Construction method of boundary forest model, multi-working-condition soft computing model updating method for complex industrial process and application thereof

Abstract

The invention discloses a construction method of a boundary forest model, a multi-working-condition soft computing model updating method for a complex industrial process and an application thereof, belonging to the field of computer application. In order to solve the problem that prediction values are unreliable due to the fact that leaf nodes of a tree integration model easily generate blank areas in an output range, when a current training set under a certain working condition is known, the construction method comprises the steps: setting different leaf node minimum sample numbers, and establishing K tree integration models with different leaf node boundaries by using different leaf node minimum samples; predicting output values of all samples in the current training set by using a treeintegration model, and forming a prediction matrix by the predicted output values; according to the prediction output value of the prediction matrix, constructing a correlation matrix of the prediction output value and a real output value; and calculating a fusion weight vector, using the weight vector, and fusing the tree integration models with different boundaries into a boundary forest model,so that the leaf nodes of different tree models cover each other, and the blank area of a single tree on an output boundary is filled, and a reliable prediction value is generated.

Description

technical field [0001] The invention belongs to the field of computer application and relates to integrated learning, online learning algorithm and key variable online soft sensor method based on boundary forest. Background technique [0002] Soft computing (Soft Computing) model is one of the most effective tools to perform the task of predicting key variables (for example, the end temperature of molten steel in the electric arc furnace refining process). Also known as soft sensor model or soft sensor, it is a virtual sensor rather than a hardware instrument, which has the advantages of easy implementation and economical feasibility [1,2] . In essence, soft computing models belong to data-driven models, which dig deep into the nonlinear relationship between input variables and key variables, and enable data to provide its potential but useful information through function mapping. However, in practical applications such as complex industrial processes, there is often a str...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem that the leaf nodes of the tree integration model are easy to produce blank areas on the output range, resulting in unreliable prediction values, the invention proposes a method for building a boundary forest model; in order to solve the problem of data time validity, the invention also proposes a method for The multi-working-condition soft computing model update method for complex industrial processes can improve the reliability of online prediction of key variables in complex industrial processes, and realizes the establishment of accurate and reliable data for massive, nonlinear, high-noise and time-effective data. soft sensor model

Images