METHOD FOR DETERMINING THE POSITION OF A SWITCHING ROLLER

Description

[0001] The invention relates to a method for determining the position of a rotating switching drum by means that generate current or voltage deviations at a driving electric machine, wherein the electric machine acts on the switching drum and is configured to rotate the switching drum by 360 degrees and the means are arranged distributed over the outer circumference or the end face of the switching drum. State of the art

[0002] The travel path of a shift drum is limited by housing-mounted mechanical end stops. The shift drum position can be uniquely referenced and thus taught by approaching both end stops and checking the calculated path length. This process can be repeated cyclically throughout the gearbox's service life, guaranteeing consistent accuracy.

[0003] If these end stops are removed, the aforementioned method of learning is no longer possible. With a "360° rotary switch," the end stops are omitted so that the rotary switch can move to any position from either direction.

[0004] One solution consists of a position sensor on the shift drum, which reports the absolute position to a control unit. No further learning is then necessary.

[0005] However, an additional sensor always means undesirable additional component costs due to the sensor itself and the wiring.

[0006] Furthermore, the requirements regarding accuracy and robustness are very high. The shift drum position, and therefore also the sensor's accuracy, is absolutely crucial for certain transmission tasks such as shifting and parking lock functions. This accuracy must also remain consistently high throughout the transmission's entire service life.

[0007] A method using an analog armature current signal and its evaluation is known, for example, from DE 195 11 307 C1. To prevent interference pulses superimposed on the armature current signal from being included in the evaluation of a current ripple count, the analog armature current signal is typically processed accordingly before digitization, for example, by frequency filtering. These measures serve to ensure that a current ripple signal as free of interference pulses as possible is available for digitization and subsequent counting evaluation. In this context, interference pulses are non-commutator-induced pulses that superimpose on the armature current signal.

[0008] From EP 1 208 638 B1, a method is known that is particularly suitable for use in position detection of an adjustment device for opening and closing a window pane, a sunroof, or for adjusting a seat in a motor vehicle. Position determination, for example of the pane, can then be carried out with sufficient accuracy solely by evaluating the current ripple signal without the use of additional sensors. However, in such an application, it is necessary to determine the reference current ripple during motor start-up and thus before any movement of the window, so that the desired correction can be carried out simultaneously with the movement, without having to first store the entire current ripple signal and then correct it after the movement has finished with regard to determining the position of the driven element.

[0009] German patent application DE 10 2013 100 213 A1 discloses a method for adapting transmission control elements to precisely locate synchronization contact points. By rotating a shift drum, a shift fork actuates the synchronizers as it passes each one. The transmitted torque is measured, and the measured feedback current of the actuators is determined.

[0010] DE 10 2010 013 962 A1 discloses a method for detecting blockage events. Position sensors are used to measure the position of the shift drum.

[0011] From DE 199 14 394 A1 a method for determining the position of a rotating switching drum is known by means that generate current or voltage deviations at the driving electrical machine.

[0012] EP 3 770 467 A1 discloses a method for calibrating at least one rotatable switching drum in a drive train of a motor vehicle.

[0013] The object of the invention is to provide a position determination of a switching drum that operates independently of stops and sensors on the basis of current peaks. Description of the invention

[0014] The problem is solved by a method according to claim 1, here a method for determining the position of a rotating switching drum by means that generate current or voltage deviations at the driving electric machine, wherein the electric machine acts on the switching drum and is configured to rotate the switching drum by 360 degrees, wherein the means are arranged distributed over the outer circumference or the end face of the switching drum, and wherein the means generate a mechanical resistance during the rotation of the switching drum, and wherein the evaluation of the measured current or voltage deviations is carried out by means of pattern recognition.

[0015] The invention eliminates the need for an additional position sensor for the switching drum. Referencing is achieved solely through pattern recognition in the current or voltage signal of the switching drum motor.

[0016] The current signal, which runs continuously as a current over time, is already present in a controller as a measured quantity and can therefore be used without further ado.

[0017] The solution according to the invention improves the function of the switching drum determination, provides greater switching comfort, reduces wear through reliable determination of the switching drum position, and saves costs and energy, thus also reducing CO2 emissions.

[0018] The cost savings are achieved by eliminating the need for an additional sensor.

[0019] It is advantageous that pattern recognition is achieved by rotating the switching drum and measuring the electrical machine, allowing input variables available when the vehicle is stationary to be learned. A previously defined current profile, known in relation to the switching drum position, is detected via pattern recognition during the switching drum's movement, thus enabling a conclusion to be drawn about the switching drum's position.

[0020] Pattern recognition according to the present invention is achieved using a neural network. Neural networks have the advantage that, if properly trained, they are generally less sensitive to noisy data or disturbances in the input patterns than conventional algorithms.

[0021] It is advantageous that pattern recognition continuously optimizes the switching drum position determination during operation. The evaluation of the current-time signal based on pattern recognition can run during normal operation, thereby continuously monitoring positional accuracy. In contrast, position determination using end stops only occurs in very specific situations.

[0022] The evaluation of the measured input variables takes place after the measurement in a time period from the past and therefore never in real time.

[0023] Pattern recognition can incorporate further characteristic patterns in the current-time signal, thereby continuously increasing accuracy.

[0024] The pattern recognition can learn on its own and detects wear-related changes in the current-time signal, thus increasing robustness.

[0025] Advantageously, means are provided that generate current or voltage deviations at the driving electrical machine, wherein at least one pressure means is provided that presses a ball into bores in the end face or outer circumference of the switching drum via a spring tension. Description of the characters

[0026] Figure 1 shows a switching drum arrangement, Figure 2 shows a flowchart, Figure 3 shows an example of pattern recognition.

