中文摘要：

在纤维、造纸、塑料及金属镀膜工业中，对线状、带状及面状弹性材料卷取张力的控制是关系其产品质量的关键技术。介绍了一种动态贝叶斯网络训练算法的多智能体技术，该方法有效地减弱了张力控制系统中各轴速度、张力之间的耦合作用。仿真结果表明，该算法具有良好的自学习和自适应性能。

英文摘要：

In the field of fiber, paper making, plastic and metal plating, one of the key techniques to guarantee the product quality is the control of winding tension of linear, strip and area materials. A kind of multi-agent technology with dynamic Bayesian networks training algorithm was introduced. This method effectively weakened the coupling action between the velocity and tension of the tension control system. Simulation results indicated the favorable self-learning and adaptive effect of the control algorithm.