Pulverizing loop ball mill load control method for cement factory

Abstract

The load control method for ball mill in powdering loop cement plant adopts a double-layer control structure, including an upper layer of self optimizing algorithm and a lower layer of multidimensional fuzzy control algorithm. The self optimizing algorithm finds out the optimal inside material storage quantity inside the ball mill as the controlling set value; and the multidimensional fuzzy control algorithm controls the material compounding links to realize the automatic regulation of the material feed quantity, so as to reach the required components in the cement production, lower the raw material amount and the cement screen residue, raise grinding fineness, ensure cement quality, increase output of ball mill and lower production cost.

Description

Technical field [0001] The invention relates to a control method, in particular to a load control method of a grinding loop ball mill in a cement plant. Background technique [0002] In the modern cement industry, the production and processing process of raw (clinker) is of great significance to the entire cement production. The quality of cement products depends on the quality of raw (clinker) products, and the production efficiency of raw (clinker) products also has a very important impact on the benefit of the entire cement production. In the raw (clinker) production workshop of a cement plant, the mixing of raw materials and the grinding of a ball mill are the two main production links. The accuracy of the mixing and batching process of raw materials determines the quality of raw (clinker) products. The grinding process of the ball mill needs to consume a lot of electric energy. Whether it can operate normally under a stable and efficient working condi...

AI Technical Summary

Problems solved by technology

However, since one of the main characteristics of the grinding circuit is the large lag, the PID algorithm cannot fundamentally solve the problem of the large delay, so this method also lacks a certain theoretical basis

Some use predictive control, but this type of control needs to know the model of the controlled object, and the grinding circuit is too complicated, and its accurate mathematical model is not easy to obtain, so this type of algorithm is not easy to apply

The neural network does not need to know the model of the control object, but because it needs a training set and the training process is cumbersome, and the grinding circuit has certain requirem

Images