An Index for Rail Weld Health Assessment in Urban Metro Using In-Service Train

Table 1

Sources for errors and measures to cover it.

Action in procedure	Potential and occurred error	Effect on results	Root cause	Taken action
Sensor arrangement	(i) Improper accelerometer	High	Unable to record desired pattern and frequencies	(i) A preliminary test is one to determine proper accelerometer capacity.
	(ii) Low accuracy of arrangement	Medium	Effects of vehicle dynamic on results	(i) Other references are checked for an optimum arrangement.; (ii) The leading wheel is used to minimize vehicle effect.
Data acquisition	(i) Unwanted noise	Medium	Electromagnetic and mechanical vibration	(i) Static test is done to determine background noise.; (ii) Shielded cables are used.
	(ii) Vehicle vibration	Medium	Vehicle dynamics	(i) Vehicle dynamics effects are determined as velocity-dependant vibration and cleared from the results if needed.
Data processing	(i) High computational volume	Low	(i) High volume of measured data; (ii) Modern methods required higher computation time	(i) Optimum parameters for windowing, method, etc., are chosen to minimize calculation time.
Pattern extraction and index definition	(i) Vehicle location error; (ii) Change in velocity; (iii) Repeatability; (iv) Comprehensiveness of results	Medium	(i) Odometry and signalling error; (ii) Normal operation of the train; (iii) Possible fake impacts in results; (iv) Change in train and track characteristics	(i) An innovative odometry algorithm is developed to control fault location error based on rail welds locations.; (ii) Normalizing the acceleration with velocity and setting a limit velocity for low speed.; (iii) Repeated data acquisition on different tracks with different train and track characteristics is made to check the repeatability and comprehensiveness of the presented algorithm.