# I. Introduction he 25KV, 1?, 50 Hz (Industrial Frequency) A.C. System was adopted for Indian Railways for high density traffic routes since 1957 collaboration with SNCF(French national Railways), since it was superior to other types of Electric Traction systems[1to 7]. The Traction system is divided into three major areas .i.e. i) Power Supply Installations (PSI). ii) Over Head Equipment (OHE) and iii) Remote Control (RC). All the above three are inter linked with one another and failure on any one of them may cause disruption in traffic (train movements) and causing loss of punctuality of trains. The major important area of Railway Electric Traction is Over Head equipment (OHE). The OHE consists of electrical conductors, huge number of insulators, various fittings, and numerous attachments to hold and maintain it in its place. A failure occurring in any one of the numerous parts of OHE could result in a breakdown causing heavy disruptions in railway traffic and causing indefinite delay of train services. So, there is a need of reliable and failure free OHE for smooth passage of Pantograph. A scheduled maintenance of the OHE with meticulous inspection of every OHE installed parts is necessary to avoid any types of failure. The adjustments at the crossovers or near the overlaps spans have to be checked for any deviations from the specified standards leading to the pantograph getting entangled with the OHE. Also a periodic examination of the OHE parameters as per the design, healthiness of various components and its geometry is necessary to attain zero defects and to achieve high reliability. All OHE breakdowns .i.e. whether it is major or minor must be handled with equal urgency for preventing unduly disruptions and for renewal of the services. The design of overhead equipment and pantograph ensures constant contact is maintained with contact wire by the pantograph with adequate pressure to obtain current. Improper adjustment of stagger and heights of contact wire at turnout or cross -over results in pantograph entanglement with overhead wires while moving on the main line [4 & 5]. To avoid this, turnouts are to be adjusted such that the contact wire of secondary line remains 5 cm above the main line, OHE at obligatory structure and also the contact wires shall not be less than 30 cm up to 10m from obligatory structure. The emergency crossovers in between Up and Dn. main lines are also equipped with similar to turnouts. In case of panto entanglement the pantograph and OHE are both damaged and electric traction traffic is dislocated. In this paper, one of the OHE defects [2] that can cause 43% among the OHE failures is presented. i.e. due to improper adjustment of 309 Nos Cross over and 1190 Nos. turn out OHEs of Secundrabad division of South Central Railway (Indian Railway). # II. Related Work Cross Over and Turn-out : ). This means a crossover will have two turnouts, which are connected by a cross over may be parallel or not, may be curved or may be having gradient in one track, or may be at different levels. In the case of electrified tracks, at turnouts and crossings additional design parameters are obligatory to be considered for smooth take of pantograph from one OHE to another OHE (figure .3). Since one cross over is constituted of two turnouts, each turnout parameters are to be designed as per the convergent or divergent point. The OHEs of the two mainline tracks are designed as per the normal principles of design irrespective of type of crossover. But the cross -over OHEs are to be designed in correlation with main line OHE[ 1& 7]. Another important parameter of turn out/ Crossover OHE is that, the gradient of contact wires. Every cross over will have two elementary sections of the two tracks, separation of which is achieved by erection of section insulator. The erection of section insulator has got certain technical parameters to be followed during erection [5] which should not be infringed in any case by simply following the turnout heights of contact wire, otherwise results in panto entanglement. The cross over OHE should be designed keeping the technical parameters for consideration 1) The heights of both mainline tracks OHE are to be adjusted such that there should be minimum allowed gradient of the