Dense phase pulverized coal concentration is a key performance indicator for the operation of pulverized coal boiler. In this paper, terahertz time-domain spectroscopy is applied to the quantitative analysis of pulverized coal concentration. To improve the stability and accuracy of terahertz spectrum quantitative analysis of pulverized coal mixture with complex chemical components, genetic algorithm (GA) and partial least squares regression (PLS-R) are introduced into terahertz time-domain spectrum quantitative analysis of dense phase pulverized coal-high density polyethylene mixture. The optimal set of spectral variables is constructed by GA, and the quantitative analysis model of pulverized coal concentration is established by partial least squares. Experimental results show that the correlation coefficient and root mean square error of sample prediction set obtained by GA and PLS-R are 0.9568 and 1.0345, respectively. Compared with the quantitative analysis model established by traditional interval partial least squares methods (iPLS-R and biPLS-R), the model has higher accuracy and stability, which provides the basis for the application of terahertz time-domain spectrum in the quantitative analysis of dense phase pulverized coal concentration.