基于一种新型联想记忆神经网络的传感器故障诊断

联想记忆网络是一种反馈型神经网络。由于反馈型网络会收敛于某个稳定状态，因此，可用于联想记忆。神经网络具有高度的并行处理能力和极强的非线性映射能力，可以实现故障与征兆之间复杂的非线性映射关系，因此在机械故障诊断领域中显示了很大的应用潜力。本文以模拟人脑由部分记忆而联想整体的特点为基础，通过引入联想记忆衰减因子，改进神经网络结构和学习算法．应用于系统的故障诊断。

The Sensor Fault Diagnosis Based on a New Associative Memory Neural Network

Associative memory network is a feedback type neural network.Due to the feedback type network will be converged to a certain stable state, therefore, it could be used for the associative memory. The neural network has a high ability of parallel process and very strong ability of nonlinear image, it can realize the complex non-linear reflective relations between the fault and symptom, so it reveals great applicable potentialities in mechanical fault diagnostic area. In this paper, based on the characteristic to simulate human brain by means of the part of memory to associate with the whole, through introducing associative memory diminished factor to improve the neural network structure and learning algorithm and also can be used in the fault diagnosis in this system.