关键词: 双层多目标模型, 应急医疗设施选址, 伤员转运, 心理煎熬成本, 新型人类学习算法

Abstract: Earthquake disasters were frequent in recent years, causing significant economic losses and casualties. The outbreak of massive earthquakes has brought about a large number of casualties, resulting in a high demand for emergency relief in a short period of time. Therefore, in the face of sudden earthquake disaster, how to carry out timely emergency rescue in a short period of time, rational arrangement of emergency medical facilities, and efficient and orderly organization of casualty transfer is particularly important. Existing studies are mostly from the perspective of vehicle scheduling to the total mileage and time of casualty transfer, the survival of casualties, and the total cost of the program as the objective function of the problem modeling and solving, but less consider the psychological state of the casualty’s impact on the effectiveness of the rescue; at the same time, the current research focuses on the location of separate medical facilities or casualty transfer problems, but rarely on the integration of the two problems research.
Based on this, this paper comprehensively considers the deterioration of injuries, convoys and helicopters involved in casualty transfer, road damage and other factors, to construct a two-layer multi-objective optimization model. The upper-level objective is to minimize the total time for transporting medical supplies and transferring casualties, while the lower-level objective is to maximize the cumulative survival rate of casualties and minimize the cost of psychological suffering of casualties. By adopting the trauma index scoring method, the model classifies the injured waiting for rescue at each rescue point after the earthquake into two categories, light and heavy, and assigns the corresponding rescue priority levels. The injury deterioration function and the casualty survival function are also proposed to portray the survival rate of casualties who are treated at different time. We consider that in the process of casualty transfer, while waiting for the vehicle to rescue, the injured have anxiety, despair and other negative feelings, which will negatively affect the earthquake rescue and deteriorate their physical conditions. This paper portrays the psychological costs of casualties through the casualty psychological ordeal cost function.
Given that the model is an NP-hard problem, an improved human learning algorithm is proposed based on the basic human learning algorithm. The algorithm extends the coding method from binary coding to integer coding by introducing the Split algorithm with the crossover operator. Meanwhile, in combination with the theory of contrastive cognition, the hash table-based solution memorization mechanism and community learning operator are introduced to improve the diversity of the solution set by avoiding repeated searches of the solution and increasing the learning approach; the contrastive selection strategy is introduced in order to improve the convergence of the solution set. In this paper, the Solomon dataset and related studies are used to generate small, medium and large scale of examples and select multi-objective genetic algorithms and simple human learning algorithms to compare with the algorithms in this paper. The results show that the algorithm in this paper can effectively solve the small, medium and large scale of cases and outperforms the remaining two comparative algorithms. Compared to the remaining two algorithms, the average cumulative survival rate of the casualty is large, and the average total transit time and the cost of psychological suffering of the casualty are small on average. Finally, taking the Jishishan earthquake as an example, an optimization scheme for the siting of emergency medical facilities and the transfer of the injured is proposed.
The effect of aftershocks and uncertainty in the number of casualties on the casualty transfer scheme can be considered subsequently to establish a multi-stage casualty number uncertainty model for emergency medical facility siting-casualty transfer.

Key words: bi-level multi-objective modeling, siting of emergency medical facilities, casualty transport, cost of psychological suffering, novel human learning algorithms