Claims:We claim:
1. A controller (110) to estimate load parameter (120) of an Electric Vehicle (EV) (100), said EV (100) equipped with components comprising at least one rotary electric machine (104) to drive said EV (100), at least one battery pack (102) to supply electrical energy to said rotary electric machine (104) and optionally to accessories of said EV (100), said controller (110) and control units of said components are connected through a Controller Area Network (CAN) (106) of said EV (100), said controller (110) configured to,
measure characteristic parameters from said CAN (106) during a trip, characterized in that,
detect at least one quasi-steady state condition of said EV (100) based on a primary parameter within said measured characteristic parameter, and
process secondary parameter within said measured characteristic parameters and an unloaded weight of said EV (100) through a load model (112) to estimate said load parameter (120), said load parameter (120) is at least one selected from a group comprising an estimated weight of said EV (100) and a road gradient.

2. The controller (110) as claimed in claim 1, wherein said primary parameter is at least one selected from a group comprising a vehicle speed, a vehicle acceleration and a throttle position, and said secondary parameter is at least one selected from a group comprising a phase current drawn by said rotary electric machine (104), battery current supplied from said at least one battery pack (102) to said rotary electric machine (104), a torque applied by said rotary electric machine (104), a power (electric/mechanical) consumed by said rotary electric machine (104), a battery power and combination thereof, wherein said secondary parameters is further selectable from a battery voltage, and an efficiency of said rotary electric machine (104).

3. The controller (110) as claimed in claim 1, wherein said load parameters (120) are used for at least one selected from a group comprising estimation of range of said EV (100) for a current trip, an estimation of abuse of said EV (100), an estimation of aging of parts and components of said EV (100), enablement of energy consumption based payment models, selection of optimal operating points for extracting higher efficiency out of said EV (100) based on estimated weight, enablement of better acceleration and drive quality while driving over gradients, enhancement of hill hold and hill assist features with estimated gradient, modulation and distribution of optimal brake force based on estimated weight and gradient, adjustment of suspension based on estimated weight, adjustment of tire pressure based on estimated weight, classification of payload of said EV (100), torque co-ordination / modulation between wheels and the like.

4. The controller (110) as claimed in claim 1, wherein said load model (112) is developed and pre-trained using any one but not limited to a Gaussian Process data model, a linear model, a polynomial model, a regression model, parameter estimation, curve fitting, Least Mean Square Error (LMSE), Recursive least square approximation , a Sum Squared Error (SSE), an absolute error, Machine Learning (ML) model, Deep Learning (DL) model, self-learning models and a combination thereof.

5. The controller (100) as claimed in claim 1, wherein while said road gradient is determined to be zero, said controller (110) configured to
estimate weight of said EV (100) from said load model (112) during said at least one quasi-steady state condition in a trip, and
classify said trip into single rider or two rider or over weight condition based on said estimated weight.

6. A method for estimating load parameter (120) of an Electric Vehicle (EV) (100), said EV (100) equipped with components comprising at least one rotary electric machine (104) to drive said EV (100) and at least one battery pack (102) to supply electrical energy to said rotary electric machine (104) and accessories of said EV (100), said controller (110) and control unit of said components are connected through a Controller Area Network (CAN) (106) of said EV (100), said method comprising the steps of:
measuring characteristic parameters from said CAN (106) during a trip, characterized by,
detecting at least one quasi-steady state condition of said EV (100) based on a primary parameter within said measured characteristic parameters, and
processing secondary parameter within said measured characteristic parameters and an unloaded weight of said EV (100) through a load model (112) for estimating said load parameter (120), said load parameter (120) is at least one selected from a group comprising an estimated weight of said EV (100) and a road gradient.

7. The method as claimed in claim 6, wherein said primary parameter is at least one selected from a group comprising a vehicle speed, a vehicle acceleration and a throttle position, and said secondary parameter is at least one selected from a group comprising a phase current drawn by said rotary electric machine (104), battery current supplied from said at least one battery pack (102) to said rotary electric machine (104), a torque applied by said rotary electric machine (104), a power (electric/mechanical) consumed by said rotary electric machine (104), a battery power and combination thereof, wherein said secondary parameters is further selectable from a battery voltage, and an efficiency of said rotary electric machine (104).

