DYNAMIC RESCUE FORCES DISPATCHING AND EMERGENCY RESOURCE ALLOCATION FOR URBAN WATERLOG DISASTERS

An urban waterlog disaster can pose a serious threat to human life and property safety. This thesis makes shortest path planning, dynamic scheduling of rescue forces (firefighters and fire engines) and applies reinforcement learning to dynamic emergency resource allocation in face of multi-point urban waterlog disasters triggered by short-term intensive rainfall. To dispatch emergency rescue forces dynamically from multiple fire stations to multiple flooding sites, a two-step approach is proposed. The first step is to acquire the optimal travel times from fire stations to flooding sites under the extension of floods. We build a path selection model for rescue vehicles in times of floods. A customized A* algorithm is developed to solve the model with a high search efficiency, which can also adapt to the time-dependent road network under disaster scenarios. Additionally, a preference-based customized A* algorithm is designed to select paths that contain users’ preferences, providing an alternative for the decision-maker to select rescue paths. The second step is to estimate the demand for firefighters at each flooding site based on the population density and the detected real-time waterlog depths. Taking the optimal travel times obtained in the first step are used as input, we build a bi-objective dynamic dispatching model to dispatch emergency rescue forces dynamically fire stations to multiple urban flooding sites. A case study in Futian District of Shenzhen, China is presented to demonstrate the practicability of our rescue planning. To make emergency resource distribution and allocation dynamically for urban waterlog disasters, a 3-layer emergency logistics network is built to control the flow of relief supplies: from warehouses to transfer stations to local flooding sites. A multi-period dynamic integer programming model is formulated to solve this emergency location, allocation and distribution problem. Two scenarios are taken into account: one is deterministic case and another one is stochastic case. The deterministic model is solved by both Gurobi Optimizer and Reinforcement Learning Approach (without Exploration), whereas an 𝜖- Greedy Reinforcement Learning Approach is developed to solve the stochastic case. A hypothetical disaster case in Luohu District of Shenzhen, China is implemented to verify the effectiveness of our model and algorithms. This study actually builds an dynamic emergency response framework for urban wa-terlog disasters, providing quick auxiliary decision support for decision-makers.

[1] HU S L, HAN C F, MENG L P. A scenario planning approach for propositioning rescue centersfor urban waterlog disasters[J]. Computers & Industrial Engineering, 2015, 87: 425-435.
[2] Ministry of Emergence Management of China. National natural disasters in the first three quar-ters of 2021[EB/OL]. 2021. https://www.mem.gov.cn/xw/yjglbgzdt/202110/t20211010_399762.shtml.
[3] HOLLINGSWORTH J. At least 15 dead after heavy rainfall and flooding in northernchina[EB/OL]. 2021. https://edition.cnn.com/2021/10/12/china/flooding-china-shanxi-ntl-hnk/index.html.
[4] GAN N, YEUNG J. ’once in a thousand years’ rains devastated central china, but there is littletalk of climate change[EB/OL]. 2021. https://edition.cnn.com/2021/07/23/china/china-flood-climate-change-mic-intl-hnk/index.html.
[5] YANG L, LIU Q, YANG S H, et al. Evacuation planning with flood inundation as inputs[J].2015.
[6] CHANG M S, TSENG Y L, CHEN J W. A scenario planning approach for the flood emergencylogistics preparation problem under uncertainty[J]. Transportation Research Part E, 2007, 43(6): 737-754.
[7] COMFORT L K, KO K, ZAGORECKI A. Coordination in rapidly evolving disaster responsesystems: The role of information[J]. American behavioral scientist, 2004, 48(3): 295-313.
[8] London Fire Brigade. Flood response[EB/OL]. 2021. https://www.london-fire.gov.uk/about-us/services-and-facilities/techniques-and-procedures/flood-response/.
[9] Ministry of Public Security of the PRC. GA/T 1340-2016 classification of fire alarms andemergency rescue operations[EB/OL]. 2016. http://www.doc88.com/p-9922813877509.html.
