中文摘要：

借助多元统计方法识别2000年1月-2016年12月江西省生猪价格波动的成因。选择生猪价格波动率作为警情指标，借助时差相关法，确定风险预警指标，建立BP人工神经网络模型，预警生猪价格波动的风险。结果表明，玉米价格、仔猪价格、猪肉价格、活鸡价格、豆粕价格、生产者预期和疫情为主要警情指标，其中先行性指标是仔猪价格和生产者预期，同步指标是玉米价格、猪肉价格和活鸡价格，豆粕价格和疫情是滞后指标；除个别样本点外，BP神经网络模型输出的生猪价格预警值和实际价格数据比较接近，生猪价格波动风险预警模型具有良好的预警效果。根据不同级别的风险程度提出相应对策建议，以便政府能够根据不同警情采取具有针对性的措施，使生猪生产、市场平稳健康发展。

英文摘要：

By multiple statistics method to recognize the cause for Jiangxi hog price fluctuation from January 2000 to December 2016, this paper selects hog price fluctuation rate as warning index, determine risk early warning index by time-difference correlation analysis, and establishes BP neural network to early warn the risk of hog price fluctuation. The results reveal that： corn price, piglet price, pork price, chicken price, soybean meal price, producers expectation and epidemic are the major alert indicators, in which the piglet price and producers expectation are leading indicators of hog price fluctuation. Corn price, pork price and chicken price are the coincident indicators. Soybean meal price and epidemic are the lagging indicators. Except individual sample point, the hog price alert value output by BP neural network model are more close to the actual price data, and the alert model has excellent alert effect. Finally, this paper puts forward relevant countermeasure according to different levels for government to take aiming measures against differnt warnings to stabilize the healthy development of hog market.