Bayesian Inference with Markov Chain Monte Carlo-Based Numerical Approach for Input Model Updating

Stochastic, discrete-event simulation modeling has emerged as a useful tool for facilitating decision making in construction. Owing to the rigidity inherent to distribution-based inputs, current simulation models have difficulty incorporating new data in real-time, and fusing these data with subjective judgments. Accordingly, application of this valuable technique is often limited to project planning stages. To expand implementation of simulation-based decision-support systems to the execution phase, this research proposes the use of Bayesian inference with Markov chain Monte Carlo (MCMC)-based numerical approximation approach as a universal input model updating methodology of stochastic simulation models for any given univariate continuous probability distribution. Found capable of (1) fusing actual performance with expert judgment, (2) integrating actual performance with historical data, and (3) processing raw data by absorbing uncertainties and randomness, the proposed method will considerably improve the resilience, reliability, accuracy, and practicality of stochastic simulation models, thereby enabling the application of stochastic simulation in the execution phase of construction.