[10] YUAN Y, WANG D. Path selection model and algorithm for emergency logistics manage-ment[J]. Computers & Industrial Engineering, 2009, 56(3): 1081-1094.
[11] AHMADI M, SEIFI A, TOOTOONI B. A humanitarian logistics model for disaster relief op-eration considering network failure and standard relief time: A case study on san franciscodistrict[J]. Transportation Research Part E, 2015, 75: 145-163.
[12] TIRKOLAEE E B, AYDıN N S, RANJBAR-BOURANI M, et al. A robust bi-objective mathe-matical model for disaster rescue units allocation and scheduling with learning effect[J]. Com-puters & Industrial Engineering, 2020, 149: 106790.
[13] SUN Y. A reliability-based approach of fastest routes planning in dynamic traffic networkunder emergency management situation[J]. International Journal of Computational IntelligenceSystems, 2011, 4(6): 1224.
[14] ZHANG X, ZHANG Z, ZHANG Y, et al. Route selection for emergency logistics management:A bio-inspired algorithm[J]. Safety science, 2013, 54: 87-91.
[15] ZHAO J, GUO Y, DUAN X. Dynamic path planning of emergency vehicles based on traveltime prediction[J]. Journal of Advanced Transportation, 2017, 2017: 1-14.
[16] GAI W, DU Y, DENG Y. Multi-objective route planning model and algorithm for emergencymanagement[M]//Decision-making Analysis and Optimization Modeling of Emergency Warn-ings for Major Accidents. [S.l.]: Springer, 2019: 113-150.
[17] MENG X Z, ZHOU H, HU X B. Many-to-many path planning for emergency material trans-portation in dynamic environment[J]. 2020 IEEE Symposium Series on Computational Intelli-gence (SSCI), 2020: 276-280.
[18] FIEDRICH F, GEHBAUER F, RICKERS U. Optimized resource allocation for emergencyresponse after earthquake disasters[J]. Safety Science, 2000, 35(1): 41-57.
[19] GONGQ,BATTAR. Allocationandreallocationofambulancestocasualtyclustersinadisasterrelief operation[J]. Iie Transactions, 2007, 39(1): 27-39.
[20] YU L, ZHANG C, YANG H, et al. Novel methods for resource allocation in humanitarianlogistics considering human suffering[J]. Computers & Industrial Engineering, 2018, 119: 1-20.
[21] YU L, YANG H, MIAO L, et al. Rollout algorithms for resource allocation in humanitarianlogistics[J]. IISE Transactions, 2019, 51(8): 887-909.
[22] XIANG Y, ZHUANG J. A medical resource allocation model for serving emergency victimswith deteriorating health conditions[J]. Annals of Operations Research, 2016, 236(1): 177-196.
[23] HU C L, LIU X, HUA Y K. A bi-objective robust model for emergency resource allocationunder uncertainty[J]. International Journal of Production Research, 2016: 7421-7438.
[24] ZHOU Y, LIU J, ZHANG Y, et al. A multi-objective evolutionary algorithm for multi-perioddynamicemergencyresourceschedulingproblems[J]. TransportationResearchPartELogistics& Transportation Review, 2017, 99: 77-95.
[25] CAO C J, LI C D, YANG Q, et al. A novel multi-objective programming model of reliefdistribution for sustainable disaster supply chain in large-scale natural disasters[J]. Journal ofCleaner Production, 2018: 1422-1435.
[26] SHAVARANI S M. Multi-level facility location-allocation problem for post-disaster humani-tarian relief distribution: A case study[J]. Journal of Humanitarian Logistics and Supply ChainManagement, 2019, 9(2).
[27] COOKRA,LODREEEJ. Dispatchingpoliciesforlast-miledistributionwithstochasticsupplyand demand[J]. Transportation Research Part E Logistics & Transportation Review, 2017, 106:353-371.
