KPIs abnormal root cause positioning method based on LSTM-CNN causal discovery model

• Main Authors: WANG PEIPENG, ZHAO YUN, ZHANG XIUGUO, WANG TENGLONG, CAO ZHIYING, SHANG ZIJING
• Format: Patent
• Language: Chinese; English
• Published: 23.06.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides a KPIs abnormal root cause positioning method based on an LSTM-CNN causal discovery model, and the method comprises the steps: carrying out the processing of abnormal KPIs data based on an improved Pearson algorithm, and obtaining an abnormal KPIs correlation matrix; analyzing the causal relationship between the abnormal KPIs based on an LSTM-CNN causal discovery model to obtain an abnormal KPIs causal relationship matrix and construct a causal relationship graph between the abnormal KPIs; calculating the importance value of each abnormal KPIs according to the processed abnormal KPIs causal relationship adjacency matrix and the abnormal KPIs correlation matrix; the obtained importance degree values of all the abnormal KPIs are arranged in a descending order, and Top-K abnormal KPIs are selected as root cause KPIs indexes causing system faults and output. According to the invention, the accuracy of KPIs abnormal root cause positioning is improved. 本发明提供一种基于LSTM-CNN因果发现模型的KPIs异常根因定位方法，包括：