|
CSME 2023/10
Volume 44 No.5
:
463-472
|
|
Prediction Method of Ball Mill State Parameters Based on FWA-LSSVM Model
Xiao-Yan Luoa, Hua-Zhi Xua, Wen-Cong Tanga and Wen-Hai Lua
aDepartment of Mechanical Engineering, Jiangxi University of Science and Technology, Ganzhou, Jiangxi 341000, China.
|
Abstract:
Aiming at the problem that it is difficult to ac-curately judge the working state parameters during the grinding process of the ball mill, a method for predict-ing the working state parameters of the ball mill based on fireworks algorithm optimized LSSVM (Least Squares Support Vector Machine) is proposed. Firstly, the LSSVM algorithm is employed to establish the predictive model for the working state parameters of the ball mill. Then, the FWA (Firework Algorithm) al-gorithm is used to optimize the radial basis kernel function parameters and penalty factors of the LSSVM model. Afterwards, time-domain features, frequency-domain features, and entropy features are extracted from the vibration signals of the mill shell to generate a set of feature vectors; finally, feature vectors are used as the input of FWA-LSSVM, and the ratio of material to ball, rotation speed and filling rate are used as the output to establish a mill state parameter prediction model. The superiority of the method is proved by grinding experiments. The results showed that the LSSVM model optimized with the FWA algorithm had less error between the predicted and actual values of filling speed, Material-ball ratio and rotational speed than the GA (Genetic Algorithm) and PSO (Particle Swarm Optimization) optimization algorithms, indi-cating that the mill state parameter prediction model has higher precision and stability.
|
|
Keywords: mill vibration, firework algorithm, least squares support vector machine, state parameter prediction.
|
Download PDF
|
|
*Corresponding author; e-mail:
|
©
2023
CSME , ISSN 0257-9731
|