How to automatically and autonomously control a compaction machine.

Abstract

The compactor (C) has a distance measuring device (WMS, GPS, 32) for accurately acquiring the position of the track construction machine on the track, and a method for automatic autonomous control of the compactor (C) is described, including acquisition of signals of actuators of the working units (23, bv, 18, 26) of the compactor (C). In order to advantageously provide an automatic tamping behavior, during tamping, ballast bed data are acquired via sensors (23, bv, 18, 26), from which actual ballast bed variables are obtained, stored for subsequent work steps and analyzed using a machine learning device (17, ML), wherein an analysis of the ballast bed condition data (EF7, S9, A3) is made based on the machine learning technical method (ML, 17), the ballast bed variables are analyzed with respect to the reduction in compaction force occurring in the longitudinal direction of the track, and from the analysis, work instructions (EF7, S9, A3) for an optimal work method are determined and stored, wherein in subsequent work steps the compactor executes the work instruction data fully automatically and autonomously depending on its actual position on the track and the work instruction data corresponding to the actual position.

Description

[Technical field]

[0001] The present invention relates to a method for automatic autonomous control of a railway construction machine, in which the railway construction machine has a distance measuring device, including an accurate correspondence to the track, thereby obtaining the position of a working unit of a compactor, transmits position-accurate work instructions for each sleeper area to be tamped to a control computer of the compactor, and the compactor executes the work instruction data in a fully automatic and autonomous manner depending on its current position on the track and the work instruction data corresponding to the current position. [Background technology]

[0002] Most railway tracks are constructed as ballast superstructures. In this case, the sleepers are located in the ballast. Under the action of the forces of the wheels of the trains running over it, the ballast is rounded, partially destroyed and worn away. This causes irregular settlements in the ballast and deviations in the track profile, both lateral and vertical. Settlements of the ballast bed cause irregularities in height, cant (in curves), twist, gauge and alignment. The irregularities themselves also increase the action of the forces acting to destroy the ballast.

[0003] If the ride comfort or safety thresholds of the values ​​for these configurations set by the railway administration are exceeded, maintenance works are planned and carried out in due time. Currently, track construction machines are often used to remove and straighten the track irregularities for this configuration. To control the process, there are measuring systems that obtain the current track condition for the variables of straightening, elevation, twist and cross gradient. To be able to allow the track to be used again after this type of maintenance work, the superstructure machines are equipped with so-called inspection measuring or inspection recording systems. So-called inspection tolerances are set by the railway administration for the quality of the track condition after improvement by the superstructure machines or by other methods. The inspection tolerances represent the minimum requirements for the quality of improvement for the obtained configuration.

[0004] In addition to the inspection tolerances there exist safety-related clearance tolerances, which represent the boundaries that must be respected so that the track on which the work has been done can be safely cleared again for train traffic. Compliance with the tolerances is verified by an inspection recording system. The track construction machine (track compactor) has a lead car operator who is responsible for controlling the machine with respect to the target geometry and for recording the track condition remaining after the maintenance work by means of a recording system.

[0005] Currently, track maintenance is planned on the basis of the track geometry, which is obtained regarding the rail condition. Track measurement vehicles run on the track at regular intervals and obtain the condition regarding the track geometry. In this case, the track condition is divided into sections, often of about 200 m length, and standard irregularities of elevation, direction, cant and twist are obtained. Besides these statistics, specific individual irregularities are measured. If the statistics exceed the predefined ride comfort tolerances, maintenance work is planned and carried out. As soon as an individual irregularity exceeds a predefined critical value, measures are taken and the irregularity is immediately removed, otherwise train traffic would be endangered and slow-down areas would have to be set up or the road would have to be closed.

[0006] The maintenance machine receives as set values ​​the track target shape, the previously recorded and measured track irregularities and the area to be maintained. No other settings are made.

[0007] For the tamping operation, there is a second operator who is the tamping operator. Regardless of the type of ballast defect, the second operator usually performs standard tamping. Methods such as multi-tamping, raising of points, etc. are left to the second operator.

[0008] How exactly the tamping operator tamps the track, e.g. which tamping pressure he uses, which clamping time he uses, whether he performs multiple tampings of the same sleeper, whether he works on sections with more or less rise, whether he chooses a more or less open width, etc., is left to the judgement, experience and motivation of the operator.

