WO2024184909A1

WO2024184909A1 - A method for reducing torque ripples in a traction motor and a system thereof

Info

Publication number: WO2024184909A1
Application number: PCT/IN2024/050144
Authority: WO
Inventors: Thirunavukkarasu SENTHIL; Balaji Sreenivasan; Datta Rajaram Sagare; Chaitanya Rajendra Zanpure
Original assignee: Tvs Motor Company Limited
Priority date: 2023-03-06
Filing date: 2024-02-14
Publication date: 2024-09-12

Abstract

The present invention provides a method (200) and system (100) for reducing torque ripples in a traction motor (150). The method (200) has the steps of determining, after every rotation of the traction motor (150), a set of first rotor values (Id, Iq); comparing, the set of first rotor values (Id, Iq) with a set of first prestored rotor values to detect occurrence of torque ripple; receiving, plurality of the set of first rotor values (Id, Iq) from plurality of vehicles; generating and updating, a set of second prestored rotor values corresponding to rotor values for which torque ripple does not occur; comparing, the set of first rotor values (Id, Iq) with a set of second prestored rotor values; determining, a set of second rotor values (Id',Iq'); and performing, excitation of the traction motor (150) using the set of second rotor values (Id', Iq') and based upon a throttle demand.

Description

TITLE OF INVENTION

A Method for Reducing Torque Ripples in a Traction Motor and a System Thereof

FIELD OF THE INVENTION

[001 ] The present invention relates to control of a traction motor. More particularly, the present invention relates to method and system for reducing torque ripples in a traction motor. BACKGROUND OF THE INVENTION

[002] In conventional electrically propelled vehicles, the traction motors that create traction force for the vehicle are known to create torque ripples, especially when operating at lower speeds. There are many known causes of such ripples in torque such as cogging torque, airgap flux harmonics, mechanical imbalances and the like. These torque ripples are undesirable and have certain adverse effects. For example, it makes the ride uncomfortable at lower speeds.

Further, the torque ripple can cause speed ripple or excite driveline resonances. In the case of an electric or hybrid vehicle, this can result in vehicle oscillations which are a disturbance to the occupants.

[003] It is known in the art to employ mechanical solutions to the design of the motor, such as the winding configuration, the geometry of the stator teeth, the geometry of the rotor barrier, the skew of the rotor and the like. However, these mechanical solutions are known to affect the engine torque density. Further, passive damping techniques, such as providing the vehicle with structural reinforcements or sound attenuating materials, can be used to reduce some of the adverse effects of torque ripple and mitigate acoustic noise. However, these vibration dampening methods may address the issue of ride comfort but are costly and do not directly address the problem of torque ripple produced by the motor.

[004] Mechanical imbalances due to the aging of the electrical machines also cause torque ripple. These imbalances happen over the entire lifecycle of the electric vehicle due to wear and tear of the mechanical parts and the like.

[005] Existing closed loop control solutions for reduction of torque ripples in traction motors are prone to introducing errors in the feedback, which further produces errors in the control of traction motor, and hence, erroneous reduction of torque ripples. Further, in conventional systems, the control solution for reduction of torque ripples is hardcoded into a controller, meaning that data or parameters are fixed in a program in such a way that they cannot be altered without modifying the program. Such hardcoding allows no flexibility and dynamic fine tuning of parameters according to riding conditions.

[006] Thus, there is a need in the art for a method and system for reduction of torque ripples in a traction motor which addresses at least the aforementioned problems.

SUMMARY OF THE INVENTION

[007] In one aspect, the present invention relates to a method for reducing torque ripples in a traction motor of an electric vehicle. The method has the steps of: determining, by a motor controller, continuously after every rotation of the traction motor, a set of first rotor values, wherein the set of first rotor values (Id, Iq) include a first magnetic flux value (d) and a first rotor angle value (q). The method further has the steps of comparing, by a machine learning unit, the set of first rotor values (Id, Iq) with a set of first prestored rotor values to detect occurrence of a torque ripple; receiving, by a remote machine learning unit, a plurality of the set of first rotor values (Id, Iq) from a plurality of vehicles; generating and updating, by the remote machine learning unit (130), a set of second prestored rotor values based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles, wherein the set of second prestored rotor values correspond to rotor values for which torque ripple does not occur; comparing, by a remote machine learning unit, the set of first rotor values (Id, Iq) with the set of second prestored rotor values when an occurrence of torque ripple is detected; determining, by the remote machine learning unit, a set of second rotor values (Id’, Iq’) from the set of second prestored rotor values based on the comparison with the set of first rotor values (Id, Iq); sending, by the remote machine learning unit, the set of second rotor values (Id’, Iq’) to the motor controller; and performing, by the motor controller, excitation of the traction motor using the set of second rotor values (Id’, Iq’) and based upon a throttle demand.