8. The method as claimed in claim 6, wherein said load parameters (120) are used for at least one selected from a group comprising estimating the range of the EV (100) for the current trip, estimating abuse of the EV (100), estimating aging of parts and components of the EV (100), enabling energy consumption based payment models, selecting optimal operating points for extracting higher efficiency out of the EV (100) based on estimated weight, enabling better acceleration and drive quality while driving over gradients, enhancing hill hold and hill assist features with estimated road gradient, modulating and distributing optimal brake force based on estimated weight and road gradient, adjusting suspension based on estimated weight, adjusting tire pressure based on estimated weight, classifying payload of said EV (100), torque co-ordination / modulation between wheels and the like.

9. The method as claimed in claim 6, wherein said load model (112) is developed and pre-trained using any one but not limited to a Gaussian Process data model, a linear model, a polynomial model, a regression model, parameter estimation, curve fitting, Least Mean Square Error (LMSE), Recursive least square approximation , a Sum Squared Error (SSE), an absolute error, Machine Learning (ML) models, Deep Learning (DL) models, self-learning models and a combination thereof.

10. The method as claimed in claim 6, wherein while said gradient is zero, said method comprises
estimating weight of said EV (100) from said load model (112) during said at least one quasi-steady state condition in a trip, and
classifying said trip into single rider or two rider or over weight condition based on the estimated weight.
, Description:Complete Specification:
The following specification describes and ascertains the nature of this invention and the manner in which it is to be performed:

Field of the invention:
[0001] The present invention relates to a controller to estimate load parameter of an Electric Vehicle (EV) and method thereof.

Background of the invention:
[0002] Existing methods of vehicle weight detection employ the use of sensors to detect additional parameters adding additional cost, hardware units and complexity and have not been applicated for two wheeler electric vehicles or other vehicles.

[0003] A patent literature US2016355189 discloses a systems and methods for vehicle load detection and response. A torque-speed curve or data of load that is used as a standard to determine an external condition in which an electric vehicle is operating such as incline or no incline, head wind or no headwind, high temperature or low temperature. The system compares samples of actual torque-speed of load data to the standard. Based on the comparison, the system determines the external condition (going up a hill, traveling into a headwind, operating at high temperature) or an abnormal operation of the vehicle powertrain, for example, low tire pressure, elevated friction, wheels out of alignment. Based on the determination, the system takes an action to govern a maximum torque output of the motor to control temperature of the vehicle battery; to raise a wind deflector; to govern maximum speed of the vehicle to reduce danger resulting from low tire pressure, elevated powertrain friction or out of alignment wheels; or to initiate an indication of abnormal conditions.

Brief description of the accompanying drawings:
[0004] An embodiment of the disclosure is described with reference to the following accompanying drawing,
[0005] Fig. 1 illustrates a block diagram of a controller to estimate load parameter of an Electric Vehicle (EV), according to an embodiment of the present invention, and
[0006] Fig. 2 illustrates a method for estimating load parameter of the EV, according to the present invention.

Detailed description of the embodiments:
[0007] Fig. 1 illustrates a block diagram of a controller to estimate load parameter of an Electric Vehicle (EV), according to an embodiment of the present invention. The EV 100 equipped with components comprising at least one rotary electric machine 104 to drive the EV 100, at least one battery pack 102 to supply electrical energy to the rotary electric machine 104 and optionally to accessories of the EV 100. The controller 110 and control units of the components are connected through a Controller Area Network (CAN) 106 bus (or other similar networking structure) of the EV 100. The controller 110 configured to measure characteristic parameters of the EV 100 from the CAN 106 during a trip, characterized in that, the controller 110 configured to detect at least one quasi-steady state condition of the EV 100 based on a primary parameter within the measured characteristic parameter, and process secondary parameter within the measured characteristic parameters and an unloaded weight of the EV 100 through a load model 112 to estimate the load parameter 120. The load parameter 120 is at least one selected from a group comprising an estimated weight of the EV 100 and a road gradient.