[0009] Prior to use of the superstructure machine, the actual track conditions are measured and compared to the target track conditions using various known measurement systems. The height and orientation differences are transmitted to the compactor as track corrections along with the track target shape.

[0010] There are compaction machines that are specialized for tamping points (separable tamping units, so-called split head units, auxiliary lifting devices for branching tracks, swivel compaction tools, etc.) and compaction machines that are preferably designed for rail tamping. Compaction machines are known both with periodic and continuous working advance. Besides, there are also single and multiple type compaction machines. Multiple type compaction machines tamper several sleepers at once in one working cycle. Moreover, multiple type compaction machines may also be used to tamper only one sleeper.

[0011] The planning of the construction site is currently carried out on the basis of the track geometry measured by the track measuring vehicle, but whatever the cause of the track irregularity, for example the ballast can be broken, rounded and crushed in the area of ​​the rail joints, but similar irregularities occur when the base is very hard, which are called "white points". At this point the running dynamics lead to the crushing of the ballast, which is indicated by the occurrence of mineral dust. The ballast is very susceptible to damage if it is installed for a long time. Fine components, organic materials or large parts of the earth pushed upwards from the base could fill the intermediate spaces of the ballast granules. In practice it is known that the track condition in this type of ballast structure cannot be permanently corrected with a track compactor. In practice it is also known that individual irregularities occur randomly distributed on the track. About 40% of these localized irregularities can be permanently repaired. 60% of these irregularities recur within a short period of time. Tracks with good ballast conditions are tamped on average about every 4 years. Individual irregularities, which indicate the destruction of the ballast, require maintenance procedures approximately every one to three months. With each tamping, the ballast is damaged via the tamping tool due to the high compaction forces. The use of the compactor generates maintenance costs and disrupts the operation. As long as the work is being carried out, trains cannot run, and in this case a deceleration point is set up for the adjacent track where the work is being carried out. It is also known that heavy soiling of the ballast bed requires high compaction forces. In fact, there is no room for movement between the ballast bed particles, since they are filled with fine material. This increases the compaction forces that must be expended to move the ballast. At the same time, the stability of the correct track shape is reduced on such soiled track beds, since the frictional forces and meshing between the ballast bed particles are smaller. Before the tamping operation, the track shape irregularities are recorded in electronic form, often by independent measuring methods, stored and transmitted to the compactor computer. Track geometry irregularities typically have wavelengths between 10 and 25 m with amplitudes between 10 and 40 mm. Longer wave irregularities in the range 25 to 70 m with higher irregularity amplitudes also occur.

[0012] The tamping unit fixes the condition of the track during maintenance measures. The maintenance measures are carried out via a tamping tool, the so-called tamping pick, which is inserted into the trackbed next to the sleeper and compacts the ballast under the sleeper via a linear closing movement combined with a compaction vibration. In this case, a linear closing movement by a hydraulic cylinder is typically combined with a vibration amplitude that is mechanically generated by an eccentric shaft. The new fully hydraulic tamping drive generates the linear closing movement and vibration simultaneously.

[0013] In WO 2019091681, a track construction machine is known. The track construction machine acquires network data and transmits the net data to a system center. The track construction machine has a sensor system and collects raw data. On the basis of this, it should be planned when and where the use of the track construction machine can be performed. During the navigation operation, raw data for activating the network data is acquired, i.e. data such as structural changes or disturbance values, but not specific ballast variables introduced during tamping. The course of the compaction force is not obtained from the collection of network data. A fully hydraulic drive for a tamping unit is disclosed, for example, in AU 513973. For the closed-loop and open-loop control of this drive, the clamping movement is acquired by means of an on-board distance sensor. In this case, the tamping pressure is measured via a pressure sensor. This allows variables such as compaction work, ballast hardness, ballast contamination, compaction force, compaction time and ballast stiffness to be measured and derived, as described in AU patent 520117. In AU patent 515801 it is known how an optimum tamping time can be set depending on the measurements. Via this fully hydraulic tamping drive a free and continuous adjustment of the opening width of the tamping tool is also possible.