[008] In an embodiment of the invention, the method further has the steps of receiving, by the machine learning unit, the plurality of the set of first rotor values (Id, Iq) from the plurality of vehicles; updating, by the machine learning unit, the set of first prestored rotor values to detect occurrence of a torque ripple based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles; predicting, by the machine learning unit, an occurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are not found in the set of first prestored rotor values; and predicting, by the machine learning unit, a non-occurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are found in the set of first prestored rotor values.

[009] In an embodiment of the invention, the method has the steps of receiving, by a vehicle telematics unit, the set of first rotor values (Id, Iq) from the machine learning unit; establishing, by the vehicle telematics unit, communication with the remote machine learning unit based on a geolocation of the vehicle; sending, by the vehicle telematics unit, the set of first rotor values (Id, Iq) to the remote machine learning unit in addition to one or more vehicle parameters; and receiving, by the vehicle telematics unit, the set of second rotor values (Id’, Iq’) from the remote machine learning unit.

[010] In a further embodiment of the invention, the method has the steps of generating and updating, by the remote machine learning unit, a truth table based on the plurality of set of first rotor values (Id, Iq) received from the plurality of vehicles, wherein the truth table includes the set of second prestored rotor values.

[011] In a further embodiment of the invention, the method has the steps of determining, by the remote machine learning unit, based on the throttle demand, that whether a torque ripple can or cannot be felt by a rider of the vehicle; changing, by the remote learning machine unit, the set of second rotor values (Id’, Iq’) to be sent to the motor controller, if it is determined that the torque ripple can be felt by the rider.

[012] In a further embodiment of the invention, determining of the set of first rotor values (Id, Iq) and excitation of the traction motor is performed using a Field Oriented Control (FOC) technique.

[013] In another aspect, the present invention relates to a system for reducing torque ripples in a traction motor of an electric vehicle. The system has a motor controller that is configured to determine continuously after every rotation of the traction motor, a set of first rotor values (Id, Iq), wherein the set of first rotor values (Id, Iq) include a first magnetic flux value (d) and a first rotor angle value (q). The system further has a machine learning unit configured to compare the set of first rotor values (Id, Iq) with a set of first prestored rotor values to detect occurrence of a torque ripple, The system has a remote machine learning unit configured to receive a plurality of the set of first rotor values (Id, Iq) from a plurality of vehicles and generate and update a set of second prestored rotor values based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles. Herein the set of second prestored rotor values corresponding to rotor values for which torque ripple does not occur. The remote machine learning unit is further configured to compare the set of first rotor values (Id, Iq) with a set of second prestored rotor values when an occurrence of torque ripple is detected. The remote machine learning unit is further configured to determine a set of second rotor values (Id’, Iq’) from the set of second prestored rotor values based on the comparison with the set of first rotor values (Id, Iq) and send the set of second rotor values (Id’, Iq’) to the motor controller. Herein, the motor controller is configured for performing excitation of the traction motor using the set of second rotor values (Id’, Iq’) and based upon a throttle demand.

[014] In a further embodiment of the invention, the machine learning unit is configured to receive the plurality of the set of first rotor values (Id, Iq) from the plurality of vehicles and update the set of first prestored rotor values to detect occurrence of a torque ripple based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles. The machine learning unit is further configured to predict an occurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are not found in the set of first prestored rotor values, and predict a non-occurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are found in the set of first prestored rotor values.

[015] In a further embodiment of the invention, the system has a vehicle telematics unit. The vehicle telematics unit is configured to receive the set of first rotor values (Id, Iq) from the machine learning unit, establish communication with the remote machine learning unit based on a geolocation of the vehicle; send the set of first rotor values (Id, Iq) to the remote machine learning unit in addition to one or more vehicle parameters; and receive the set of second rotor values from the remote machine learning unit.