[28] WANG F, PEI Z, DONG L, et al. Emergency resource allocation for multi-period post-disasterusing multi-objective cellular genetic algorithm[J]. IEEE Access, 2020, 8.
[29] WANG Y. Multiperiod optimal allocation of emergency resources in support of cross-regionaldisaster sustainable rescue[J]. International Journal of Disaster Risk Science, 2021, 12(3): 394-409.
[30] SHIN K, LEE T. Emergency medical service resource allocation in a mass casualty incident byintegrating patient prioritization and hospital selection problems[J]. IISE Transactions, 2020,52(10): 1141-1155.
[31] SHEU J B. An emergency logistics distribution approach for quick response to urgent reliefdemand in disasters[J]. Transportation Research Part E Logistics & Transportation Review,2007, 43(6): 687-709.
[32] ALEMD,CLARKA,MORENOA. Stochasticnetworkmodelsforlogisticsplanningindisasterrelief[J]. European Journal of Operational Research, 2016, 255(1): 187-206.
[33] ZHANG J, LIU H, YU G, et al. A three-stage and multi-objective stochastic programmingmodeltoimprovethesustainablerescueabilitybyconsideringsecondarydisastersinemergencylogistics[J]. Computers & Industrial Engineering, 2019, 135: 1145-1154.
[34] YANG L, JONES B F, YANG S H. A fuzzy multi-objective programming for optimization offire station locations through genetic algorithms[J]. European Journal of Operational Research,2007, 181(2): 903-915.
[35] WEI X, QIU H, WANG D, et al. An integrated location-routing problem with post-disasterrelief distribution[J]. Computers & Industrial Engineering, 2020, 147: 106632.
[36] ROLLAND E, PATTERSON R A, WARD K, et al. Decision support for disaster manage-ment[J]. Operations Management Research, 2010, 3(1-2): 68-79.
[37] ZHANG S, GUO H, ZHU K, et al. Multistage assignment optimization for emergency rescueteams in the disaster chain[J]. Knowledge-based systems, 2017, 137: 123-137.
[38] ARAYA-CÓRDOVA P, VÁSQUEZ Ó C. The disaster emergency unit scheduling problem tocontrol wildfires[J]. International Journal of Production Economics, 2018, 200: 311-317.
[39] REZAPOUR S, NADERI N, MORSHEDLOU N, et al. Optimal deployment of emergencyresources in sudden onset disasters[J]. International Journal of Production Economics, 2018,204: 365-382.
[40] RAUCHECKER G, SCHRYEN G. An exact branch-and-price algorithm for scheduling rescueunits during disaster response[J]. European Journal of Operational Research, 2019, 272(1):352-363.
[41] SUN H, WANG Y, XUE Y. A bi-objective robust optimization model for disaster responseplanning under uncertainties[J]. Computers & Industrial Engineering, 2021, 155(3): 107213.
[42] ALUTAIBI K, ALSUBAIE A, MARTI J. Allocation and scheduling of firefighting units inlarge petrochemical complexes[C]//International Conference on Critical Infrastructure Protec-tion. [S.l.: s.n.], 2015.
[43] RODRIGUEZ S A, RODRIGO A, AGUAYO M M. A simulation-optimization approach forthe facility location and vehicle assignment problem for firefighters using a loosely coupledspatio-temporal arrival process[J]. Computers & Industrial Engineering, 2021, 157: 107242.
[44] GOSAVI A. Reinforcement learning: A tutorial survey and recent advances[J]. INFORMSJournal on Computing, 2009, 21(2): 178-192.
[45] KHADILKAR H. A scalable reinforcement learning algorithm for scheduling railway lines[J].IEEE transactions on intelligent transportation systems, 2018.
[46] JIANG Z, GU J, FAN W, et al. Q-learning approach to coordinated optimization of passengerinflow control with train skip-stopping on a urban rail transit line[J]. Computers & IndustrialEngineering, 2018, 127(JAN.): 1131-1142.
