EP0568167A1 - Method for determining the rolling resistance of railway vehicles - Google Patents

Abstract

For the online determination of the rolling resistance of a train section, the actual running path/running time behaviour of the section is recorded during freewheeling; during this process, the times when individual axles of the section pass over contacts are determined. In addition, the computational time points when the contacts are passed over are determined from the simulation of the process. If the actual running behaviour of a section deviates from its computationally calculated running behaviour, false values for the variables taken into account in a simulation model are set in the simulation model. By listing the numerical values for the influence which changes to these variables have on the computationally calculated times of passing over contacts, in a matrix and application of the orthogonalisation method to this matrix, correction values for the variables are obtained. Using these correction values, new times of passage over the contacts and new values for the influence of parameter changes are determined and, from these, new correction values are determined again for the variables. These processes are repeated until a sufficiently accurate adaptation of the computationally determined running path/running time behaviour to the appropriate running path/running time behaviour is obtained. The value of the rolling resistance which is then used for the simulation gives the actual rolling resistance of the train section in question. <IMAGE>

Description

The invention relates to a method according to the preamble of claim 1. Such a method is, for. B. from DE-PS 12 36 256 known.

In order to ensure that individual vehicles or groups of vehicles, hereinafter referred to as departments, that run over the discharge hill of a maneuvering system reach certain route points at a certain speed, the departments are braked to predetermined exit speeds on their way. The determination of these outlet speeds is extremely problematic because it requires knowledge of a large number of plant-specific and department-specific data. This data is only partially known from the beginning or can be determined with sufficient accuracy. The rolling resistance is one of the department-specific data, which so far can only be determined with great imprecision; this essentially includes the frictional forces between the wheels and the rails. The rolling resistance, which differs from department to department, is extremely dependent on the weather and therefore only ever applies to a specific department and the weather conditions prevailing during the investigation.

There are a variety of approaches for determining the rolling resistance of running departments. This is usually done by measuring the running behavior of each department in front of a brake using a three-point measuring section in a track section of constant inclination (Signal + Draht 83 (1991) 12, page 196).

It is also already known to determine the running resistance of a department by determining its running behavior in a piece of track whose inclination varies (DE-PS 11 92 858). This is done by assigning functionally defined inclination values to departments of different lengths. To simplify this, it is assumed that the inclination acting on a department is only dependent on the length of the department; in fact, it depends on the running position of the individual axes and the inclination of the track there.

Usually, the rolling resistance of a department running through a measuring section in free running is determined by inferring from the distance from the track switching means and the times of their actuation the entry and exit speeds of the departments into or out of a measuring section. These speed values are squared and then subtracted from each other. The resulting value is divided by the double distance between the measuring bases for the speed determination, multiplied by the value of a reduced acceleration due to gravity, and then subtracted from the inclination of the measuring section (DE-PS 12 36 256). The values for the rolling resistance of a department determined using the aforementioned relationships are quite imprecise for various reasons. The measuring base is usually quite short due to the constant inclination of the track. As a result, the values of the entry and exit speeds into and out of the measuring section do not differ too much, so that any measurement errors have a strong percentage effect. Measurement errors arise from the gauntlet of the vehicle axles, So a wobble movement of the wheel sets which, depending on the respective inclination of the axle at the measuring point, leads to an early or late response of the track switching means to a vehicle with axles running absolutely perpendicular to the longitudinal axis of the track. In known measuring arrangements, the period of the wobbling movement caused by the skewing movement is approximately in the order of magnitude of the measuring section for determining the rolling resistance, ie the measuring fluctuations caused by the wobbling movement of the vehicles are particularly pronounced. The errors caused by them when determining the rolling resistance can be of the order of magnitude of a department's rolling resistance, namely up to 2 ‰.

The rolling resistances that can be determined with the known devices generally include not only the actual rolling resistance of a vehicle, which is essentially speed-independent, but also other resistances, which are largely speed-dependent, essentially track-dependent resistances due to horizontal and vertical deviations of the loaded rails within the vehicle Measuring section from the ideal line. Furthermore, the measuring devices regularly do not take into account, or only inadequately, the influence of the wind on the running departments, neither when determining the rolling resistance nor when the departments later run / runtime control to achieve predetermined running goals.