[0008] The estimated weight of the EV 100 corresponds to the weight of the EV 100 during driving conditions or a trip or a drive cycle. The load parameter 120 are those parameters which influences the weight of the EV 100 during the driving. In other words, the load parameters 120 define the load which acts on the EV 100 due to payload on the EV 100 and/or due to terrain over which the EV 100 is driven. Further, the quasi-steady state (or steady state) is a condition which comprises but limited to driving at a speed value within a certain specified range for a duration of time, or even transient events like accelerating from 0% throttle to a certain range of throttle (for example from 0 to 30% throttle) as a ramp. The quasi-steady state is a window to initiate estimations using the controller 110.

[0009] In accordance to the present invention, the controller 110 comprises but not limited to a memory element such as Random Access Memory (RAM) and/or Read Only Memory (ROM), Analog-to-Digital Converter (ADC) and a Digital-to-Analog Convertor (DAC), clocks, timers and at least one processor (capable of implementing machine learning) connected with each other and to other components through communication bus channels. The memory element is pre-stored with logics or instructions or programs or applications or load models 112, modules and threshold values, reference values and conditions which is/are accessed by the processor as per the defined routines. The internal components of the controller 110 are not explained for being state of the art, and the same must not be understood in a limiting manner. The controller 110 may also comprise communication units to communicate with the cloud server through wireless or wired means such as Global System for Mobile Communications (GSM), 3G, 4G, 5G, Wi-Fi, Bluetooth, Ethernet, serial networks, and the like.

[0010] In accordance to an embodiment of the present invention, the controller 110 is at least one of an internal control unit and an external control unit. The internal control unit is at least one of a Vehicle Control Unit (VCU), a Motor Control Unit (MCU) and Battery Control Unit (BCU). The external control unit is a control unit externally interfaced with the controller 110 through wired or wireless means such as Universal Serial Bus (USB), type-C, Bluetooth™, Wi-Fi etc. The external control unit is selectable from at least one of a smartphone, a portable computer, a wearable device, a cloud, and the like. In an embodiment, the internal control unit estimates the load parameters 120. In an alternative embodiment, the external control unit interfaced with the internal control unit estimates the load parameters 120. In yet another alternative, both the internal control unit and the external control unit together shares the processing load and estimates the load parameter 120 of the EV 100.

[0011] According to the present invention, the controller 110 is implementable for EV 100 comprising two-wheelers, three-wheelers such as auto-rickshaws, a four wheeler such as cars, and multi-wheel vehicles.

[0012] The characteristic parameters are combination of primary parameters and secondary parameters, which in turn are combination of measured variables and derived variables of the EV 100. For example, speed is measured, and acceleration is derived from the speed. Only one example is provided and the same is relatable with other characteristic parameters as well as known in the art. Further, the measured variables are preferred over the derived, but both are usable to enhance the estimation accuracy of the load model 112.

[0013] According to an embodiment of the present invention, the primary parameter is at least one selected from a group comprising a vehicle speed, a vehicle acceleration, and a throttle position. The secondary parameter is at least selected from a phase current drawn by the rotary electric machine 104, battery current supplied from the battery pack 102 to the rotary electric machine 104, a torque applied by the rotary electric machine 104, a power consumed by the rotary electric machine 104 (electric/mechanical), battery power, and combination thereof. The secondary parameter is further selectable from battery voltage, and efficiency of the rotary electric machine 104. Specifically, the primary parameter is used for the determination of the quasi-steady state condition, whereas the secondary parameter is used in the load model 112. The characteristic parameters are measured using respective sensors already available in the EV 100, such as speed sensor 108, throttle position sensor 116, voltage and current sensors 118 and the like. Some of the characteristic parameters are either estimated using the available sensor values or modeled as known in the art.