[47] JIANG Z, FAN W, LIU W, et al. Reinforcement learning approach for coordinated passengerinflow control of urban rail transit in peak hours[J]. Transportation Research Part C EmergingTechnologies, 2018, 88(MAR.): 1-16.
[48] ŠEMROV D, MARSETIČ R, ŽURA M, et al. Reinforcement learning approach for trainrescheduling on a single-track railway[J]. Transportation Research Part B: Methodological,2016, 86: 250-267.
[49] KWONIH,CHANGOK,JINJ,etal. Case-basedmyopicreinforcementlearningforsatisfyingtarget service level in supply chain[J]. Expert Systems with Applications, 2008, 35(1-2): 389-397.
[50] JIANG C, SHENG Z. Case-based reinforcement learning for dynamic inventory control in amulti-agent supply-chain system[J]. Expert Systems with Applications, 2009, 36(3p2): 6520-6526.
[51] KARA A, DOGAN I. Reinforcement learning approaches for specifying ordering policies ofperishable inventory systems[J]. Expert Systems with Applications, 2018, 91: 150-158.
[52] YU L, ZHANG C, JIANG J, et al. Reinforcement learning approach for resource allocation inhumanitarian logistics[J]. Expert Systems with Applications, 2021, 173: 114663.
[53] KLAINE P V, NADAS J P, SOUZA R D, et al. Distributed drone base station positioning foremergency cellular networks using reinforcement learning[J]. Cognitive Computation, 2018,10(5): 790-804.
[54] SU Z P, JIANG J G, LIANG C Y, et al. Path selection in disaster response management basedon q-learning[J]. International Journal of Automation and Computing, 2011, 8(1): 100-106.
[55] WALUNJKAR G, RAO A K. A reinforcement learning approach for evaluation of disasterrelief[C]//2019 International Conference on Cutting-edge Technologies in Engineering (ICon-CuTE). [S.l.: s.n.], 2020.
[56] NADI A, EDRISSI A. A reinforcement learning approach for evaluation of real-time disasterrelief demand and network condition[J]. World Academy of Science, Engineering and Technol-ogy, International Journal of Environmental, Chemical, Ecological, Geological and Geophysi-cal Engineering, 2016, 11(1): 5-10.
[57] NADI A, EDRISI A. Adaptive multi-agent relief assessment and emergency response[J]. In-ternational Journal of Disaster Risk Reduction, 2017, 24: 12-23.
[58] DIJKSTRA E W, et al. A note on two problems in connexion with graphs[J]. NumerischeMathematik, 1959, 1(1): 269-271.
[59] HART P, NILSSON N, RAPHAEL B. A formal basis for the heuristic determination of mini-mum cost paths[J]. IEEE Trans. Syst. Sci. Cybern., 1968, 4: 100-107.
[60] IKEDA T, HSU M Y, IMAI H, et al. A fast algorithm for finding better routes by ai search tech-niques[C]//ProceedingsofVNIS’94-1994VehicleNavigationandInformationSystemsConfer-ence. [S.l.]: IEEE, 1994: 291-296.
[61] CHABINI I, LAN S. Adaptations of the a* algorithm for the computation of fastest pathsin deterministic discrete-time dynamic networks[J]. Intelligent Transportation Systems IEEETransactions on, 2002, 3(1): 60-74.
[62] NANNICINI G, DELLING D, SCHULTES D, et al. Bidirectional a* search on time-dependentroad networks[J]. Netw., 2012, 59(2): 240–251.
[63] DELLING D, NANNICINI G. Core routing on dynamic time-dependent road networks[J].INFORMS journal on computing, 2012, 24(2): p.187-201.
[64] MAGNANTI T, AHUJA R, ORLIN J. Network flows: theory, algorithms, and applications[J].PrenticeHall, Upper Saddle River, NJ, 1993.
