Sustainable Strengthening Strategy For PC Bridges Based On Theory Of Planned Behavior And Multi-attribute Utility

Climate change is intensifying,and the ensuing rise in global temperatures will have huge negative impacts on global ecological security and economic development,therefore sustainable development cannot be ignored.Transportation,industry and construction account for more than 60%of the major sources of greenhouse gases generated by human activities,while energy consumption for the construction and maintenance of existing buildings accounts for about 30-40%of the global energy usage.Therefore,the construction industry(including infrastructure construction)is an important influencing factor in the process of sustainable development,which needs to be studied in depth.As China's economy continues to grow and its infrastructure work progresses steadily,China has become the country with the largest number of bridges in the world.Meanwhile,the number of degraded bridges in China is growing.The main-tenance of degraded bridges will not only bring huge economic costs,but also have a significant impact on the environment and society,causing a major impact on sustainable development.Due to the lack of financial resources,the lack of human resources,and the limitations of management tools and concepts,the allocation of maintenance resources is often based on subjective will,and lacks objective and scientific decision-making methods.In the context of climate change,and with the development of sustainable development,more and more factors need to be consid-ered in bridge maintenance strategy decisions.During the decades-long maintenance period,decision makers need to consider not only the safety and economy of the maintenance strategy,but also the environmental and social costs associated with the maintenance and strengthening activities.The decision maker in the traditional risk decision method is considered to be completely rational,which means the decision maker can choose the optimal strategy from the perspective of maximizing his own interests without subjective emotions.However,in real life,it is difficult to exist a fully rational person,and in the bridge maintenance strategy decision,the decision maker is influenced by his or her own knowledge,personal experience,and subjective emotions.When considering the sustainability of a maintenance strategy,many engineers face new problems and challenges due to the increased number of influencing factors to con-sider and the limited knowledge in environmental and social aspects.In order to develop a sustainable bridge maintenance strategy decision making approach that brings the results obtained from the normative model closer to the decisions made by decision makers in reality,the psychological factors of engineers in choosing maintenance strategies must be studied and incorporated into the model.A com-prehensive consideration of the economic,environmental,and social influences in bridge maintenance strategy decision making,as well as an in-depth analysis of the psychological factors of decision makers to identify the important influences,will not only improve the performance of the decision making method,but also lead to more sustainable maintenance strategies.The research work in this paper includes the following aspects.(1)The time-varying reliability of PC bridges is calculated based on the CO₂concentration and temperature variations under climate change conditions.Based on the multi-attribute utility theory,we analyzed the cost,economic,environmental and social attributes of PC bridges under the RCP8.5 climate change scenario,and established a sustainable PC bridge maintenance strategy optimization model,con-sidering four common maintenance methods(external prestressing method,FRP cloth/plate application method,cross-section enlargement method and adhesive steel method)based on performance control.The multi-attribute utility theory-based sustainable PC bridge maintenance strategy optimization model calculates the sustainable utility and cost utility of different maintenance strategies,and solves the multi-objective optimization with the objective functions of maximizing the cost utility and minimizing the sustainable utility,and using the non-dominated genetic algorithm with increasing elite strategy,density estimation and fast non-dominated ranking.(2)Based on the theory of planned behavior,the driving mechanism of bridge maintenance decision behavior is studied by extending the model factors of tradi-tional behavior theory.Taking bridge maintenance decision behavior as the research object,the influencing factors of maintenance reinforcement behavior are summa-rized from the perspectives of social psychology and social statistics.Based on this,the planned behavior theory model of bridge strengthening behavior is con-structed,and behavioral attitude,subjective norm,perceived behavioral control,knowledge,and perceived risk are selected as influencing factor indicators,and data are collected through questionnaire surveys to establish structural equation models to analyze the influence of each factor on maintenance decision behavior.(3)For the established PC bridge sustainable maintenance strategy optimiza-tion model,the attribute weights in the multi-attribute decision problem are ob-tained by hierarchical analysis and entropy value method,respectively,and then the comprehensive weights are calculated to ensure the accuracy of the weights to the maximum extent.The risk attitude and perceived risk of decision makers are analyzed,and the moderating effect of perceived risk on risk attitude is verified,and the sustainable PC bridge maintenance strategy optimization model is updated by the modified risk attitude.(4)To address the shortcomings of the computational complexity and high com-putational and time costs of the theoretical prediction model of planned behavior for sustainable maintenance behavior,the model data was trained and tested by er-ror backward propagation algorithm,support vector regression algorithm and light gradient boosting decision tree algorithm,and the corresponding machine learning prediction model was established to simplify the behavior prediction process and rank the importance of the factors influencing behavior.The relevance and sen-sitivity of the model variables were analyzed by maximum information correlation analysis as well as global sensitivity analysis.