[016] In a further embodiment of the invention, the remote machine learning unit is configured to generate and update a truth table based on the set of first rotor values (Id, Iq) received from a plurality of vehicles, wherein the truth table including the set of second prestored rotor values. [017] In a further embodiment of the invention, the remote machine learning unit is configured to determine that whether a torque ripple can or cannot be felt by a rider of the vehicle, based on the throttle demand; and change the set of second rotor values (Id’, Iq’) to be sent to the motor controller, if it is determined that the torque ripple can be felt by the rider.

[018] In a further embodiment of the invention, the motor controller determines of the set of first rotor values (Id, Iq) and performs excitation of the traction motor is performed using a Field Oriented Control (FOC) technique.

BRIEF DESCRIPTION OF THE DRAWINGS

[019] Reference will be made to embodiments of the invention, examples of which may be illustrated in accompanying figures. These figures are intended to be illustrative, not limiting. Although the invention is generally described in context of these embodiments, it should be understood that it is not intended to limit the scope of the invention to these particular embodiments.

Figure 1 illustrates a method for reducing torque ripples in a traction motor, in accordance with an embodiment of the present invention.

Figure 2 illustrates further method steps involved in the method for reduction of torque ripples in a traction motor, in accordance with an embodiment of the present invention. Figure 3 illustrates a system for reducing torque ripples in a traction motor, in accordance with an embodiment of the present invention.

Figure 4 illustrates implementation of a Field Operation Control loop in a motor controller, in accordance with an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

[020] The present invention relates to control of a traction motor. More particularly, the present invention relates to a method and system for reducing torque ripples in a traction motor. The system and method of the present invention are typically used in a vehicle such as a two wheeled electric or hybrid vehicle, or a three wheeled electric or hybrid vehicle, or a four wheeled electric or hybrid vehicle, or other multi-wheeled electric or hybrid vehicles as required. [021] Figure 1 illustrates method steps involved in a method 200 for reducing torque ripples in a traction motor 150 of an electric vehicle 10. As illustrated, at step 202, a set of first rotor values (Id, Iq) are determined continuously after every rotation of the traction motor 150 by a motor controller 110. As an example, the traction motor 150 is a three-phase PMSM I BLDC based setup, which requires Field Operation Control and Rotating Magnetic Field to operate in forward and reverse directions. Herein, the set of first rotor values (Id, Iq) include a first magnetic flux value (d) and a first rotor angle value (q). Further, Id refers to a direct axis along which torque produced by the traction motor is controlled and Iq refers to the quadratic axis along which speed of the traction motor is controlled.

[022] Thereafter, at step 204, the set of first rotor values (Id, Iq) is compared with a set of first prestored rotor values by a machine learning unit 120 to detect or predict an occurrence of a torque ripple. When occurrence of a torque ripple is detected at step 204, the method 200 moves to step 206 wherein, a plurality of the set of first rotor values (Id, Iq) are received by a remote machine learning unit 130 from a plurality of vehicles. Thereafter, at step 208, a set of second prestored rotor values are generated and updated based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles. Herein, the set of second prestored rotor values correspond to rotor values for which torque ripple does not occur. The set of second prestored rotor values are generated and updated based on inputs received from a plurality of vehicles by the remote machine learning unit 130. Thus, in effect, a crowdsourced repository of rotor values for which torque ripples does not occur is generated and dynamically updated, which is then referred to as the set of second prestored rotor values.

[023] Thereafter, at step 210, the set of first rotor values (Id, Iq) are compared with the set of second prestored rotor values by the remote machine learning unit 130. Herein, as explained hereinbefore, the set of second prestored rotor values correspond to rotor values for which torque ripple does not occur.

[024] Thereafter, at step 212, a set of second rotor values (Id’, Iq’) is determined from the set of second prestored rotor values by the remote machine learning unit 130 based on the comparison with the set of first rotor values (Id, Iq). Since the set of second rotor values (Id’, Iq’) belongs to the second prestored rotor values, torque ripple does not occur for the set of second rotor values (Id’, Iq’). Thereby, a set of second rotor values (Id’, Iq’) which are comparable with the set of first rotor values (Id, Iq) and for which torque ripple does not occur, is obtained.