[0014] According to an embodiment of the present invention, the load parameters 120, once determined, are used by the controller 110 for at least one selected from a group comprising estimation of range of the EV 100 for a current trip, an estimation of abuse of the EV 100, an estimation of aging of parts and components of the EV 100, enablement of energy consumption based payment models, selection of optimal operating points for extracting higher efficiency out of the EV 100 based on estimated weight, enablement of better acceleration and drive quality while driving over gradients, enhancement of hill hold and hill assist features with estimated gradient, modulation and distribution of optimal brake force based on estimated weight and gradient, adjustment of suspension based on estimated weight, adjustment of tire pressure based on estimated weight, classification of payload on the EV 100, a torque co-ordination/ modulation between wheels and the like.

[0015] Few of the usage mentioned above are self-explanatory and few of them are described with more details now. The energy consumption based payment models comprises charging by actual consumed energy and not time or distance travelled. The actual energy consumed is influenced by the payload on the EV 100 and the road gradient. The selection of optimal operating points comprises for example, adapting selective torque maps for overloaded EVs 100 to ensure better mileage and ride quality. The modulation and distribution of optimal brake force comprises for example, adjusting braking force required to stop the momentum of much heavier EVs 100 which is different from a single driver use case, and therefore adjustment of the brake power for different weights and gradients is done to enhance drivability. The torque co-ordination/ modulation between wheels corresponds to scenarios such as when load on EV 100 is less and if there are more than two rotary electric machine 104 on the EV 100, the torque is distributed or modulated between the two rotary electric machines 104 in such a way that each works at most efficiency region, instead of both operating at same torque and inefficient point.

[0016] In accordance to an embodiment of the present invention, the load model 112 is developed and pre-trained using any one but not limited to a Gaussian Process data model, a linear model, a polynomial model, a regression model, parameter estimation, curve fitting, Least Mean Square Error (LMSE), Recursive least square approximation, a Sum Squared Error (SSE), an absolute error, Machine Learning (ML) models, Deep Learning (DL) models, self-learning models and a combination thereof. The ML model is at least one selected from a group comprising logistic regression, linear regression, naive bayes, k-nearest neighbors, decision tree, random forest, support vector machines, k-means clustering, etc. Similarly, the DL model is at least one selected from a group comprising, multilayer perceptron, convoluted neural networks, recursive neural networks, etc. These are just examples and the present invention are not limited to the same.

[0017] In accordance to an embodiment of the present invention, while the road gradient is determined to be zero, the controller 110 configured to estimate weight of the EV 100, specifically a two-wheeler, from the load model 112 during the at least one quasi-steady state condition, and classify the trip into single rider or two rider based on the estimated weight. The classification is also applicable for other types of vehicles as mentioned before.

[0018] According to the present invention, in any EV 100 such as two-wheeler, in the presence of additional load or while driving on a gradient (uphill slope), the characteristics parameters such as phase current, vehicle torque, acceleration, power, battery current, etc. varies. There exists variables of the EV 100 that exhibits a distinguishable difference in value for cases comprising a single rider load on the EV 100, two riders/ heavy rider/ loaded with luggage, the EV 100 is being driven over a gradient / slope. The values of characteristic parameters such as acceleration, power, current, etc. are already available in the vehicle CAN bus 106. The load model 112 utilizes these characteristics parameters to determine load parameter and thereby classify whether there is a single rider payload, two rider payload / overloaded or whether the EV 100 is being driven on the gradient. The estimation and prediction is performed only after the EV 100 has begun operation and sufficient CAN 106 data is available for analysis in the trip.

[0019] The trip is the drive between a start of the EV 100 (speed becomes greater than 0 kmph) till stop of the EV 100 (speed = 0 kmph). The controller 110 performs estimation/prediction when the EV 100 is in a quasi-steady state driving condition (which is when the vehicle speed value is within a defined band of variation for a fixed duration of time). In the presence of the gradient, the quasi-steady state condition is noted, and the time duration is measured (in order to finally evaluate what proportion of the drive cycle was over the gradient). When no gradient is detected, the load model 112 estimates the weight of the EV 100 and then utilized this estimation to provide one of the usages such as perform classification to determine whether the trip has been a single rider trip or a two-rider trip. The estimations are performed in each and every trip.