crossover OHE. 2) Adjustment of crossover OHE should be done at a time in accordance with the as erected drawings supplied by the construction organization during erection and commissioning of OHE. Any incorrect adjustments of stagger and height at turn out or cross over OHE causes entanglement of the pantograph with overhead wire during its movement on main line or else entanglement with main line during its movement on overhead line. This problem may be overcome by making sure that height of the contact wire at turn-out or cross-over near obligatory locations is maintained 50 mm above the mainline contact wire and nearly 9m additional distance while pantograph moves on main line to make sure no contact is made with contact wire at turn out and cross over where the track separation is nearly 150 to 700mm. # Procedure for Adjustment of Turn Out and Cross Over: A pre check that is necessary prior to the adjustment of Turn-out and Cross-over (figure .2) near any location is inspection of ATDs of main and loop lines for their free movement. The other steps in the procedure are, 1. Measure fitting of the obligatory mast from L/L & M/L tracks, track separation and perform turn-out adjustments at obligatory point according to SED. 2. Arrange 'G' Jumpers at a distance of 5.6m from obligatory mast at the points of cross-over and turnout in the parallel run side direction. 3. Fix the contact wire at turn-out at a height 50mm above the M/L contact wire near obligatory mast located at the Cantilever. 4. Perform according to schedule, adjustments of the A&B droppers and tune distances of the A&B droppers from the obligatory mast for M/L and for turn out OHE. 5. Remove hogging on the M/L contact wire with adjustments of the length of 'B' and adjacent droppers to 10 m from the obligatory mast on M/L OHE contact wire in the direction of turn outside. 6. Perform height adjustments of L/L contact wire one mast previous to the obligatory mast considering M/L & L/L track level diff. 7. Execute in the direction of turn out side near 10 mts. distance adjustment of the height of loop line contact wire to min +30mm (with related adjustments of loop line B and the adjacent droppers) according to the height of main line contact wire. 8. Maintain at least 50mm diff. in the contact wire height at M/L & L/L OHE at obligatory mast. 9. Ensure panto does not come in contact with the contact wire of L/L cross over during operations of the tower wagon on M/L. 10. Running tower wagon at turn out track so that M/L contact wire achieves take in/take off panto pan of 650mm + /-20cm from the center of panto of the tower wagon. 11. Maintain for a cross type turn out near obligatory a height diff. of +1.5 cms, at 5 mtrs and 10 mts distance from mast. In running the tower car on M/L ensure panto does not come into contact with contact wire at turn out and if necessary prevent contact by adjustments of A&B dropper. # III. Heuristic Scale to Strayed Proneness (HSSP) of Cross-Over OHE a) Dataset Preprocessing The OHE cross-over preventive maintenance log record contains 25 attributes with the values of type categorical. The 25 th attribute is remark, which is descriptive, which is replaced by a Boolean value to represent the state of the respective record log. The dataset that used here in this experimental model is formed from the real time logs collected from the South Central Railway, Secundrabad division. The data set is the combination of records labeled as true and false. In order to balance the computational overhead, we aimed to select optimal attributes from the records labeled as true and also from the records labeled as false. Hence forth, initially we convert all alphanumeric values to numeric values and continuous values to be converted to categorical as follows. The value of Hamming Distance obtained here is to denote the difference between unique values of same attribute from records labeled true and false. This is one of the significant strategy to assess the difference between to elements in coding theory. This strategy is applied to identify the distance between the unique values observed for an attribute in record set labeled as true and labeled as false. The hamming distance between given vectors formed by the values of optimal attributes