[025] Thereafter, at step 214, the set of second rotor values (Id’, Iq’) are sent by the remote machine learning unit 130 to the motor controller 110. Thereafter at step 216, excitation of the traction motor 150 is performed by the motor controller 110 using the set of second rotor values (Id’, lq’) and based upon a throttle demand. Thus, since the traction motor 150 is being excited corresponding to the set of second rotor values (Id’, lq’) for which torque ripple does not occur, torque ripples are reduced in the traction motor 150. In an embodiment, determining of the set of first rotor values (Id, lq) and excitation of the traction motor 150 is performed using a Field Oriented Control (FOC) technique.

[026] Figure 2 illustrates further method steps 300 involved in the method 200, in accordance with an embodiment of the present invention. As illustrated, at step 302, the set of first rotor values (Id, lq) are determined continuously after every rotation of the traction motor 150. Thereafter, at step 304, the set of first rotor values (Id, lq) are fed to a Supervised Machine Learning (SML) module in the machine learning unit 120. Thereafter, the set of first prestored rotor values is updated by the machine learning unit 120 to detect occurrence of a torque ripple based on the plurality of set of first rotor values (Id, lq) from the plurality of vehicles. Herein, the plurality of set of first rotor values (Id. lq) may be received by the machine learning unit 120 from the plurality of vehicles via a telematics system or through the remote machine learning server or through an intermediate external device any other means.

[027] At step 306, it is predicted that whether a torque ripple occurs at the determined set of first rotor values (Id, lq). Herein, at step 306, an occurrence of torque ripple is predicted by the machine learning unit 120 when on comparison, the set of first rotor values (Id, lq) are not found in the set of first prestored rotor values. Similarly, a non occurrence of a torque ripple is predicted by the machine learning unit 120 when on comparison, the set of first rotor values (Id, lq) are found in the set of first prestored values. The continuous updating of the first set of prestored rotor values based on rotor values from the plurality of vehicles increases the accuracy in the prediction of occurrence of torque ripple. [028] At step 308, a communication is established between a vehicle telematics unit 140 and the remote machine learning unit 130 based on the geolocation of the vehicle. In an embodiment, the remote machine learning unit 130 is a server to which the vehicle telematics unit 140 can connect. Since more than one remote machine learning units 130 are present, it is important for the vehicle telematics unit 140 to connect with the nearest remote machine learning unit 130 for reliable connection throughout different locations. The establishment of communication between the vehicle telematics unit 140 and the remote machine learning unit 130 based on the geolocation of the vehicle ensures that the vehicle telematics unit 140 is connected with the remote machine learning unit 130 present in the vicinity of the geolocation of the vehicle 10. The first set of rotor values (Id, Iq) are received by the vehicle telematics unit 140 and at step 310, the first set of rotor values (Id, Iq) along with one or more vehicle parameters are sent by the vehicle telematics unit 140 to the remote machine learning unit 130. [029] At step 312, a truth table is generated and updated by the remote machine learning unit 130 based on the set of first rotor values (Id, Iq) received from a plurality of vehicles, wherein the truth table includes the set of second prestored rotor values. An unsupervised or supervised machine learning module is utilised for generation and updating of the truth table, wherein the machine learning module receives rotor values from the plurality of vehicles and generates and updated rotor values for which torque ripple does not occur. Thereafter at step 314, the truth table is sent to the vehicle 10 via the vehicle telematics unit 140. The truth table is in essence, the crowdsourced set of second prestored rotor values along with the determined set of second rotor values (Id’, Iq’) based on the set of first rotor values (Id, Iq).

[030] Reference is made back to step 306, wherein, if an occurrence of a torque ripple is predicted, the method moves to step 316, wherein the control of the traction motor 150 is executed using the set of second rotor values (Id’, lq’) from the remote machine learning unit 130, obtained from step 314. Thereafter, the method moves to step 318 where the excitation of the traction motor 150 is done based on throttle demand and the set of second rotor values (Id’, lq’). After step 310, the method moves back to step 304 for continuous operation.

[031] Reference is made back to step 306, wherein if an occurrence of a torque ripple is not predicted, the method moves to step 320, wherein the excitation of the traction motor 150 is done based on the throttle demand. Thereafter for continuous operation, at step 322 and step 324, the first set of rotor values are forecasted by the machine learning unit 120 and the occurrence of a torque ripple is predicted through the SML module. Thereafter, if occurrence of a torque ripple is predicted, the method moves to step 316 for continuous operation. Similarly, if non-occurrence of a torque ripple is predicted, the method moves to step 326 for excitation of traction motor 150 according to the throttle demand, after which the method moves to step 204 for continuous operation.

