A Health Monitoring Index for Rotating Machinery Based on the Logarithmic Interval Quantile Root Mean Square Value Ratio

In rotating machinery condition monitoring, dimensional and dimensionless indicators, as conventional classic health indicators (HI), are extensively applied in evaluating and monitoring the health status of equipment. However, dimensional indicators are insensitive to early faults, while dimensionless indicators exhibit a decreasing and oscillating trend after detecting early faults. To solve this problem, a health indicator for rotating machinery condition monitoring, based on the log interval quantile root mean square ratio (LIQRR), is proposed. This indicator constructs interval quantiles based on the improved quartile method, and then obtains normal and anomalous subsets. The root mean square (RMS) values of the different subsets are calculated and ratioed, resulting in the logarithmic interval quantile RMS ratio. The key parameters of this process are optimized using the Honey Badger Algorithm (HBA). Experiments were carried out on the XJTU-SY dataset and four industrial fault operation datasets. The results indicate that the proposed indicator, compared to existing HI, does not show decreasing and oscillating behavior after detecting early faults, exhibits strong monotonicity, and is unaffected by variations in load and rotational speed, making it applicable for real-time condition monitoring of rotating machinery.