Study on Co-pyrolysis Products Based on Multivariate Regression Analysis and Neural Network Models

The study investigates the normality of INS and the yields of tar, water, and coke. The Wilcoxon signed-rank test -was employed, yielding a significance value of 0.828, indicating that INS has no significant impact on tar yield. Subsequently, paired sample t-tests were conducted, resulting in significance values of 0.245 and 0.000 for water yield and coke yield, respectively. These tests revealed that INS has no significant impact on water yield, but does significantly affect coke yield. Then, mathematical models were established. Multivariate regression analyses were performed. The analysis revealed that the significance level of Z was remarkably high. It was further discovered that the presence of an interaction effect between nonane insoluble (INS) and the mixing ratio. Based on the fitted linear regression equation, the path parameters indicated that the variable Z has the greatest impact on tar yield. Finally, a comprehensive product index was established based on weights. Subsequently, a random forest regression model was developed. The results showed that the combination of small algae/Huainan coal with a ratio of 40/100 can achieve a better product utilization rate and energy conversion efficiency. The predicted results for tar yield and INS solubles yield were 0.3901 and 0.2341, respectively.