[0020] In accordance to the present invention, a working of the controller 110 is explained and the same must not be understood in limiting manner. Consider the EV 100 to be a scooter. The unloaded weight (or kerb weight or dry weight) of the scooter is already known and fed into the controller 110 with certain tolerance band. When the scooter is driven say with one driver and one pillion rider. The controller 110 receives the input from a speed sensor 108 and determines the quasi-steady state conditions such as speed within a predetermined range and for a predetermined time. For example, the scooter was driven at 30km/hr. for 10 seconds. Alternatively, the scooter was accelerated from 0% throttle to certain range of throttle, say 30% as a ramp, in which case the throttle position is used as the characteristic feature to determine the quasi-steady state condition. Once the quasi-steady state is detected/confirmed, the controller 110 processes the phase/window for which the quasi-steady state is active/present. The quasi-steady state condition is possible to be active in parts and not continuous, for example, the quasi-steady state condition is detected in first 10 seconds of the trip, then fifth minute, then 7th minute and the like. In case the road gradient is detected which is either based estimated value or a gradient sensor 114, then the same is discarded for any processing by the controller 110. The controller 110 takes in the values of the characteristic parameters during the quasi-steady state conditions until the sufficient data requirement is met and processes through the load model 112. The load parameters 120 are estimated and then used for various purposes such as to estimate the aging or wear of the components based on the whether the vehicle is overloaded, or being driven aggressively. Alternatively, the weight along with the energy consumed is used to bill the rider if used in the context of motorcycle taxis or renting system. In another example, if there are three riders in the scooter of a renting system, then the controller 110 makes a note of this incident and later charges or penalizes a registered rider.

[0021] In an embodiment, the controller 110 is used to estimate only the weight of the EV 100 and the gradient is determined using a gradient sensor 114. The remaining operation and functionality of the controller 110 remains the same. The gradient sensor is shown in the Fig. 1 and is used if needed. The same does not convey and must not be understood in a manner that the gradient sensor 114 is mandatory. In another embodiment, controller 110 is used to estimate only the road gradient and weight is determined with weight sensors (not shown) mounted on the seat or other suitable positions in the EV 100. The remaining operation and functionality of controller 110 remains the same. The same does not convey and must not be understood in a manner that the weight sensor is mandatory.

[0022] Fig. 2 illustrates a method for estimating load parameter of the EV, according to the present invention. The EV 100 is equipped with components comprising at least one rotary electric machine 104 to drive the EV 100 and at least one battery pack 102 to supply electrical energy to the rotary electric machine 104 and accessories of the EV 100. The controller 110 and control unit of the components are connected through the Controller Area Network (CAN) 106 of the EV 100. The method comprises plurality of steps of which a step 202 comprises measuring characteristic parameters from the CAN 106 during the trip. The method is characterized by a step 204 comprising detecting at least one quasi-steady state condition of the EV 100 based the primary parameter within the characteristic parameters. The quasi-steady state condition is detected based on vehicle speed/acceleration as the characteristic parameter or throttle position or both. Any other characteristic parameter is also possible to be used. A step 206 comprises processing secondary parameter of the measured characteristic parameters and the unloaded weight of the EV 100 through the load model 112 and estimating the load parameter 120. The load parameter 120 is at least one selected from the group comprising the estimated weight of the EV 100 and the road gradient. The method is executed by the controller 110 and compliments the description of the controller 110 as provided in Fig. 1. The characteristic parameters comprises the combination of the primary parameter and the secondary parameter.

[0023] According to the method, the primary parameter is at least one selected from a group comprising a vehicle speed, a vehicle acceleration, and a throttle position. The secondary parameter is at least selected from a phase current drawn by the rotary electric machine 104, battery current supplied from the battery pack 102 to the rotary electric machine 104, a torque applied by the rotary electric machine 104, a power consumed by the rotary electric machine 104 (electric/mechanical), battery power, and combination thereof. The secondary parameter is further selectable from battery voltage, and efficiency of the rotary electric machine 104. Specifically, the primary parameter is used for the determination of the quasi-steady state condition, whereas the secondary parameter is used in the load model 112. The characteristic parameters are derived or measured using respective sensors already available in the EV 100, such as speed sensor 108, throttle position sensor 116, voltage and current sensors 118 and the like.