The object of the invention is to provide a method according to the preamble of claim 1, which allows the rolling resistance of running departments to be determined very precisely without this being a measuring section with constant inclination and arranged at a precise distance from each other Track switching equipment required. The method according to the invention should also still be applicable if one of the track switching devices for the determination of traffic events fails. It should do without the mandatory use of speed measuring devices to determine the vehicle speeds in the measuring section and should not require detailed knowledge of vehicle-specific parameters such as the center distances for its use.

The invention solves this problem by the characterizing features of claim 1.

Advantageous refinements and developments of the method according to the invention are specified in the subclaims.

The above-mentioned measurement errors caused by the skewing of the wheel axles and insufficiently defined rails when determining rolling resistance are largely avoided by the method according to the invention in that the influence of partial track subsidence and track displacements extends over the length of the measuring section, which will generally be greater than before, approximately compensates and that a large number of event messages are triggered in the measuring section for each axis, in which the measurement time falsifications caused by the spinning gear approximately cancel each other out. Only track switch means that are already located in the system and in principle arbitrarily distributed can be used as track switch means; their position can be determined by subsequent measurement.

The invention is explained in more detail below.

The method according to the invention is based on the consideration that the values for the model parameters included in any simulation model for describing the route / runtime behavior of freely running departments have only been correctly selected if the runway / runtime behavior of the departments calculated from the simulation model covers the actual route / runtime behavior of these departments. The comparison of the simulated with the actual driving behavior of a department according to running time or speed will regularly show certain deviations. H. the values used for the variables in the simulation model do not correspond to the actual values of these variables. The method according to the invention now provides for the values of the variables used in the simulation model to be optimized step by step using a mathematical method until the simulated driving behavior finally matches the actual driving behavior. This process takes a finite period of time after the actual running behavior of the vehicles has been recorded; this period of time is available the more in shunting mode, the earlier the running behavior of the departments in free running z. B. already detected shortly after the summit. The model parameters to be determined are required for the control of the brakes further ahead in the running path and the running time there is many times longer than the time for the adaptation of the running behavior of a department determined in the simulation model to its actual running behavior. The runtime is adjusted in online mode.

The description of the running behavior of a department in the free run happens regularly via an energy balance, in which the kinetic and the potential energy of the Department is described in different running positions; the numerical difference between these energies corresponds to the influence of all the resistances acting on the department when passing through the measuring section. In addition to the rolling resistance to be determined, which is essentially independent of speed, these resistances include a speed-dependent running resistance, which is dependent on the respective route position and essentially describes the obstacles caused by wind, track bends, wheel links, rail twists and bends and other comparable influences the rail act on the vehicle. The axle weights of the departments are recorded on the track-side sensors when the individual wheel sets pass the sensors.

The method according to the invention uses track switching means for the run / runtime description of the departments in the free run, hereinafter referred to as contacts, which are arranged mainly in the direction of travel behind the summit mountain. Their distance from one another must be known, but it is not necessary for the contacts to have a predetermined distance from one another. The measuring section defined by the contacts also need not have a certain constant gradient. Rather, the gradient can in principle be chosen as desired, but it must be measured with sufficient accuracy in order to be able to determine the slope downforce that arises there for the individual running positions of the department. The method can also be used on measuring sections that contain sections with inclines. The method according to the invention manages to determine the advancing speed of the department in the measuring section without speed measuring devices; rather, the entry speed of the department into the measuring section is initially preferably estimated; the Changes in the department's speed when advancing in the measuring section result from arithmetic trajectory tracking of the axes, taking into account the track gradient at the respective running position of the axes. The center distances to be used as the basis for route tracking may or may not be known in advance; they can be estimated at least approximately from the rollover times of the contacts and the distance between the contacts. The estimated values for rolling resistance, vehicle speed and center distances, which are required for the simulation of the running of the vehicle, are gradually calculated and optimized more precisely as the method according to the invention is used.

In order to adapt the running behavior of a department calculated from the simulation model to the actual running behavior of the department, the actual running behavior of the department must first be determined. This happens because the rollover times of the individual axes are detected via the individual contacts. The rollover time and the associated axis are recorded for each contact. For an explanation, reference is made to FIG. 1. There is schematically shown a track in the area of a drain mountain peak with a department consisting of two coupled vehicles. Each vehicle has two axles A with a first digit identifying the position of the vehicle and with a following digit following a point and identifying the position of the respective axle in the vehicle. For redundancy reasons, at least four rail contacts K1 to K4 of any type and, in principle, of any arrangement follow, preferably behind the run-off summit. For the individual contacts, the associated traffic events are indicated by the identification of the contact, the axis and the rollover time marked. The rolling over of the first contact K1 by the first axis A1.1 marks the time to for the start of the tracing.

