Auto-Regression and Statistical Process Control Techniques Applied to Damage Indication in Telecommunication Masts

The authors and ESAT Digifone, a telecommunication company in the Republic of Ireland are examining the potential of using measured vibration data for structural integrity assessment of telecommunication masts and towers. At present 'climb-down' inspections, which are non-scientific, qualitative and relatively expensive, are undertaken. ESAT Digifone currently owns and maintains approximately 200 telecommunication towers. It is anticipated that this number will rise to 1000 over the next few years and more quantitative methods of assessment, which could assist in prioritising these structures for repair and maintenance work, would be attractive management and quality control tools. This paper discusses auto-regressive and statistical pattern recognition techniques for damage identification in the context of telecommunication masts. An auto regressive function, developed using optimisation techniques, is used to model the measured response of a mast or tower and the error between the measured and generated data is calculated. Statistical process control techniques are employed to characterize the variation of the mean and the standard deviation of this error term. Subsequent measurements on the same structure are passed through the auto-regressive model and an error term calculated. Statistically similar and dissimilar error distributions are representative of similar and altered structural responses respectively.