[0024] Further, as per step 208, the load parameters 120, once determined, are used for at least one selected from the group comprising estimating the range of the EV 100 for the current trip, estimating abuse of the EV 100, estimating aging of parts and components of the EV 100, enabling energy consumption based payment models, selecting optimal operating points for extracting higher efficiency out of the EV 100 based on estimated weight, enabling better acceleration and drive quality while driving over gradients, enhancing hill hold and hill assist features with estimated gradient, modulating and distributing optimal brake force based on estimated weight and gradient, adjusting suspension based on estimated weight, adjusting tire pressure based on estimated weight, classifying payload of the EV 100 , torque co-ordination/ modulation between wheels and the like.

[0025] The load model 112 is developed and pre-trained using any one but not limited to a Gaussian Process data model, a linear model, a polynomial model, a regression model, parameter estimation, curve fitting, Least Mean Square Error (LMSE), Recursive least square approximation, a Sum Squared Error (SSE), an absolute error, Machine Learning (ML) models, Deep Learning (DL) models, self-learning models and a combination thereof.

[0026] According to the method of the present invention, while the gradient is zero, the method comprises estimating weight of the EV 100, specifically the two-wheeler, from the load model 112 during the at least one quasi-steady state condition, and classifying the trip into single rider or two rider based on the estimated weight. The method is also applicable for other types of EVs 100 as explained before.

[0027] According to another method, once the EV 100 is started, the primary parameter of the EV 100 is monitored by the controller 110 for determination of the quasi-steady state condition. Once determined, the method comprises processing the secondary parameters through the load model 112 to estimate the road gradient. Once the road gradient is determined, the method for estimating weight comprises processing only those values of characteristic parameter which has zero gradient. In other words, the method eliminates or discards the values of the secondary parameters in the quasi-steady state condition where the gradient is detected to be non-zero and positive. The presence of road gradient is either through the load model 112 or through the gradient sensor 114. The remaining data is taken and processed through the load model 112 for estimating the payload/weight of the EV 100. Once the weight is estimated, the method comprises classifying the payload into single rider or multiple rider as one of the options. The different options are already explained above in the usage of load parameters 120.

[0028] According to the present invention, the controller 110 and method for sensor-less load/weight and gradient prediction/estimation for EVs 100 is disclosed. The controller 110 estimates payload using variables related to drive units of the EV 100. The controller 110 uses empirical formula, data based model or physics based model as an estimator. Further, the controller 110 uses weight estimator for accurate prediction of remaining mileage by the EV 100. The controller 110 and method also adapts control strategy for the rotary electric machine 104 for better vehicle dynamics and energy consumption assessment to avoid abuse. The controller 110 and method also enables to assess any sort of abuse with over loading of the EV 100 and indicate to user.

[0029] The controller 110 and method aims to identify vehicle payload (number of riders or heavy rider/ weight) and whether the EV 100 is being driven over a slope/gradient without the use of any additional sensors on EV 100. Using existing values of variables relevant to drivetrain systems of EV 100 (data already available on CAN bus 106), the controller 110 and method detects whether there is a single rider/ two riders or heavy rider on the EV 100. The present invention enables the detection of number of drivers of the EV 100 and conditions of driving over a gradient without the use of any additional sensors, hardware, or modifications on the EV 100. The controller 110 and the method uses load model 112 (an estimator) which is already trained with offline data. Hence, estimated weight and road gradient as output are provided instantaneously. The input required for estimation are already available in control unit of the rotary electric machine 104 and does not require any signal from standalone sensor or from any other controller 100 on EV 100. The present invention is implementable in the existing controllers 110 of the EVs 100 or over cloud application and does not require any additional hardware.

[0030] It should be understood that embodiments explained in the description above are only illustrative and do not limit the scope of this invention. Many such embodiments and other modifications and changes in the embodiment explained in the description are envisaged. The scope of the invention is only limited by the scope of the claims.