To compare the calculated and actual running behavior of the department, the rollover times of the contacts through the individual axes are determined from the simulation model, which result from the values of the model parameters taken into account in the simulation. One of these model parameters is the department's initial speed at the beginning of the trace. This initial speed is estimated. This can e.g. B. happen by the time evaluation of the passage of successive rail contacts and taking into account their distance. The value of an average speed between these two contacts to be determined could, for. B. used as a first estimate of the department's initial speed.

In order to be able to calculate the rollover times of the individual contacts through the individual axes, it is necessary to track the travel of the individual axes over the measuring section. This can be done in such a way that the center distance of the departments is initially determined more or less roughly when passing through the individual contacts, so that a statement can also be made about the approximate position of the other axes of the department with each contact inspection event. The track gradients given there apply to the running positions of the individual axes in the measuring section. The downward slope forces on the department and thus result from these track gradients the speed changes of the department when passing through the measuring section. As already explained in the introduction to the description, the rollover times calculated from the simulation model will differ more or less from the determined rollover times. This is due to the estimated values used for the simulation, namely the initial speed of the department at the start of the tracing, the estimated value for the rolling resistance used for the simulation and the values of the center distances used for the path tracking. These values are now to be modified to adapt the calculated running behavior to the actual running behavior. For this purpose, the simulation model is linearized in the vicinity of the current estimated values for the model parameters and a linear compensation for determining parameter corrections is carried out with this local substitute model using the mathematical orthogonalization method (Schwarz, HR; Rutishauser, H .; Stiefe, E .: " Numerics of symmetric matrices ", BG Teubner, Stuttgart 1968, pages 93 to 102). With this method, the parameters that may have been very inaccurate in the simulation are gradually optimized. For this it is necessary to determine the influence of the changes of these parameters on the running behavior, ie on the running times of the departments for the individual contacts.

The differential equations given in FIG. 2 with the associated initial values for the start of the sequence tracking apply to the route tracking of a department in the measuring section and the influence of the variables on the transit times at predetermined running positions. By numerically solving the last two equations, the influence of a change in the assumed rolling resistance can be determined for each arithmetically determinable driving event and determine the assumed initial speed on the calculated running time. The influence of the change in the assumed center distances follows from the reciprocal of the calculated speed of the department (second differential equation). These values are entered line by line in the matrix according to FIG. 3 for the respective run times supplied from the model and the actual run times. The indices used behind the determination expressions in FIG. 3 indicate that the numerical value in question is a calculated value and to which traffic event this value applies. In addition to these values, which indicate the influence of changed variables on the running times, the matrix contains further values which contain the dependence of the assumed center distances on the running times. A consecutive number after the wheelbases 1 indicates the wheelbase, 11 is the distance between the first and second wheelsets and lq is the distance between the penultimate and last wheelsets of the department.

By entering the numerical values obtained by partial differentiation for the influence that changes in the parameters used for the simulation model have on the running time of an axis for a contact, the matrix shown in FIG. 3 is finally obtained, the larger the larger is the number of contacts whose travel events are evaluated and the larger the number of axes in a department. In general, three contacts are sufficient to use the method according to the invention. The accuracy of the method depends largely on the length of the measuring section, less on the number of contacts; There is already redundancy with three contacts. From a certain The number of axes can no longer be used because it must then be assumed that the department no longer goes into free run, but is either still pushed or already braked, so that the simulation model is no longer valid. In this case, rolling resistance replacement values based on empirical values must be assumed for the further vehicle control.

The mathematical orthogonalization method is applied to the values of the matrix shown in FIG. This method results in correction values Δw _R , ΔvO, Δll to Δlq for all parameters included in the matrix. These values indicate the manner in which the parameter values previously used in the simulation model are to be corrected in order to achieve a better adaptation to the actual running behavior, ie to minimize the deviation between the actual and calculated rollover times. The correction values found in each case by the orthogonalization are introduced into the simulation model as additive variables and change the values of these variables used up to now. A new simulation run then takes place, which in turn calculates the rollover times of the contacts through the individual axes. These rollover times differ from the values found in the first simulation run; they are closer to the actual rollover times than assumed in the first simulation step. Here, too, the influences that changes in the variables now changed on the transit times are again recorded numerically and brought into a matrix according to FIG. 3. By orthogonalization again, correction values for the parameters are determined and a third iteration cycle is started, in which the Rollover times of the contacts can be determined by the individual axes. In this way, the calculated runtimes are increasingly aligned with the actual runtimes. It is no longer sensible to carry out further iteration steps very quickly, because the simulation model with the values then found largely corresponds to the actual running behavior of the department, that is to say that the running times then calculated largely correspond to the actual running times of the department. The termination criterion for the iteration cycle for optimizing the runtime is preferably reached when the value of the squares of deviation t² or a forecast for the value of the squares of deviation after the parameter correction has been carried out between the mathematically determined and the actual runtimes leads to a relative change in the squares of deviation, which is below a predetermined value Value of z. B. 1 ‰.

