The invention discloses an online self-adaptive working
condition monitoring and fault diagnosis method for a process industrial course and belongs to the technical field of fault monitoring and diagnosis of complex industrial courses. The method comprises the following steps of firstly, analyzing historical
observation data under a normal working condition, introducing an elastic regression network combining Lasso constraints with
Ridge constraints to establish an industrial course fault monitoring model on the basis of sparse
principal component analysis, and then obtaining a course controllimit to industrial course fault monitoring statistics; during online monitoring of industrial course faults, adopting an order-1 matrix correcting
algorithm for resolving a
covariance matrix of the online
monitoring data, conducting
recursion updating on a load matrix of the sparse monitoring model to obtain the course control limit to the course fault monitoring statistics matched with the working condition, and achieving self-adaptive fault detection in the
process industry course; finally, according to the detected faults, adopting a contribution plot method for obtaining specific causes of the faults. By means of the method, the faults of the
process industry course with complex and changeable working conditions can be self-adaptively monitored for a long time; the method has the advantages of low calculation complexity, high precision, a low report missing rate and the like.