[032] In an embodiment, the method 200 further has the step of determining that whether a torque ripple can or cannot be felt by a rider of the vehicle 10 based on the throttle demand by the remote machine learning unit 130. It is important to check that whether even after excitation of traction motor 150 using the set of second rotor value (Id’, lq’), is the torque ripple being felt by the rider. The throttle demand of the vehicle 10 is an indicator of whether the torque ripple is being felt by the rider. If the torque ripple is being felt by the rider, the method 200 further has the step of changing the set of second rotor values (Id’, lq’) by the remote machine learning unit 130 to be sent to the motor controller 110. This ensures that if the torque ripples are being felt even after excitation based on the set of second rotor values (Id’, lq’), the set of second rotor values (Id’, lq’) are changed, thus ensuring dynamic reduction in the torque ripple felt by the rider.

[033] In another aspect, the present invention relates to a system 100 for reducing torque ripples in a traction motor 150 of an electric vehicle 10. Figure 3 illustrates the system 100 for reducing torque ripples in the traction motor 150. As illustrated, the system 100 comprises a motor controller 110. The motor controller 110 is not a general purpose computer unit. The motor controller 110 includes suitable logic, circuitry, interfaces and/or codes that are configured to generate a rotating magnetic field for Field Operation Control of the traction motor 250. The motor controller 110 is configured to determine continuously after every rotation of the traction motor 150, a set of first rotor values (Id, lq). The set of first rotor values (Id, lq) include a first magnetic flux value (d) and a first rotor angle value (q).

[034] The system 100 further has a machine learning unit 120. The machine learning unit 120 is not a general-purpose computer unit. The machine learning unit 120 includes suitable logic, circuitry, interfaces and/or codes that are configured to compare the set of first rotor values (Id, lq) with a set of first prestored rotor values to detect occurrence of a torque ripple. The machine learning unit 120 functions on a Supervised Machine Learning (SML) Model for estimating and predicting the torque ripples. The SML model is capable of functioning on any one of, but not limited to the following techniques: Logistic Regression, Decision tree, Random forest, Naive Bayes, KNN, or XG Boost.

[035] The system 100 further has a remote machine learning unit 130. The remote machine learning unit 130 is not a general-purpose computer unit. The remote machine learning unit 130 includes suitable logic, circuitry, interfaces and/or codes that are configured to receive a plurality of the set of first rotor values (Id, lq) from a plurality of vehicles, and generate and update a set of second prestored rotor values based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles. Herein, the set of second prestored rotor values correspond to rotor values for which torque ripple does not occur. The set of second prestored rotor values are generated and updated based on inputs received from a plurality of vehicles by the remote machine learning unit 130. Thus, in effect, a crowdsourced repository of rotor values for which torque ripples does not occur is generated and dynamically updated, which is then referred to as the set of second prestored rotor values.

[036] The remote machine learning unit 130 is further configured to compare the set of first rotor values (Id, Iq) with a set of second prestored rotor values when an occurrence of torque ripple is detected. Herein, the set of second prestored rotor values correspond to rotor values for which torque ripple does not occur. The remote machine learning unit 130 further determine a set of second rotor values (Id’, Iq’) from the set of second prestored rotor values based on the comparison with the set of first rotor values (Id, Iq) and send the set of second rotor values (Id’, Iq’) to the motor controller 110.

[037] Thereafter, the motor controller 110 is configured for performing excitation of the traction motor 150 using the set of second rotor values Id’, Iq’ and based upon a throttle demand. The remote machine learning unit 130 functions on an Unsupervised Machine Learning Model (USML) or Supervised Machine Learning (SML) Model for obtaining the set of second rotor values (Id’, Iq’). The USML or SML model is capable of functioning on any one of, but not limited to the following techniques: Logistic Regression, Decision tree, Random forest, Naive Bayes, KNN, or XG Boost.

[038] Thus, since the motor controller 110 excites the traction motor 150 corresponding to the set of second rotor values (Id’, Iq’) for which torque ripple does not occur, torque ripples are reduced in the traction motor 150. In an embodiment, the motor controller 110 determines of the set of first rotor values (Id, Iq) and excitation of the traction motor 150 is performed using a Field Oriented Control (FOC) technique.

