Abstract: The original DCS alarm system of the grinding system in a cement production line suffers from imperfect architecture, multiple monitoring blind spots. Operators have long relied on manual experience to judge equipment operating conditions, resulting in delayed fault responses. Accordingly, a voice alarm system was developed and applied to this production line. The system was implemented in three phases: first, sorting out and optimizing DCS operational interlocks and adding monitoring-point sensors; second, completing system development and hardware deployment; and finally, carrying out continuous iterative optimization. Technically, multi-source data such as equipment vibration, temperature, electric current and material level are accurately collected via edge computing. A multi-dimensional alarm classification system and priority scheduling mechanism are established, and a speech-synthesis module is integrated to realize dual voice broadcast via onsite columnspeakers and the central control room. Feature-recognition models are built for typical abnormal working conditions such as roller press material collapse and mill over-filling, which automatically trigger voice alarms and link interlock protection. Environment-temperature adaptive threshold adjustment is introduced for key parameters (e.g., bearing temperature) to reduce false alarms. After commissioning, the number of alarm items decreased from 299 to 184. The false-alarm rate was lower than 1% during 60d online operation. The system realizes the transformation from manual on-site watch-keeping to online monitoring, from empirical judgment to data-driven identification, and from audible?visual prompts to voice broadcast. It effectively improves fault-response efficiency and production stability.
马纯辉, 郭智彬, 冯锐, 杨国进, 邓选成. 水泥生产线粉磨系统语音报警系统的开发和应用[J]. 水泥技术, 2026, 1(5): 53-61.
Ma Chunhui, Guo Zhibin, Feng Rui, Yang Guojin, Deng Xuancheng. Development and Application of Voice Alarm System for Grinding System of Cement Production Line. Cement Technology, 2026, 1(5): 53-61.
