Multi-step prediction method for clean coal quality in dense medium coal separation process based on LSTM (Long Short Term Memory)

Abstract

The invention provides an LSTM (Long Short Term Memory)-based clean coal quality multi-step prediction method in a dense medium coal preparation process, which comprises the following steps of: firstly, obtaining useful data, and reconstructing the useful data into a plurality of production group data; coal flow separation transportation dead time T is calculated; model construction: constructing an LSTM neural network; model training: adjusting a network structure of the LSTM neural network through values and trends of training set data loss and test data set loss; storing the adjusted model; and new data is obtained and substituted into the stored model for output prediction. According to the method, the space-time relationship between the coal dressing process flow data and the to-be-predicted output variable is fully considered, so that a subsequent machine learns a prediction model result, and the method can be better suitable for an industrial process of dense medium coal dressing with a complex space-time sequence relationship. Good prediction precision is obtained, and a powerful basis is provided for precise control and product quality stability in the subsequent coal dressing production process.

Description

technical field [0001] The invention relates to the field of quality prediction, in particular to an LSTM-based multi-step prediction method for clean coal quality in a dense medium coal preparation process. Background technique [0002] According to statistics, my country's 2020 CO 2 Emissions are about 10.3 billion tons, 92% of which are CO 2 Emissions come from coal, oil and natural gas (approximately 9.5 billion tonnes). The characteristics of my country's energy structure of "rich in coal, short of oil, and low in gas" determine that coal will remain an important strategic energy source and material for my country for a long time to come. important part of. As the source and basis of clean coal utilization, coal preparation is an important means to speed up the adjustment of energy structure and increase the supply of clean energy. One of the indicators to evaluate the separation process is the stability of clean coal ash. Therefore, the online prediction and judgmen...

AI Technical Summary

Problems solved by technology

The application number is: CN201910900555.4 invention patent: an intelligent control system and method for dense medium separation Density setting relies heavily on experience, but this patent does not involve the use of existing big data to deal with the field of clean coal ash prediction

At the same time, in the actual production process of dense medium coal preparation, the measurement accuracy of the clean coal online ash analyzer is restricted and affected by various influencing factors, which makes the measurement results fluctuate greatly, and it is difficult to reflect the actual clean coal quality.

Images