selected (see section 3.2) from records labeled as true and records labeled as false respectively. # Global i i n m ? = Begin ({ } { }) 0 i i i i if cx cx CX cy cy CY then ? ? ? ? ? ? { } { } i i i i CZ cx cx CX cy cy CY ? ? ? ? ? ? Else 1 CZ ? End | | 1 { } CZ CX CY j hd CZ i â??" = = ? // CX Further we build a weighted graphWG such that values of ( ) FS SR as vertices and edges between these vertices under the constraints such as: a) No edge is between two vertices, if those two are values of same attribute b) An edge between two vertices that justifies the above condition is possible if those two vertices are appeared together in at least one given record. Each edge weighted by the ratio of the given records contains the two vertices of the edge. Further the closeness of the features ( ) FS SR and records SR is assessed by using bipartite graph (see fig 7) build between those records and features. | | 1 ( ) { } | | i k j i j k r f f k i k r f i ew f r k j ew r < â??" = â??" ? ? ? ? = ? ?. (Eq1) Then the feature weights from the bipartite graph are assessed as follows: Initially a matrix that contains the edge weights of bipartite graph will be formed, such that each feature weight towards each record. Further the link based ranking model [23] will be applied on bipartite graph (see fig 7) to evaluate the connected set. The confidence of each record r is proportionate to degree of all feature weights. Hence the influence of record r will be derived from these weights. Intuitively, a record with high confidence should contain many of the features. This approach is as follows. # Global FS NR b v b v b v b v b v b v b v b v b v = those contains all the unique values of all attributes of SR and NR respectively. Here Let matrix representation of records and features of set ( ) FS SR as a matrix ' M '. The value represents the edge weight between record and features that calculated by using Eq1. q p p p qp b v b v b v q NR b v a v a v q p NR Find Feature support as matrix FC by summing up the columns of each row of matrix ' M (which is transpose of matrix M ) that represents edge weights between a feature of ( ) FS SR and all records of SR . The matrix multiplication between M and FC to obtain the record support. # RC M FC = × Then the confidence of each feature { ( )} f f FS SR ? ? can be measured as follows ( ) | | 1 | | 1 { ( ) } ( ) ( ) SR i i i SR i i RC r f r c f RC r = = ? ? = ? ? | ( )| 1 { ( ) } ( ) 1 ( ) FS SR j j i j SR i i c f f r hssp r ec r = ? ? = ? ? ( ) i ec| | 2 1 { ( ) } ( ) (| | 1) SR i i SR i SR hssp r r SR hssp pd hssp SR = ? ? ? ? ? ? ? ? = ? ? Here in the above equation ( ) # SR pd hssp represents the probable deviation of the strayed proneness of records of SR The procedure that followed to assess the SR hssp (heuristic scale to strayed proneness) and ( ) SR pd hssp (probable deviation) from records of the SR will be adopted to assess the heuristic scale to strayed proneness NR hssp IV. # Experimental Study The real time data (see sec 3.1) was used in experimental study. The overall data collected is the size of 303labeled records and each record contains 24 fields. Among these 213 records were used as training set to define the scale proposed. The remaining 90 records were used to test the scale defined in training phase. The empirical study delivered promising results. The statistics explored in table 2 Total number of records found to be true positives are 68 and false positives are1 As per these results, the accuracy of the proposed heuristic scale under hamming distance ratio of 0.052 is 96.5%. The accuracy observed from the attributes selected under hamming distance ratio >0.25 also reflected the same performance accuracy, but delivered magnitude computational overhead that compared to the computational overhead observed under hamming distance ratio greater than 0.052 The observed time complexity is scalable since the completion time is incrementing with the same ratio against the increase in features count due to lower hamming distance ratio (see fig 9). Hence it is obvious to conclude that hamming distance based optimized attribute selectionis significant to minimize the computational overhead of the proposal, which is done without loss of accuracy. Results observed for statistical