Developing a Predictive Maintenance Data System for Manufacturing Plants in Kaduna State

Unplanned equipment downtime causes significant production losses in Nigerian manufacturing, and predictive maintenance systems using sensor data and machine learning can reduce these losses substantially. This study developed a predictive maintenance data system for manufacturing plants in the Kaduna Industrial Estate, Kaduna State, North West Nigeria. A professional system design methodology was applied, with site assessments at three manufacturing facilities (textile, food processing, and metal fabrication), structured consultations with 12 plant engineers and 6 data engineering specialists, and benchmarking against IIoT predictive maintenance frameworks from the Industry 4.0 literature. The system designed specifies sensor deployment requirements, time-series data ingestion pipelines, feature engineering procedures for vibration, temperature, and current signatures, and a survival analysis model using the Cox proportional hazards framework for component failure prediction. Alert threshold calibration for the Kaduna industrial environment is specified, accounting for local power fluctuation patterns that affect baseline sensor readings. A dashboard for production planners and a maintenance scheduling integration API are included. Expert review by nine industrial IoT and manufacturing analytics specialists confirmed the system's technical and operational adequacy. The study recommends a 12-month pilot at the textile facility with maintenance cost tracking as the primary evaluation metric, and advocacy for Kaduna State Investment Promotion Agency to offer data infrastructure grants to manufacturing SMEs adopting predictive maintenance systems.

Keywords: predictive maintenance, manufacturing analytics, Kaduna State, IIoT, survival analysis