[039] In an embodiment, the machine learning unit 120 is configured to receive the plurality of the set of first rotor values (Id, Iq) from the plurality of vehicles and update the set of first prestored rotor values to detect occurrence of a torque ripple based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles. Herein, the plurality of set of first rotor values (Id. Iq) may be received by the machine learning unit 120 from the plurality of vehicles via a telematics system or through the remote machine learning server or through an intermediate external device any other means.

[040] The machine learning unit 120 is further configured to predict an occurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are not found in the set of first prestored rotor values. Similarly, the machine learning unit 120 is configured to predict a nonoccurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are found in the set of first prestored rotor values. The continuous updating of the first set of prestored rotor values based on rotor values from the plurality of vehicles increases the accuracy in the prediction of occurrence of torque ripple.

[041 ] As further illustrated in Figure 3, in an embodiment, the system 100 comprises a vehicle telematics unit 140. The vehicle telematics unit 140 is configured to receive the set of first rotor values (Id, Iq) from the machine learning unit 120. The vehicle telematics unit 140 establishes communication with the remote machine learning unit 130 based on a geolocation of the vehicle 10 and sends the set of first rotor values (Id, Iq) to the remote machine learning unit 130 in addition to one or more vehicle parameters. Thereafter, the vehicle telematics unit 140 receives the set of second rotor values from the remote machine learning unit 130. The establishment of communication between the vehicle telematics unit 140 and the remote machine learning unit 130 based on the geolocation of the vehicle ensures that the vehicle telematics unit 140 is connected with the remote machine learning unit 130 present in the vicinity of the geolocation of the vehicle 10.

[042] Further, in an embodiment, the remote machine learning unit 130 is generate and update a truth table based on the set of first rotor values (Id, Iq) received from a plurality of vehicles, wherein the truth table including the set of second prestored rotor values. The truth table is in essence, the crowdsourced set of second prestored rotor values along with the determined set of second rotor values (Id’, Iq’) based on the set of first rotor values (Id, Iq).

[043] Further, the remote machine learning unit 130 is configured to determine that whether a torque ripple can or cannot be felt by a rider of the vehicle (10), based on the throttle demand and change the set of second rotor values (Id’, Iq’) to be sent to the motor controller 110, if it is determined that the torque ripple can be felt by the rider.

[044] As illustrated in Figure 4, the FOC (Field Oriented Control) loop is implemented in software of the motor controller 110. In FOC, 2-dimensional parameters magnetic flux ‘d’ and rotor angle ‘q’ from PI Controllers is converted into 3-dimensional parameters using Inverse Clarke & Park transformation 204. The 3-dimensional parameters are thereafter converted into 3 phase currents iu, iv, iw though a Pulse Width Modulation (PWM) Generator. After the traction motor 150 is rotated once for the next rotation, flux and the current rotor angle is fed back after Clarke 208 and Park transformation 210 again to the PI Controllers. Before the ‘d’ and ‘q’ coordinates are fed back to PI Controllers, in the FOC loop, in the feedback of ‘d’ & ‘q’ a machine learning technique is introduced such that error’s detected in the magnetic flux and the rotor angle is learnt and eliminated cycle by cycle. Accordingly, the same are fed to Inverse Clarke & Park conversion such that PWM pulses generated is used for switching IGBT’s. Id, Iq mapping is compared in any SML models and a suitable value will be retrieved and fed back to the system 100 to reduce I eliminate the torque ripples.

[045] Advantageously, the present invention provides a method and a system for reduction of torque ripples in a traction motor. Utilizing of the SML models after the Id, Iq values are generated from the motor controller allows SML of the machine learning unit, and thereafter SML/USML of the remote machine learning unit to obtain the set of second rotor values based on which the machine controller excites the traction motor. The set of second rotor values being obtained from the dynamically updated and generated set of second prestored rotor values ensures that the rotor values are accurate and are dynamically updated based on larger and updated data set. The dynamic updating of the set of first prestored rotor values also enhances the accuracy in the prediction of the occurrence of a torque ripple. Further, the torque ripples are differentiated by means of whether ripple can be felt by the customer or not and if so, changing the excitation, which provides for a dynamic manner of reducing the torque ripples which makes the ride more comfortable.