metrics under divergent hamming distance thresholds. The statistical metrics [8], such as precision, recall, and F-measure were used along with prediction accuracy. The result obtained for these metrics under divergent hamming distance thresholds are explored in table 13 and visualized in fig 10 V. # Hamming # Conclusion Heuristic scales to assess strayed proneness of the OHE provided on railway cross-over tracks has been proposed in this paper. The Hamming Distance Analysis of the recorded attributes is devised to obtain the optimal attributes, which is promising to simplify the process of defining Heuristic Scale to Strayed Proneness of the line cross-over's. The reinforcement relation between records and attribute values is analyzed to define the proposed heuristic scale. In order to this the proposed model is using a weighted graph that built by using optimal attribute values as vertices and their associativity scope as edge weight. The other significance of the proposed heuristic scale is that a given report of a line cross over is assessed by couple of heuristic scales called SR hssp and NR hssp , which are built from the respective records of type strayed and normal. The experiments were done using real time data collected from Secundrabad division of South Central Railway zone. The exploration of the results concluding that the Hamming Distance Analysis is promising and significant to select optimal attributes of the records dataset. The heuristic scales proposed are observed to be robust and is with minimal process complexity and retains the maximal prediction accuracy. In future the evolutionary computational approach like GA, CUCKOO search can be devised. 1![Fig. 1 : Electric Locomotive with panto raised condition. The figure.1 shows the electric locomotive with pantograph in raised condition. The Pantograph is provided on loco roof to receive the required power from overhead contact wire to the electric locomotive. If the pantograph has free movement over the contact wire without any obstacles will ensures the locomotive works efficiently. However if broken fragment of the pantograph arise in the path of overhead wires or else broken overhead wires come in the path of the pantograph it will result in panto entanglement as shown in figure 4 and 5.](image-2.png "Fig. 1 :") 234![Fig. 2 : Testing of turn out OHE with Tower Wagon](image-3.png "Fig. 2 :Fig. 3 :Fig. 4 :") 5![Fig. 5 : Electric Locomotive with Broken Pantograph](image-4.png "Fig. 5 :") 6![Fig. 6 : Cross-over & Turn out layout of Track A cross over is a track which diverges from one track and converges to another track (Figure.6). This means a crossover will have two turnouts, which are connected by a cross over may be parallel or not, may](image-5.png "Fig. 6 :") 7![Fig. 7 : An example bipartite graph between features and records. The edge between a feature { ( )} i i f f FS SR ? ? and record { } j r SR ? is the average of the edge weights between i f and all other features in j r found in weighted graphWG (see Eq1)](image-6.png "Fig. 7 :") ![represents the all possible unique values of the optimal attributes 1 { },......{ } p b NR b NR ? ? respectively. Further the values of the set ( ) FS SR and ( ) FS NR are referred as respective features of SR and NR .](image-7.png "") ![r in above equation represents the edge count connected to record i r Then the heuristic scale of strayed proneness by strayed records SR hssp can be found as follows: records count Further find the probable deviation of the " SR hssp " of records from SR as follows:](image-8.png "") 8![Fig. 8 : A line chart that representing the attributes scope under different hamming Distance thresholds a) Performance AnalysisThe results obtained for hamming distance threshold is greater than 0.082 are (i) false negatives:4 (strayed record claimed as normal),(ii) true negative are 17 (claimed records normal that are actually normal), true positives: 66,(iii) false positives: 3, and the prediction accuracy is 92.3%.The experiments also conducted on the same data set under hamming distance ratio >0.052 and >0.025, the results are as follows:Total records Tested 30% (90 records)(70 strayed records and 20 normal records) hamming distance ratio is greater than 0.052 Total number of records found