Experiments have shown that the method according to the invention converges to the extent that the termination criterion is given after only three iteration steps. The value for the relative change in the squares of deviations from the measured and calculated runtimes is a system-specific value, ie it differs from drain system to drain system. According to an advantageous embodiment of the method according to the invention, it is advantageous to record the absolute value of the square of deviation when an iteration cycle is terminated and to compare it with the termination values previously determined in the treatment of other departments. In this way, information can be provided on the one hand about vehicles whose calculated running behavior does not adapt to the actual running behavior in the same way as that of the other departments; such departments could e.g. B. partially filled tank wagons sloshing content. Long-term observations could also cause influences due to B. Changes in the rail bedding or imprecise switching track switching means can be detected.

The use of the method according to the invention makes it possible to determine the rolling resistance of departments with high precision without the need for a measuring section with a constant gradient or track switching means arranged at a predetermined distance from one another. The method according to the invention even allows the rolling resistance to be determined very precisely in the event of failure of one or more of the contacts, by using the measured values of another contact instead of the measured values of this contact. If, for example, the contact K1 fails in the exemplary embodiment in FIG. 1, the contact K2 takes over its function, the first axis of each sequence marking the start of the trace as it passes this contact K2. For reasons of redundancy, it is therefore advantageous to know from the outset the distance of a fifth or possibly even sixth contact to the other contacts and to keep the track inclinations in a correspondingly expanded track area for the determination of the slope downforce. To determine these downhill forces, the track inclinations must also be known in an area outside the actual measuring section, which is defined by the length of the respective department.

As explained at the beginning, the determination of the rolling resistance is necessary in order to be able to carry out a somewhat exact braking of the target. Especially when it comes to braking with a department's slow running speed close to the target point great advantage to take into account the influence of wind on the department concerned, both when determining the rolling resistance and for the later braking of the running target. To take the wind influence into account when determining the rolling resistance of a department, the wind direction and the wind strength in the area of the measuring section are detected and from this as well as from department-related data such as the front surface of the department the influence of the wind on the speed of the department is determined and for the simulation of the assumed vehicle speed subtracted or added. This gives a value for the rolling resistance that is independent of wind influences. For the braking of the running target it is necessary to take into account the influence that the wind has on a department on the way to the target point. However, very different wind conditions can prevail near the target track than in the measuring section. It is therefore necessary to measure the strength and direction of the wind along the entire route so that it can be taken into account when controlling the brakes. This is advantageously done by introducing the found values for wind direction and wind strength into a polynomial approach, with which the wind influence applicable there can be calculated for each location of the route. For the area of the directional tracks, it is advantageous to reduce the wind values determined in each case compared to the values determined by free-standing transducers, in order to take into account the influence of the wind shading by departments already in the target tracks.

In the method for determining the rolling resistance of running departments, which is described in more detail above, estimates of the running time are first used for the simulation of the running time / travel path behavior of running departments Rolling resistance, the initial speed and the center distances of the departments. Even if the initial values are not all too well estimated, the method according to the invention converges sufficiently quickly as a rule after three iteration steps. Of course, it is possible to use values for these estimated values that were determined by a measuring process. This could, for. B. include a speed measurement in the area of the measuring section by a radar device, which is there anyway or there could be a track-bound speed measuring device which determines the speed of a department in a measuring section from the detuning of a track circle by the first axis of a department. Regarding the center distances, quite good estimates could be taken from the corresponding pre-reported data, provided that this data is available or this data could be supplied by a radar device for tracking the long distance. It is also an advantage, at certain standardized center distances, as z. B. on bogies, instead of the estimated values to access the then exactly existing distance values.

A very important advantage of the method according to the invention is that for the determination of the rolling resistance of running departments, only time values, that is to say primary values, and not secondary values derived therefrom, such as speeds or accelerations, have to be used. All such secondary values would be associated with measurement errors which would be far greater than the time measurement errors, which are preferably given exclusively by the raster of the temporal sampling of the sensor messages. This is also a reason why the method according to the invention converges very quickly and stably.