[0014] The tamping operator is currently responsible for selecting the correct settings of the tamping unit, e.g. tamping pressure, feed time, descent speed of the tamping unit, opening width, tamping depth, track elevation or multi-tamping, etc. Subsequent planning of the work, e.g. planning of the tamping work itself and preparation of work such as ballast replacement in areas of localized irregularities, localized drainage improvements, etc. This increases track maintenance costs and reduces the sustainability of the obtained track condition.

[0015] The track sections with high track stiffness form high points that change in height only slightly with train traffic. The more the track bed stiffness varies, the greater the force interaction between the wheels and the rails, the higher the track load, and the earlier the track geometry deteriorates. The singular short fault points in the track tend to elongate longitudinally under the action of high dynamic forces on the track, increasing the degree of track irregularity, and creating subsequent irregularities caused by the track vehicles.

[0016] Sensors are known that are able to determine the position of the sleepers on the track as the compactor passes over them. By means of devices of this kind the machine can be correctly positioned for tamping in a fully automatic manner. Thus, in practice, machines that operate fully automatically are known.

[0017] Storing track geometry target data in a database of the infrastructure manager is partially realized, as is the possibility of downloading this or reproducing the results. Machine learning systems are background art. Machine learning is a broad term for computer-aided, experience-based knowledge creation. For this, algorithms build statistical models based on training data. Patterns and regularities in the learning data are recognized. Thus, the system can also evaluate unknown data. A GPS system on board the compactor allows a precise assignment of sleepers to track kilometers and recorded measurement parameters via GPS coordinates.

[0018] Virtual GPS correction data services are known that transmit correction data to a suitable GPS receiver. This requires only a mobile GPS-supported measuring vehicle moving on the track. RTK-GPS has the advantage that the absolute position can be determined very precisely using the RTK correction data (approximately 5 mm for position and 10 to 15 mm for height). The more satellites and satellite systems that are received simultaneously by the GPS receiver, the more accurate the results are. Modern satellite receivers simultaneously receive and evaluate the satellite systems GPS, GLONASS, GALILEO, BeiDou, QZSS, IRNSS and SBAS. The satellite receiver can transmit data to the correction service and receive correction data on a second channel. An accuracy in the range of 5 to 15 mm is too inaccurate for the calculation of correction values ​​for track compactors for lifting or straightening, but is sufficient to set absolute reference points for the track geometry. With a similar accuracy, sleepers on the track and elsewhere can be located and GPS coordinates provided. These points or sleepers are precisely and specifically located by the superstructure machinery equipped with a RTK-GPS system. [Prior art documents] [Patent documents]

[0019] [Patent Document 1] International Publication No. 2019091681 [Patent Document 2] Austrian Patent No. 513973 [Patent Document 3] Austrian Patent Invention No. 520117 [Patent Document 4] Austrian Patent No. 515801 Summary of the Invention [Problem to be solved by the invention]

[0020] The problem underlying the present invention is to provide a method for automatic autonomous control of a track construction machine, which avoids the aforementioned drawbacks. The method should provide the track construction machine with not only target shape data and track state correction data in general, but also with accurate, locally clearly assigned work instructions, so that the track construction machine autonomously tamps with high quality and also adapted to the characteristics and requirements of the ballast bed, thus avoiding human defects. At the same time, ballast bed variables are acquired by the tamping machine during the work, which should be analyzed by a computer and, after the work is finished, preferably transmitted to an infrastructure manager in order to prepare the next track maintenance work. [Means for solving the problem]

[0021] The present invention solves the problem set out by the features of claim 1. Advantageous developments of the invention are described in the dependent claims. In particular, the invention solves the problem by acquiring ballast bed data via sensors during tamping, acquiring actual ballast bed variables from the ballast bed data, storing them for a further work process and analyzing them with the aid of a machine learning device, whereby an analysis of the ballast bed condition data is made on the basis of machine learning technology methods, the ballast bed variables are analyzed with respect to the occurring reduction in compaction force in the longitudinal direction of the track, and from the analysis work instructions for an optimal work method are determined and stored, whereby the tamping machine executes the work instruction data in a fully automatic and autonomous manner in a further work process as a function of the actual position on the track and the work instruction data corresponding to the actual position.

