Health vector (h)........................ 159 Stochastic prediction of severity levels . . . . . . . . . . . . 160 7.5.2 Action space A ......................... 161 7.5.3 Transition probability function P ............... 161 7.5.4 Reward function R....................... 162 7.6 Experimentalsetup........................... 163 7.6.1 Setup .............................. 163 7.6.2 Comparison of maintenance strategies . . . . . . . . . . . . 164 7.7 Results.................................. 164 7.7.1 Implementation and hyper-parameter tuning . . . . . . . . 164 7.7.2 Policy analysis: overview . . . . . . . . . . . . . . . . . . . 165 7.7.3 Policy analysis over episode . . . . . . . . . . . . . . . . . . 168 7.8 Discussionandconclusions. . . . . . . . . . . . . . . . . . . . . . . 171 7.9 References................................ 172 Discussion, Conclusions & Recommendations 175 8 Discussion 177 8.1 Reliability Modelling: Data-driven Inference of Fault Tree models . 177 8.2 Markov Process-based Prognostics: Multi-state deterioration modelling179 8.3 Maintenance Optimisation: Maintenance optimisation of multi-state components ............................... 182 8.4 Moving towards comprehensive Prognostics and Health Management: ClosingThoughts............................ 184 8.5 References................................ 186 9 Conclusions & Recommendations 187 9.1 Conclusions ............................... 187 9.2 Recommendations ........................... 188 9.2.1 Automatic Inference of Fault Tree Models . . . . . . . . . . 188 9.2.2 Multi-State Deterioration Modelling of Sewer Mains . . . . 189 9.2.3 Strategic Maintenance Planning for Sewer Mains using ReinforcementLearning....................... 191 9.3 References................................ 192 Appendices 195 A Appendix: Introduction 197 A.1 Example of a Multi-State Deterioration model with two states . . . 197 A.2 Example of a Multi-State Deterioration model with three states . . 200 B Appendixes: FT-MOEA 203 B.1 Data-driven methods to infer FTs from data . . . . . . . . . . . . . 203
RkJQdWJsaXNoZXIy MjY0ODMw