[046] Further, the present invention provides for the establishment of communication between the vehicle telematics unit and the remote machine learning unit based on the geolocation of the vehicle, which ensures that the vehicle telematics unit is connected with the remote machine learning unit present in the vicinity of the geolocation of the vehicle. This ensures a stable and reliable connection between the vehicle and the remote machine learning unit irrespective of the location of the vehicle, ensuring consistent reduction in torque ripples. [047] Further, the present system and method allows for reduction in torque ripples which result in improved performance of the traction motor of the Electric Vehicle. Further, the present invention allows for a robust technique which is capable of being fit on to or used for any vehicle with a Field Oriented Control Loop. Reduction in torque ripples further leads to improvement in handling of the vehicle, and thus comfort for the rider.

[048] In addition, as opposed to conventional systems, in the present invention, due to provision of a machine learning unit and a remote machine learning unit, errors detected in the magnetic flux and the rotor angle is learnt and eliminated cycle by cycle and accordingly fed to Inverse Clarke & Park conversion such that PWM pulses generated is used for controlling the motor. Id, Iq mapping is compared in SML models and suitable values are retrieved and fed back to the system to reduce I eliminate the torque ripples. By implementing the Logistic Regression or any other Supervised Machine Learning algorithm before the actual Id, Iq calculation in FOC, errors in the feedback of flux ’d’ and rotor angle ‘q’ can be eliminated. Furthermore, in the present invention, since the set of second prestored rotor values is generated and updated dynamically, the set of rotor values for excitation of traction motor for which torque ripple does not occur is obtained dynamically. Therefore, torque ripple caused by mechanical imbalances due to the aging of the electrical machines is also addressed and avoided.

[049] In light of the abovementioned advantages and the technical advancements provided by the disclosed method and system, the claimed steps as discussed above are not routine, conventional, or well understood in the art, as the claimed steps enable the following solutions to the existing problems in conventional technologies. Further, the claimed steps clearly bring an improvement in the functioning of the device itself as the claimed steps provide a technical solution to a technical problem.

[050] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer-readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.

[051] While the present invention has been described with respect to certain embodiments, it will be apparent to those skilled in the art that various changes and modification may be made without departing from the scope of the invention as defined in the following claims.

List of Reference Numerals

10: Vehicle

100: System for reduction of torque ripples

110: Motor Controller

120: Machine Learning Unit

130: Remote Machine Learning Unit

140: Vehicle Telematics Unit 150: Traction Motor

200: Method for reduction of torque ripples

Claims

CLAIMS:

1. A method (200) for reducing torque ripples in a traction motor (150) of an electric vehicle (10), the method (200) comprising the steps of: determining, by a motor controller (110), continuously after every rotation of the traction motor (150), a set of first rotor values (Id, Iq), the set of first rotor values (Id, Iq) including a first magnetic flux value (d) and a first rotor angle value (q); comparing, by a machine learning unit (120), the set of first rotor values (Id, Iq) with a set of first prestored rotor values to detect occurrence of a torque ripple; receiving, by a remote machine learning unit (130), a plurality of the set of first rotor values (Id, Iq) from a plurality of vehicles; generating and updating, by the remote machine learning unit (130), a set of second prestored rotor values based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles, the set of second prestored rotor values corresponding to rotor values for which torque ripple does not occur; comparing, by the remote machine learning unit (130), the set of first rotor values (Id, Iq) with the set of second prestored rotor values when an occurrence of torque ripple is detected; determining, by the remote machine learning unit (130), a set of second rotor values (Id’, Iq’) from the set of second prestored rotor values based on the comparison with the set of first rotor values (Id, Iq); sending, by the remote machine learning unit (130), the set of second rotor values (Id’, Iq’) to the motor controller (110); and performing, by the motor controller (110), excitation of the traction motor (150) using the set of second rotor values (Id’, lq’) and based upon a throttle demand.

2. The method (200) as claimed in claim 1 , comprising the steps of: receiving, by the machine learning unit (120), the plurality of the set of first rotor values (Id, lq) from the plurality of vehicles; updating, by the machine learning unit (120), the set of first prestored rotor values to detect occurrence of a torque ripple based on the plurality of set of first rotor values (Id, lq) from the plurality of vehicles; predicting, by the machine learning unit (120), an occurrence of torque ripple when on comparison, the set of first rotor values (Id, lq) are not found in the set of first prestored rotor values; and predicting, by the machine learning unit (120), a non-occurrence of torque ripple when on comparison, the set of first rotor values (Id, lq) are found in the set of first prestored rotor values.

