoning process from anywhere.As a result, the research has developed a Web-based safety assessment system to show how to utilize construction images to assess safe construction practice, store information from assessment process, and reuse this information for safety knowledge enhancement. Two experiments using 69 images and a set of detailed images have confirmed the application of a Web-based safety assessment system and verified its reliability.Another benefit from the safety assessment system is the safety likelihood scores obtained, which can be used to detect safety trends that are developing in construction project over time. These trends were used to predict the likely safety of the construction practices in use on the project in future so it can be used as indicators to monitor and control safety in construction projects. With this process construction images can be used as sources of safety related information and the safety assessment system can be used in future for predicting, monitoring and controlling of on-site safety.Areas needing future research are suggested, including providing advance search features in the assessment system to retrieve closer relevant cases for case-based reasoning and automated hazard recognition and identification feature to avoid accident occurrence as the result of human carelessness.
|dc.subject||safety related information|
|dc.title||The use of construction images in a safety assessment system|
|curtin.faculty||Faculty of Science and Engineering, Department of Civil Engineering|