false negative are 2(strayed record claimed as normal) and found to](image-9.png "Fig. 8 :") 9![Fig. 9 : The completion time of defining HSSP under divergent hamming distance ratios](image-10.png "Fig. 9 :") 10![Fig. 10 : performance analysis of the prediction accuracy under divergent hamming distance thresholds given.](image-11.png "Fig. 10 :") 1IDDescriptionValuesF1Main line contact wire height onin H meters (decimal fraction).one sideF2Main line contact wire height onin H meters (decimal fraction).another sideF3T/Out Contact wire height on onein H+50meters (decimal fraction)sideF4T/Out Contact wire height onin H+50meters (decimal fraction)another sideF5Stagger of main line wire staggerin 200millimeters (decimal fraction)on one sideF6Stagger of main line wire staggerIn200 millimeters (decimal fraction)on another sideF7Stagger of T/out wire stagger onIn300 millimeters (decimal fraction).one sideF8Stagger of T/out wire stagger onIn300 millimeters (decimal fraction).another sideF9sag of section insulatorin Zero mmF10Take-off from one sideIn 650 to 720mmF11Take-off from other sideIn650 to 720 mmF12Point take-off from one sidein 4meters (decimal fraction)F13Point take-off from other sidein 4meters (decimal fraction)F14stagger of section insulator±100 mmF15Runner towards the centre of T/out. In mm (1.65 minimum)F16runner away from the centre ofIn mm (1.45 minimum)T/outF17condition of ATD of T/out & mainFree to moveLineF18Hex tie rod of limiting device[6]F19Setting distance of obligatory mastIn Metres (mi 3.0Mtrs)from one side.F20Setting distance of obligatory mastSame as abovefrom other side side.F21Track separation of obligatory mastIn mm ( 150 to 700mm)from one side.F22Track separation of obligatory mastSame as abovefrom other side.F23Distance of 'G'JumperIn 5.6meters (decimal fraction).F24Length of 'G' jumperIn4 meters (decimal fraction).F25Labeltrue/false? Let consider each attribute with alphanumericvalues, then list all possible unique values and listthem with an incremental index that begins at 1.? Replace the values with their appropriate index. ? Let consider each attribute with continuous values, and then partition them into set of ranges with minN respectively i. Hamming Distanceand max values, such that the records distributedevenly through all these ranges.b) Optimal Attribute Selection ? Partition the given network transactions as intruded ( I ) and normal ( N ) ? Find the hamming distance (see sec 3.3.1) between unique values of each attribute of I with the counter part of N ? 2hsspSR7.11324( pd hsspSR)1.538036hsspNR2.982372( pd hsspNR)0.623142The attributes of the records selected asare defined in tables 5, 6 and 7 and the same isoptimal under different hamming distances thresholdsvisualized in fig 8. 3Attribute IDHD10.08273620.05222630.07550740.03582150.027238 4Attribute IDHamming Distance Ratio10.08273630.07550760.09362770.0755690.057556100.081826110.068945120.073293140.084389150.089146160.06504210.071742220.089284230.056242 5Attribute IDHamming Distance Ratio10.08273620.05222630.07550740.035821 6Attribute IDHamming Distance Ratio10.08273660.093627140.084389150.089146220.089284 © 2016 Global Journals Inc. (US) © 2016 Global Journals Inc. (US)Table 6 : * AC Traction Manual II Part-1, 2002 * Comments/Suggestions given by RDSO * IRIEEN journals * Comments/Suggestions given by CETI/THK * Research Designs & standards organization maintenance instruction TI/MI/0028 Rev. 2(28.09 2001 * Research Designs & standards organization maintenance instruction TI/MI/0035 * Treatise on Electric Traction distribution Volume-I & II by Indian Railways Institute of Electrical Engineers, Nasik road 2007 India * OHE maintenance manuals of Centre or advanced Maintenance Technology Gwalior, India * Using multivariate statistics BGTabachnick LSFidell SJOsterlind 2001 * UJShenoy KGSheshadri KParthasarathy HPKhincha DThukaram 2004. November * MATLAB/PSB based modeling and simulation of 25 kV AC railway traction system-a particular reference to loading and fault conditions TENCON 2004. 2004 IEEE Region 10 Conference IEEE 100 * Preliminary investigations into the loco-man factor on the Indian railways RNSen AKGanguli Applied ergonomics 13 2 1982 * Railway electrification: design of overhead equipment FBKitchin JHolland Proceedings of the IEE-Part IA: Electric Railway Traction the IEE-Part IA: Electric Railway Traction 1950 97 * High Speed Protection Scheme for Traction OHE of 25 kV AC Indian Railway System BBhalja RPMaheshwari Industry Applications Conference, 2007. 