3. The method (200) as claimed in claim 1 , comprising the steps of: receiving, by a vehicle telematics unit (140), the set of first rotor values (Id, lq) from the machine learning unit (120); establishing, by the vehicle telematics unit (140), communication with the remote machine learning unit (130) based on a geolocation of the vehicle (10); sending, by the vehicle telematics unit (140), the set of first rotor values (Id, lq) to the remote machine learning unit (130) in addition to one or more vehicle parameters; and 1 receiving, by the vehicle telematics unit (140), the set of second rotor values (Id’, lq’) from the remote machine learning unit (130).

4. The method (200) as claimed in claim 1 , comprising the step of: generating and updating, by the remote machine learning unit (130), a truth table based on the plurality of set of first rotor values (Id, lq) received from the plurality of vehicles, wherein the truth table includes the set of second prestored rotor values.

5. The method (200) as claimed in claim 4, comprising the step of: determining, by the remote machine learning unit (130), based on the throttle demand, that whether a torque ripple can or cannot be felt by a rider of the vehicle (10); changing, by the remote learning machine unit (130), the set of second rotor values (Id’, lq’) to be sent to the motor controller (110), if it is determined that the torque ripple can be felt by the rider.

6. The method (200) as claimed in claim 1 , wherein determining of the set of first rotor values (Id, lq) and excitation of the traction motor (150) is performed using a Field Oriented Control (FOC) technique.

7. A system (100) for reducing torque ripples in a traction motor (150) of an electric vehicle (10), the system (100) comprising: a motor controller (110), the motor controller (110) being configured to: determine continuously after every rotation of the traction motor (150), a set of first rotor values (Id, Iq), the set of first rotor values (Id, Iq) including a first magnetic flux value (d) and a first rotor angle value (q); a machine learning unit (120), the machine learning unit (120) being configured to: compare the set of first rotor values (Id, Iq) with a set of first prestored rotor values to detect occurrence of a torque ripple; and a remote machine learning unit (130), the remote machine learning unit (130) configured to: receive a plurality of the set of first rotor values (Id, Iq) from a plurality of vehicles; generate and update a set of second prestored rotor values based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles, the set of second prestored rotor values corresponding to rotor values for which torque ripple does not occur; compare the set of first rotor values (Id, Iq) with a set of second prestored rotor values when an occurrence of torque ripple is detected; determine a set of second rotor values (Id’, Iq’) from the set of second prestored rotor values based on the comparison with the set of first rotor values (Id, Iq); send the set of second rotor values (Id’, Iq’) to the motor controller (110); wherein the motor controller (110) is configured for performing excitation of the traction motor (150) using the set of second rotor values (Id’, Iq’) and based upon a throttle demand.

8. The system (100) as claimed in claim 7, wherein the machine learning unit (120) is configured to: receive the plurality of the set of first rotor values (Id, Iq) from the plurality of vehicles; update the set of first prestored rotor values to detect occurrence of a torque ripple based on the plurality of set of first rotor values (Id, Iq) from the plurality of vehicles; predict an occurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are not found in the set of first prestored rotor values; and predict a non-occurrence of torque ripple when on comparison, the set of first rotor values (Id, Iq) are found in the set of first prestored rotor values.

9. The system (100) as claimed in claim 7, comprising a vehicle telematics unit (140), the vehicle telematics unit (140) being configured to: receive the set of first rotor values (Id, Iq) from the machine learning unit (120); establish communication with the remote machine learning unit (130) based on a geolocation of the vehicle (10); send the set of first rotor values (Id, Iq) to the remote machine learning unit (130) in addition to one or more vehicle parameters; and receive the set of second rotor values from the remote machine learning unit.

10. The system (100) as claimed in claim 7, wherein the remote machine learning unit (130) is configured to: generate and update a truth table based on the set of first rotor values (Id, Iq) received from a plurality of vehicles, wherein the truth table including the set of second prestored rotor values.

11. The system (100) as claimed in claim 10, wherein the remote machine learning unit (130) is configured to: determine that whether a torque ripple can or cannot be felt by a rider of the vehicle (10), based on the throttle demand; change the set of second rotor values (Id’, lq’) to be sent to the motor controller (110), if it is determined that the torque ripple can be felt by the rider.

12. The system (100) as claimed in claim 7, wherein the motor controller (110) determines of the set of first rotor values (Id, lq) and performs excitation of the traction motor (150) is performed using a Field Oriented Control (FOC) technique.