[0022] The automatic autonomous control of the working mode of the compactor, the tamping unit of the compactor and the lifting and levelling unit is as follows: To the control computer of the compactor, position-precisely (e.g. via GPS coordinates) working instructions are transmitted for each sleeper area to be tamped (this includes multi-tamping, larger opening width of the tamping tool, tamping pressure, lift, setting of maximum compaction force, tamping time, automatic tamping time depending on compaction or setting of working sequence at points, where for example in the point compactor split head unit, the outer part can be swiveled outwards, etc.). These working parameters were obtained in the previous working process, full or partial tamping of the track, and stored for the following working process.

[0023] The compactor is positioned precisely in the sleeper area to be tamped via automatic sleeper recognition or GPS coordinates. In the position reached, the compactor can then execute the set work instructions in a fully automatic and autonomous manner according to the set work instructions, creating new work instructions for the next step if necessary, and then moving via the automatic transfer device to the next sleeper area, where the work sequence is repeated accordingly until the entire set work area has been worked over.

[0024] The set work instructions do not have to be executed in a fully automatic manner, but can be announced to the operator for each sleeper area, whereupon the operator sets and executes the set work mode.

[0025] According to the present invention, ballast bed data and operation data are acquired during tamping by the fully hydraulic tamping drive and the tamping drive sensor, from which current ballast bed parameters (e.g. ballast bed hardness, compaction force, tamping time, tamping unit plunge time, tamping unit braking acceleration during plunge operation, current GPS position or orbit kilometers, current climb and straightening values, current climb and straightening forces, etc.) are calculated, stored and analyzed using a machine learning device having machine learning techniques.

[0026] According to the invention, during the operation a ballast bed condition record is created and displayed for the notification of the tamping operator or lead vehicle operator, and after the operation a ballast condition report is created from the measurement data, whereupon both are communicated to the infrastructure manager as a basis for work preparation for future tamping operations.

[0027] In the absence of work instructions being communicated by the infrastructure manager or responsible work planner, ongoing analysis of the ballast bed data during the tamping operation provides the tamping operator with appropriate guidance on the optimal operation method.

[0028] In this case, according to the invention, the measurement data of the tamping operation is analyzed by a rule-based expert system (KI system or other machine learning program) with regard to correlations of the compaction force with longitudinal sudden drops (individual irregularities) or statistical variables, such as standard irregularities, average values, track height irregularities, from which instructions for the optimal working mode are determined and set.

[0029] The data obtained by the hydraulic tamping drive and the tamping drive sensors allows for the derivation of useful and valuable information. For the analysis, algorithms based on the field of artificial intelligence (KI) are used. KI systems are able to find relationships and patterns in different volumes of structural data that have little or no human interpretation. KI systems create predictions on the occurrence of track condition deterioration and track irregularities, from which maintenance suggestions are provided to improve the durability of the track condition. For this purpose, other techniques of machine learning (ML) (rule-based learning) are also suitable.

[0030] Rule-based expert systems (XPS) can assist operators with specific suggestions. XPS has a great advantage in areas where there is deep expertise in the interpretation of algorithmic models and data states.

[0031] Below is an example of how a work order might look: The work order may be computer-aided and created by the works engineer. In this case the list contains all the sleepers for which tamping is to be done.

[0032] [Table 1]

[0033] Various work orders can be coordinated and scaled in coordination between the infrastructure manager and the machine operator. In this case, a work order means: EF7 Individual deviations in the track. Operation variables: 3x tamping, enlarged opening width (75cm), tamping time 3x0.65s S9 Heavily contaminated ballast area Operation parameters: single tamping, tamping pressure 180 bar, tamping time 1.2 seconds A3 Typical settings Operation variables: Maximum compaction force limit up to 35kN, tamping with automatic optimum tamping time selection, damping depth 485mm

[0034] Through the setting of generic set values ​​(operation variables to be used in the absence of specific requirements) and operation instructions for specific parts and their terminology, it is easy for the control computer of a modern compaction machine to interpret and execute the corresponding predefined instructions.

