基于神经网络的自动化生产监控系统

为了使自动化生产系统始终处于高效的工作状态,在工作过程中就必须对其进行实时监控,以及时地发现系统中的故障情况和异常现象并进行诊断和处理。基于多层感知器和线性联想记忆的神经网路提出了一种新的自动化生产系统的故障发现及诊断的理论方法。运用这种理论方法,我们可以很方便地对自动化生产系统中的多特征复杂故障、多故障及多过程故障模式进行监测和诊断。同时,这种方法解决了以往所用方法算法复杂和难于实现的困难,且对于多过程的自动化生产系统,这种方法不存在优先权的选择问题,克服了以往所采用的方法不能同时监控多个过程的局限性,增强了系统监控的实时性。经仿真研究表明:这种方法较为理想和实用。

AUTOMATIC PRODUCTION MONITORING SYSTEM BASED ON NEURAL NETWORK

In this paper, a new theoretical method for fault detection and diagnosis of automatic productionsystem is proposed, based upon multi-layer perception and linear associative memory neural network.By using this method, multiple feature complex fault, multiple pattern fault, and multiple process faultcan be detected and diagnosed conveniently. Moreover, this method is simple in algorithm and easy torealize in comparison with old methods. For multiple process automatic production system, there is noneed to select priority, thus the real time performance of process monitoring is enhanced. Simulationstudies show that this method is pragmatic and effective.