42nd IAS Annual Meeting. Conference Record of the 2007 IEEE IEEE 2007. September * Adaptive Digital relay for comprehensive distance protection of traction overhead equipment RPMaheshwari HKVerma Developments in Power system protection, Conference Publication No. 434, IEE 25-27th March 1997 * Study on microprocessor based adaptive protection relay for heavy duty electric traction system GaoShibin HeWeijun ChenXiaochuan Developments in Power System Protection, Sixth Intl. Conf., Pub. No. 434, IEE 25th-27th March 1997 * Voltage Regulation on 25 kV AC railway systems by using Thyristor switched capacitor GCelli FPilo SBTennakoon Proceedings of 9th Intl. Conference on Harmonics and Quality of Power 9th Intl. Conference on Harmonics and Quality of Power Oct. 2000 2 * Static distance protection relay with parallelogram characteristic for 25 kV AC single phase, 50 Hz traction Overhead equipment Research Designs and Standards organization, Ministry of Railways India Specification No. ETI/PSI/141 (10/90 * Voltage form factor control and reactive power compensation in a 25 kV Electrified railway system using shunt active filter based on voltage detection Pee-ChinTan RobertEMorrison DonaldGrahameHolmes IEEE Transactions on Industry Applications 39 2 March/April 2003 * Application of mho and quadrilateral distance characteristics in power systems MJDomzalski KPNickerson PRRosen Developments in Power System Protection, Seventh Intl. Conf., Pub. No 479 9-12 April 2001 * Static relays for the protection of 25 kV single-phase traction overhead equipment HKVerma 1976 Roorkee, India Department of Electrical Engineering., University of Roorkee Ph.D. dissertation * Study on microprocessor based adaptive protection relay for heavy duty electric traction system GaoShibin HeWeijun ChenXiaochuan Proc. Of 6th Int. Conf (IEE) on DPSP, Conf. Publication No. 434 Of 6th Int. Conf (IEE) on DPSP, Conf. Publication No. 434 25th -27th March 1997 * Voltage Regulation on 25 kV AC railway systems by using Thyristor switched capacitor GCelli FPilo SBTennakoon Proc. Of 9th Int. Conf. (IEEE) on Harmonics and Quality of Power Of 9th Int. Conf. (IEEE) on Harmonics and Quality of Power 1st-4th October 2000 2 * Matlab/PSB Based Modeling and simulation of 25 kV AC railway traction system-A particular reference to loading and fault condition UJShenoy KGSheshadri KParthasarathy HPKincha DThukaram Proc. Of Int. Conf. (IEEE) on TENCON Of Int. Conf. (IEEE) on TENCON 21st -24th November 2004 C * Getting Started Manual" V3, Manitoba HVDC Research Centre Inc Pscad/Emtdc 2001 Manitoba Canada * Electric Locomotives Class WAG6A for Indian Railways Training Program Manual No. TK December 1988 ) Vehicle Description, ASEA Traction ABS-72173 * SPICE for Power Electronics and Electric Power MHRashid 1993 Englewood Cliffs, N.J. Prentice Hall * Introduction to PSCAD DennisWoodford 2001 V3,, Manitoba HVDC Research Centre Inc, Manitoba Canada * Control of Electrical Drives WLeonard 1985 SpringerVerlag New York * Electric motor drives and control: past, present and future PCSen IEEE Transactions on Industrial Electronics 37 6 1990 * A Review of Iterative Harmonic Analysis for AC-DC Power Systems BCSmith JArrillaga ARWood NRWatson IEEE Transaction on Power Delivery 13 1 January 1998 * Microprocessor based protection for AC railway supplies GCWeller PJHindle Proc. Of Int. Conf (IEE) on Main Line Railway Electrification (DPSP), Conf. Publication No Of Int. Conf (IEE) on Main Line Railway Electrification (DPSP), Conf. Publication No 25th -28th Set 1989 312 * A Wavelet tour on signal processing StephaneMallat 1998 Academic Press London (UK * Adapted wavelet analysis from theory to software MVWickerhauser 1994 IEEE Press New York * Wavelet-based data compression for power disturbances using minimum description length data EYHamid ZIKawasaki IEEE Transactions on Power Delivery 17 2 April 2002 * Modeling and protection of a three phase power transformer using wavelet packet transform SASaleh MARahman IEEE Transactions on Power Delivery 20 2 April 2005