[65] CUNNINGTON D. A cognitive assistant for route selection using knowledge heuristics[C]//ACM Computer Science in Cars Symposium (2018). [S.l.: s.n.], 2018.
[66] Department of housing and urban construction. Standards for local flooding prevention andcontrol system planning[EB/OL]. 2014. https://www.docin.com/p-1009004947-f6.html.
[67] COXR,SHANDT,BLACKAM. Australianrainfallandrunoffrevisionproject10: appropriatesafety criteria for people[J]. Water Res, 2010.
[68] FISHBURN, PETER C. Exceptional paper–lexicographic orders, utilities and decision rules:A survey[J]. Management Science, 1974, 20(11): 1442-1471.
[69] MIETTINEN K. Nonlinear multiobjective optimization: volume 12[M]. [S.l.]: Springer Sci-ence & Business Media, 1999.
[70] Xinhuanet. 11 people dead in shenzhen floods[EB/OL]. 2019. http://en.people.cn/n3/2019/0414/c90000-9566570.html.
[71] ZHANG Z, MENG F, ZENG Y, et al. Identification of urban rainstorm waterlogging basedon multi-source information fusion: A case study in futian district, shenzhen[C]//E3S Web ofConferences: volume 259. [S.l.]: EDP Sciences, 2021: 01004.
[72] CHENJ,YANGST,LIHW,etal. Researchongeographicalenvironmentunitdivisionbasedonthe method of natural breaks (jenks)[J]. ISPRS - International Archives of the Photogrammetry,Remote Sensing and Spatial Information Sciences, 2013, XL-4/W3(4): 47-50.
[73] Fire brigade command center. Dispatching standard of the fire receive-disposal alarm[EB/OL].2018. https://wenku.baidu.com/view/b4a04598a66e58fafab069dc5022aaea998f418f.html.
[74] TZENG G H, CHENG H J, HUANG T D. Multi-objective optimal planning for designingreliefdeliverysystems[J]. TransportationResearchPartE:LogisticsandTransportationReview,2007, 43(6): 673-686.
[75] VITEVITCH M, ERCAL G, ADAGARLA B. Simulating retrieval from a highly clusterednetwork: Implications for spoken word recognition[J]. Frontiers in psychology, 2011, 2: 369.
[76] SHEU J B. An emergency logistics distribution approach for quick response to urgent reliefdemand in disasters[J]. Transportation Research Part E Logistics & Transportation Review,2007, 43(6): 687-709.
[77] HAIMES Y. On a bicriterion formulation of the problems of integrated system identificationand system optimization[J/OL]. IEEE Transactions on Systems, Man, and Cybernetics, 1971,SMC-1(3): 296-297. DOI: 10.1109/TSMC.1971.4308298.
[78] HOLGUÍN-VERAS J, PÉREZ N, JALLER M, et al. On the appropriate objective function forpost-disaster humanitarian logistics models[J]. Journal of Operations Management, 2013, 31(5): 262-280.
[79] METE H O, ZABINSKY Z B. Stochastic optimization of medical supply location and distribu-tion in disaster management[J]. International Journal of Production Economics, 2010, 126(1):76-84.
[80] PETTET G, MUKHOPADHYAY A, KOCHENDERFER M J, et al. Hierarchical planning forresource allocation in emergency response systems[C]//Proceedings of the ACM/IEEE 12thInternational Conference on Cyber-Physical Systems. [S.l.: s.n.], 2021: 155-166.
[81] MENG F, GONG W, LIANG J, et al. Impact of different control policies for covid-19 outbreakon the air transportation industry: A comparison between china, the u.s. and singapore[J]. PLoSONE, 2021, 16(3): e0248361.
[82] GONG X, LIANG J, ZENG Y, et al. A hierarchical multi-objective programming approach toplanning locations for macro and micro fire stations[C]//ICFIE 2021 - 8th International Confer-ence on Frontiers of Industrial Engineering. [S.l.: s.n.], 2021.