Claims (11)

Method for determining the rolling resistance of railway vehicles passing a line in free running using a simulation model for the route / runtime description of the individual departments during the course and evaluation of the driving events ascertainable by track switching means,
characterized, - that the event messages of at least three track switching devices (K1 to K3) in the path of the departments for these track switching devices (e.g. K1) according to running time (e.g. t1) and number of the wheelset (e.g. A1.1) the department that triggers this event, - that these transit times (t1) are compared and saved with the transit times (t1 *) calculated in the simulation model for passing the individual track switching means through the individual axes, or that the respective transit time deviations (t1 - t1 *) are saved, - That for each calculated runtime value (t1 *) by partial differentiations of the time according to the parameters flowing into the simulation, the influences of a change in the rolling resistance assumed in each case ${\text{((δt / δw}}_{\text{R}}{\text{)}}_{\text{l}}\text{)}$

, the assumed speed at the start of the route tracking ${\text{((δt / δVo)}}_{\text{l}}\text{)}$

and the axis distances assumed for determining the travel path positions of the individual axes $\text{((δt / δℓl) ... (δt / δℓq))}$

the departments are numerically determined to the respective runtime value and these values are compiled line by line into a matrix, - That the orthogonalization method is applied to this matrix and the resulting change values for rolling resistance (Δw _R ), initial speed (ΔVo) and center distances (Δℓl ... ℓq) are added to the values taken into account in the simulation, that a new simulation process with new matrix formation and orthogonalization then takes place with these new values, - That these process steps are repeated with continuously improved estimates for the parameters until further improvements do not result in any noteworthy improvements in results in the sense of aligning the run times determined in the simulation model to the run times actually measured, and - That at least the value of the rolling resistance (w _R ) determined in this way is used for the further route / runtime description of the relevant process. Method according to claim 1,
characterized, that the event-oriented deviations between the actually measured time values and the times calculated from the simulation model are squared and then added, - That the value of the squares of deviation and a prognosis for the value of the squares of deviation are derived from this after execution of the parameter correction - And that the iteration cycle for the parameter correction is terminated if the relative change in the squares of deviation that can be achieved thereby falls below a predetermined value. Method according to claim 2,
characterized in that the absolute value of the squares of deviation is recorded when an iteration cycle is terminated and with those previously during the treatment other departments determined termination values to detect sporadic sudden deterioration in earnings and / or longer-term tendency deterioration in earnings is compared. Method according to claim 1,
characterized in that the values introduced in the first iteration step of the method for the variables rolling resistance, initial speed and center distances are measured values and / or estimated values. Method according to claim 4,
characterized in that at least the estimated values for the initial speed and / or the center distances are obtained by inference from the actual rollover times of the track switching means. Method according to claim 1 or 5,
characterized in that for the route / runtime simulation of departments with standardized center distances, the actual center distances given by the standardization are taken into account for the route / runtime description. Method according to claim 1,
characterized in that the description of the variables in the simulation model is made by the following differential equations,

with the initial values

wherein s the current travel position of the department's top wheelset, sO the travel position of the top wheel set at the beginning of the trace, V the current speed of the department, Vo the initial speed of the department at the beginning of the trace, t the running time of the department since tracing started and w _{R are} the rolling resistance. Method according to claim 1,
characterized in that the location and speed-dependent running resistances outside the rolling resistance acting on the departments passing the measuring section are numerically taken into account in the simulation model and that for the further runtime / route control of the departments outside the measuring section the locally applicable and speed-dependent running resistances in addition to the determined rolling resistances are taken into account. Method according to claim 1,
characterized in that the influence of wind on the speed of the sections passing the measuring section is taken into account in the simulation model and that the wind conditions which occur there are taken into account for the further route / runtime control of the sections outside the measuring section. A method according to claim 8,
characterized in that the consideration of the wind influences in the determination of the parameters and the route / runtime control of the departments is carried out in such a way that wind direction and wind strength are measured at predetermined running positions and the wind-dependent forces acting there on a department in the running direction are determined and that the thus found Values are introduced into a polynomial approach that describes the wind influence that applies to each location on the route. Method according to claim 9,
characterized in that, in the area of the directional tracks, the wind influence values determined there are reduced in accordance with the wind shading by the departments already running in the directional tracks.

Images