Intelligent joint cutting strategy and method for roadheader combining LSTM deep learning and fuzzy inference control

Excavating coal mine roadways is the most hazardous and challenging aspect of underground production. While intelligent fully mechanized mining faces have advanced, the intelligentization of roadway excavation has been slow, resulting in a “mining-excavation imbalance” that hinders efficient and intelligent mining in coal enterprises. In China, roadheaders are extensively used electro-mechanical equipment for underground excavation. The efficiency and quality of the roadway excavation directly depend on the roadheader’s ability to cut coal and rock quickly and accurately. This paper proposes an intelligent joint cutting strategy for roadheaders, utilizing LSTM deep learning and fuzzy inference control to enhance efficiency and intelligence. The study includes a comprehensive analysis of joint cutting conditions, leading to the development of a joint control strategy. Additionally, a joint cutting control method is suggested, integrating LSTM deep learning neural network controller for accurate load identification and a fuzzy inference controller for intelligent speed regulation of the roadheader’s cutting head and arm. Simulation results demonstrate that the proposed method achieves intelligent joint regulation under both conventional and complex working conditions with a control process response time within 0.6 seconds, minimal overshoot, high control accuracy, and stability. Compared to advanced single control methods, the proposed joint cutting control method reduces response time and ensures stability. Designed experiments on a remotely monitored and controlled platform for roadheaders verify the accuracy and effectiveness of the method, providing a technical reference for the rapid and intelligent excavation of roadheader robots and laying a theoretical foundation for further optimization and engineering applications.