【摘要】 生鲜农产品等冷链产品市场需求快速增长，冷链物流的供给无法满足人们的需求给生鲜农产品带来新的挑战.安徽省作为一个农产品丰富的地区，生鲜农产品的供应对于满足市场需求至关重要.收集了2001～2022年生鲜农产品产量数据，采用反向传播神经网络(Back Propagation Neural Network, BP神经网络)、长短时记忆(long short-term memory, LSTM)、粒子群算法优化的长短期记忆神经网络(Particle Swarm Optimization-Long Short-Term Memory, PSO-LSTM)三种模型进行训练和验证，通过三种模型的对比分析，三种模型相对误差分别为0.13%、0.06%、0.02%.结果表明，PSO-LSTM模型预测精度最高，拟合效果最好，能够有效预测未来四年安徽省生鲜农产品冷链物流需求，以应对不断增长的冷链物流需求压力.更多还原

【Abstract】 The demand for fresh agricultural products and other cold chain products in the market is rapidly increasing, but the supply of cold chain logistics cannot meet people’s needs, bring new challenges to fresh agricultural products. As a region rich in agricultural products, the supply of fresh agricultural products in Anhui Province is crucial for meeting market demand. Therefore, this paper collected fresh agricultural product yield data from 2001 to 2022 and trained and validated three models: back propagation neural network(BP neural network), long short term memory(LSTM), and particle swarm optimization long short term memory(PSO-LSTM). Through comparative analysis of the three models, the relative errors of the three models were 0.13%, 0.06%, and 0.02%, respectively. The results showed that the PSO-LSTM model has the highest prediction accuracy and the best fitting effect, and can effectively predict the cold chain logistics demand for fresh agricultural products in Anhui Province in the next four years to cope with the growing pressure of cold chain logistics demand.更多还原