[0027] The invention aims to achieve a predefined pattern in the current signal of the switching drum. Many possibilities exist for this. According to the present invention, a current increase is achieved by a mechanical resistance in the travel path of a switching drum. The mechanical resistance can be artificially generated by component modifications or be inherent in the system itself, as in an overpressure unit, e.g., at the parking lock. As an example, the present invention incorporates an artificial resistance into the switching drum system, as shown in Figure 1 depicted.

[0028] The method relies on a type of detent in the switching drum assembly 10. A switching drum 3 has guide grooves 6 that extend around the circumference of the switching drum and serve to guide switching forks. An electric drive, an electric machine 7, engages the switching drum and rotates the switching drum 360 degrees. The axis 9 of the switching drum 3 is, by way of example, mounted on a housing via bearings 8.

[0029] In the Figure 1 A spring-loaded pressure piece 1 with a ball 2 is used for detent operation. As the switching drum 3 passes through a bore 4 on its end face 3a, the ball 2 slides into the bore 4, securing it in place. Initially, a higher force is required to re-tension the spring 5 of the pressure piece for further rotation of the switching drum 3.

[0030] In Figure 2A top view of the end face 3a of the switching drum 3 is shown, with the bores 4a, 4b, 4c and the pressure piece 1 depicted. As can be seen, the bores are located at different distances from each other on a circular line of the end face 3a, so that the pressure piece 1 gradually engages in each of the bores 4 as the switching drum 3 rotates.

[0031] The increased force of locking and pushing out the balls 2 can be measured as an increase in the switching drum current I, which is already detected at the switching drum, see Figure 3 .

[0032] Figure 3 This shows the switching drum current I over time. The current peaks Imax are clearly visible and assigned to the respective bores 4a, 4b and 4c.

[0033] Knowing the precise nature of this current rise, delta I, allows the position to be determined by pattern recognition as the switching drum rotates 360°. To minimize misinterpretations, three or more detents are incorporated.

[0034] The position determination is learned before the vehicle is used, and the position determination is continuously refined during operation.

[0035] The procedure for teaching the shift drum position is part of the process of Figure 4 to see.

[0036] The process begins at step 31 with the start of the learning process. In the first step, step 32, the previously known position is deleted from the software.

[0037] Step 33 concerns the movement, rotation of the switching drum in a predetermined direction at a predetermined speed.

[0038] When a first bore is detected, the position P1, e.g. at the current peak I 4a, is stored and in step 35 the switching drum is rotated further.

[0039] In step 36, the second position P2 is determined and saved.

[0040] Step 37 rotates the shift drum to position P3, which is then saved.

[0041] After saving the three example positions, the rotation of the switching drum 3 is stopped in step 39. In step 40, the position is verified, and in step 41, a plausibility check is performed.

[0042] The position is determined by comparing and adjusting it against predefined values. These values ​​are determined by the transmission, clutches, etc., and vary depending on the transmission design.

[0043] If everything is okay, the process ends at step 42. Otherwise, the process jumps back to the beginning.

[0044] The detent can be of any type. It can be a sphere / bore, an ovality in a rotating component, or any other type of stop or resistance. Many embodiments are conceivable. The frequency of the detent can also be varied. If several detent groups are distributed across the 360°, even a slight rotation—not a full 360°—can achieve referencing.

[0045] The pattern recognition itself can be implemented using different algorithms. According to the invention, a neural network 21 is used.

[0046] An example of this implementation is described in the Figures 5 and 6 depicted.

[0047] The results for pattern recognition are shown at the switching drum 3 itself as input variable 20. The switching drum current I, the rotational speed v, and the rotational acceleration a of switching drum 3 are used.

[0048] Other input variables used include values ​​of the electric machine 7 of the drive, as well as the rotational speed of the electric machine 7, and the acceleration of the shaft of the electric machine 7.

[0049] Neural network 21 represents the processing of input data for pattern recognition. The input data for the neural network is represented in graph 22, which depicts various input variables.

[0050] The Figure 6 Graph 22 is shown again, enlarged. It plots the switching drum rotation position 23 with the angle ω against time t.

[0051] In this embodiment, the switching drum rotates 360°. At point 24 of the switching drum's movement, a shoulder can be seen in its path, which is measured as a current peak.

[0052] Pattern recognition attempts to determine the position of the switching drum from the positions of the current ripples. At each point in time, e.g., at time 25, a defined time period is considered in the past. The system searches for unique patterns that allow conclusions to be drawn about the current position value. Pattern recognition does not provide a real-time reference value.

Claims

1. Method for determining the position of a rotating gear shift drum (3) using means which generate current or voltage deviations on a driving electric machine (7), wherein the electric machine (7) acts on the gear shift drum (3) and is configured to rotate the gear shift drum (3) through 360 degrees, characterized in that the means are arranged so as to be distributed over the outer circumference or the end face (3a) of the gear shift drum (3), and wherein the means generate a mechanical resistance during the rotation of the gear shift drum (3), and wherein the measured current or voltage deviations are evaluated via pattern recognition (21) by a neural network (21).

2. Method according to Claim 1, characterized in that the pattern recognition is learned by rotation of the gear shift drum (3) and measurement on the electric machine (7), wherein input variables that are collected when a vehicle is at a standstill are used.

3. Method according to one of the preceding claims, characterized in that the pattern recognition continuously optimizes the gear shift drum position determination during operation.

4. Method according to one of the preceding claims, characterized in that the measured input variables are evaluated out after the measurement in a time period in the past.

5. Method according to one of the preceding claims, characterized in that the means which generate current or voltage deviations at the driving electric machine (7) is at least one pressure-exerting means which presses a ball (2) into bores (4) of the end face (3a) or of the outer circumference of the gear shift drum (3) via spring tension.