Study On The Optimization Models And Algorithms For Refrigerated Food Transportation

With the continuous development of economic and social development in recent years,the food cold chain market has been expanding rapidly.As the critical process of the food cold chain,the refrigerated food transportation plays a direct and important role in affecting the freshness and even the safety of food products.Confronted with the fact that the optimization theory for refrigerated food transportation is imperfect,it is necessary to scrutinize the optimization techniques of point-to-point transportation,trans-shipment,and joint distribution of perishable food products to decrease loss ratio,reduce transportation cost,and improve customer satisfaction.This dissertation uses knowledge and techniques in operations research,logistics,and computer sciences and combining qualitative analysis and quantitative calculation to research the computational intelligence-based optimization theory for refrigerated food transportation.Therefore,the major focuses of research in this dissertation include the following aspects:(1)Research the load planning problem for long-haul point-to-point transportation.The profit is defined as the fuzzy number and defuzzified by using the graded k-preference integration method over the uncertainty fluctuation of food prices.On account of the constraints of food compatibility with regard to odor contamination,vehicle capacity,temperature demand and humidity demand for storage of various perishable food products,the mathematical model of the load planning problem for long-haul point-to-point transportation is constructed to maximize the total profit of loaded food products.According to the characteristics of the model,a discrete firefly algorithm is designed to acquire the near-optimal solutions to the load planning problem.This algorithm uses the similarity degree of load plans to measure the distance between any two fireflies.Furthermore,how the fireflies moves is redefined.(2)Research the load planning problem for short-haul point-to-point transportation.The profit and the shelf life are defined as fuzzy numbers and defuzzified by using the graded mean integration.To maximize the total profit of loaded food products,minimize the sum of shelf life values of loaded food products,and maximize the sum of matching values between the optimal temperature range of the loaded food products and the compartment temperature of refrigerated vehicles based on a similarity method,the mathematical model of the load planning problem for short-haul point-to-point transportation is constructed.Given that this model has three objective functions,this paper proposes the PDFA which integrates the discrete firefly algorithm and the PAES to acquire the Pareto set of the load plan.The framework of the discrete firefly algorithm in this hybrid algorithm is similar to that for solving the load planning problem for long-haul door-to-door transportation.The two versions of discrete firefly algorithm differ in population initialization strategy and movement scheme according to the characteristics of the perishable food short-haul nonstop transportation.Furthermore,as an integral part of the PAES for optimizing the surrounding neighborhood,a three-dimensional adaptive grid algorithm is designed to optimize the degree of crowding,which contributes to solving the problem of performance degradation caused by the overcrowding of population.In addition,algorithm performance is tested through a series of problem instances.(3)Research the multi-commodity minimum cost flow-based optimization for perishable food trans-shipment.To cope with the uncertainty of information in the period of decision making,cost,demand,supply,deterioration rate,capacity are defined as fuzzy numbers and defuzzifed by using the graded mean integration method.In this.model,many characteristics of perishable food transportation are taken into consideration,including the perishability of food and the imbalance between demand and supply.Two strategies that are excess supply and off season sales are considered in the problem.The mathematical model is constructed aiming at minimizing costs and damageable ratio according to the trade-off relation of cold chain.Given that the model of multi-objective multi-commodity flow is quite complex because of the high correlation among the flows of direct edges.This characteristic also makes it difficult for operations such as crossover and combination among flows to be implemented.To cope with this problem,this dissertation designs a harmony search-based algorithm to generate the optimal routes.In this algorithm,the flow between the supply node and the transshipment node and the flow between the transshipment node and the demand node is optimized by using the harmony search algorithm.Flow between transshipment nodes is optimized randomly.Through this method,performance degradation caused by repetitive adjustment can be avoided.(4)Research the customer satisfaction-oriented clustered cold accumulation multi-temperature joint distribution optimization problem.The cold accumulation multi-temperature joint distribution is a service which stores perishable food products with different temperatures in insulated boxes or cabins and transported them by the general type vehicle.To emphasize the timeliness of the delivery,the mathematical model of clustered cold accumulation multi-temperature joint distribution which aims at maximizing the customer satisfaction is constructed.In this model,a Z-shaped curve is used to depict the customer satisfaction degree and define the travel times as fuzzy numbers.Additionally,the capacity of each vehicle can be different and customers in multiple regions can be served by the same vehicle in this model.A time series-based discrete firefly algorithm is designed to optimize the distribution route and load plan simultaneously.By considering the feature of the clustered cold accumulation multi-temperature joint distribution optimization problem,this algorithm defines the coding scheme of the firefly,the measurement method between fireflies,the movement scheme of fireflies,and the initialization strategy based on the random searching and the greedy algorithm respectively.