[0035] The drawings illustrate the subject matter of the invention. [Brief description of the drawings]

[0036] [Figure 1] FIG. 2 shows a schematic side view of a compactor. [Diagram 2] FIG. 1 shows a schematic diagram of a fully hydraulic tamping unit. [Diagram 3] FIG. 1 shows a block diagram of the track geometry computer together with the compactor controller. [Figure 4] Ballast bed inspection records are shown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0037] FIG. 1 shows a compaction machine 38, C with a trailer 39, which travels on a railway rail S by means of track-driven running gear 34, 36. The compaction machine 38, C comprises a tamping unit 30 with a fully hydraulic drive and measuring sensors 37, a lifting and levelling unit 42, 43 for introducing lifting forces FH and levelling forces FR into the track, a work measuring system aw, bw, 35 and an inspection and recording measuring system ar, br, 35. The work measuring system aw, bw, 35 and the inspection and recording measuring system ar, br, 35 are, for example, chord measuring systems. The trailer is connected to the compaction machine via a drawbar 40. The tamping unit 30 has a standard opening width B of the tamping tool 29. The compaction machine 38, C further comprises a control system 19 and a track geometry master computer 17 with a monitor 20. Via an antenna 33, data is exchanged wirelessly with the infrastructure manager. Via the GPS system 32, the working area is precisely captured in coordinates.

[0038] FIG. 2 shows a tamping unit B with a fully hydraulic drive Z. Via sensors 23, the clamping distance 31 and the compaction force (via pressure sensors in the tamping cylinder hydraulic system) are acquired and transmitted to the control computer 18, which forwards these forces to the track geometry computer 17 for processing. Via acceleration sensors bv, the braking deceleration of the tamping unit is measured when penetrating the ballast bed. The harder the ballast bed, the greater the braking deceleration. The fully hydraulic drive allows the opening width of the tamping arm 30 by the tamping tool 29 to be adjusted from the normal opening width B to a larger width BE. This allows the larger opening width BE to be used at the point where the ballast is damaged to compact the ballast granules from the middle area and push them under the sleepers, where the partially damaged, crushed ballast powder is replaced by undamaged ballast granules, which increases the durability of the track condition. The rail S is fixed on the sleepers 41.

[0039] Figure 3 shows a block diagram of the track geometry computer 17 with the machine's control device 19. The sensors of the fully hydraulic tamping units 18, 26 are read and analyzed with the machine learning program ML. Via the monitor 20 the machine operator is informed about the ballast bed condition and can receive work instructions. At the end of the tamping operation a ballast bed report 22 and a ballast bed record 21 are generated by the track geometry computer 17 and the machine learning program ML. These data are sent wirelessly 25 to the infrastructure manager's or machine owner's database or cloud. The ballast bed variables under each sleeper are precisely captured via GPS and assigned to each sleeper. Via the distance measuring wheel WMS the local condition is assigned over the track kilometers.

[0040] Figure 4 shows a schematic representation of a ballast bed record A. Record channel 1 shows the braking deceleration bv of the tamping unit, channel 2 shows the track height deviation before the operation determined from a preliminary measurement of the actual track condition and a comparison with the target track condition, channel 3 shows the track bed hardness and channel 4 shows the resulting compaction force. Channel 5 is an event channel displaying various special track situations or track features via markers 6, 7, 8, Br. Symbol 6 represents a rail joint and symbol 7 marks a track point where the ballast has been destroyed and therefore a satisfactory compaction force cannot be achieved. Symbol 8 represents a stored image and Br represents a bridge. At each singular deviation a photograph is embedded in the record. When the machine operator activates it, the corresponding photograph 8 is displayed. 10 shows a specific irregularity with destroyed ballast, which is on the one hand visible in a steep drop in the compaction force and, since the ballast does not have a high penetration resistance at that point, also in a drop in the tamping unit braking deceleration 11. 9 forms another fault point. Another fault point occurs at the welded joint, as can be seen with the symbol 6. Such specific irregularities can be detected and recognized relatively easily by machine learning programs (or rule-based systems). Comparing the course of the height irregularity (channel 2) with the course of the ballast bed hardness (channel 3), it can be seen that they have an approximately inversely proportional behavior 12. At the hard points, height peaks are formed. Where soft points are present, subsidences (valleys) are formed. Via the correlation function it can be determined how well these two channels are correlated. If the correlation is high, this has an effect on the persistence of the track height state, since the ballast deformations are formed correspondingly strongly. The hardness of the ballast bed δ BH The larger the standard deviation δ, the stronger the variation in stiffness and the greater the interaction forces between the wheels and the rails, the less the track condition will be sustainable. On the other hand, the average value of the ballast hardness 16, 17 indicates the degree of contamination and wear of the ballast. The more the ballast bed is polluted, the higher the ballast hardness 16, 17. The compaction force (channel 4) runs proportional to the ballast hardness. Very low values ​​of the compaction force indicate new positions 14 (new ballast) or anomalies 9, 10 with bad ballast. Standard deviation δv The lower the value, the smaller the variation in stiffness and the better the continuity of the track conditions. The horizontal lines indicate track kilometers (e.g. 76400).

