Fault diagnosis model training method and device

• Main Authors: WANG JINJIANG, LIU SHUANZHONG, LI WENJIN, CHEN CHONG, ZHANG LAIBIN, JIANG AIGUO, QIN JIAN'AN, GU MING, YU HAOTIAN, XIAN MINYUAN
• Format: Patent
• Language: Chinese; English
• Published: 28.07.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a fault diagnosis model training method. The method comprises the steps of acquiring state characteristic parameters, collected by sensors in multiple top drive systems, of thetop drive systems, wherein the position of the sensor in the top drive system is determined according to a random weight genetic algorithm; respectively eliminating variable working condition characteristic parameters in each acquired state characteristic parameter by utilizing a redundant attribute projection algorithm; respectively fusing the plurality of state characteristic parameters withoutthe variable working condition characteristic parameters to obtain a plurality of fused state characteristic parameters; respectively marking fault types for the plurality of fused state characteristic parameters to obtain training data; and training a pre-established fault diagnosis model by adopting the training data. Through the scheme of the invention, the accuracy of fault diagnosis is improved. 一种故障诊断模型训练方法，包括：获取多条顶