[0041] An example of a ballast bed analysis report is shown below. This is preceded by statistical decisions that give general information about the worked section. The analysis by the machine learning system ML provides information about the persistence of the track condition and the ballast bed hardness. When faults 9, 10 occur, they are displayed with their type, exact condition, length and characteristic values. The communication of these data to the infrastructure manager or responsible work planner forms the basis for the work instructions for the next maintenance work. The analysis also provides an estimate of the track condition deterioration rate, which is important for the time planning of the next maintenance work. These data are easily put into a machine-readable format and communicated.

[0042] Ballast Bed Report statistical judgment

[0043] [Table 2]

[0044] On-track stability The ballast bed is defective. The durability of the track condition is deteriorating. The estimated rate of track deterioration is 1.6 mm / year. Ballast bed hardness The average ballast bed hardness was 254 Nm. The ballast bed is in a borderline highly contaminated state. Track screening is recommended. Major defects (with crushed / rounded ballast) have been found in the tamped areas. Replacement of 11 or more sleepers is recommended in the 76580 area. Fault location 1

[0045] [Table 3]

[0046]

Table 4

[0047]

Table 5

Claims

1. The compactor (C) has a distance measuring device (WMS, GPS, 32) for accurately acquiring the position of the track construction machine on the track, and a method for automatic and autonomous control of the compactor (C) including acquisition of signals of actuators of working units (23, bv, 18, 26) of the compactor (C), During tamping, ballast bed data are acquired via sensors (23, bv, 18, 26), from the ballast bed data actual ballast bed variables are acquired, stored for a further work step and analysed using a machine learning device (17, ML), in which an analysis of the ballast bed condition data (EF7, S9, A3) is made on the basis of a machine learning technical method (ML, 17), the ballast bed variables are analysed with respect to the reduction in compaction force occurring in the longitudinal direction of the track, and from the analysis work instructions (EF7, S9, A3) for an optimal work method are determined and stored, in which the tamping machine performs the further work step fully automatically and autonomously in accordance with the actual position on the track and the work instruction data corresponding to the actual position. A method comprising:

2. transmitting position-accurate work instructions (EF7, S9, A3) for each sleeper area (41) to be tamped to the control computer (17, 19) of the compactor (C); The compactor (C) moves to the respective longitudinal position on the track according to the precisely acquired corresponding longitudinal position on the track (WMS, RTK-GPS, 32) and the given work instruction data (EF7, S9, A3), autonomously performs the work in the sleeper area (41) to be tamped according to the work instruction data (EF7, S9, A3), and then moves automatically via the automatic traveling device to the sleeper area (41) to be next processed according to the work instruction data (EF7, S9, A3), and repeats the cycle of processing the work instruction data (EF7, S9, A3) and moving to the sleeper area (41) to be next processed until the given work area has been processed.

2. The method according to claim 1 .

3. 2. The method according to claim 1, characterized in that predetermined work instructions (EF7, S9, A3) for each sleeper area (41) are displayed (20) to the operator, who sets and executes a predetermined work mode.

4. 4. A method according to claim 2 or 3, characterized in that during operations a ballast bed condition record (Fig. 4, A) is generated, a ballast condition report is generated together with the results of the analysis of the ballast bed data by a machine learning device (17, ML), and both the ballast condition record (Fig. 4, A) and the ballast condition report are communicated (33, 24, 23, 25) to an infrastructure manager.

5. 5. The method according to claim 1, characterized in that the machine learning device (17, ML) analyzes the ballast bed data (A, FIG. 4) during the tamping operation to transmit optimal operation procedure instructions to the operator.

Images