Electric vehicle drive motor control method

Abstract

The present invention provides a control method for an electric vehicle drive motor, which mainly includes three steps of performing diversified intelligent control on the electric vehicle drive motor based on a diversified intelligent control system. The steps mainly include: identifying the driver by the on-vehicle recognition system to determine the driving strategy, and during the execution of the driving strategy, when the real-time torque and the real-time speed do not match, a fault alarm is issued. Based on the prediction with a uniquely adjustable driving style coefficient by humans, the present invention synthesizes the torque output to represent the driver's evaluation of the pedal feel, realizes an input current compensation strategy for the drive motor, and is used to complete the personal intelligent customization of electric vehicle driving; at the same time, diversified means are adopted to monitor the speed and torque, which ensures the reduction of the overall failure rate of the monitoring and realizes the classification alarm of faults.

Description

Technical Field

[0001] The present invention relates to a control method for an electric vehicle drive motor, belonging to the field of motor intelligent control. Background Art

[0002] The motor of an electric vehicle uses the ratio of the pedal step to the output torque to generate a corresponding torque following the depth of the pedal step. Essentially, it is the pedal control of the drive motor input current. Therefore, the drive motor of an electric vehicle still uses the driver's subjective "foot feeling" as the improvement standard for control.

[0003] However, the actual situation is as follows. First, different drivers have very different understandings of the foot feeling due to different driving ages, driving habits, and personal style orientations. It is not possible to simply unify a standard foot feeling. Even if unified, it will inevitably sacrifice the driving needs of some drivers. Second, the existing technology uses the relationship between torque and current control for control, and the control means are single. If the control circuit fails, there is no further remedy, and repair and replacement become inevitable. Third, the existing control method does not allow the driver to participate in the personalized formulation of control. Instead, it is a habit-cultivating mode where users can only accept the foot feeling and fault detection experience brought by these design parameters according to how the traditional manufacturer designs and manufactures.

[0004] Therefore, there is an urgent need to introduce a solution that can reflect multiple means of monitoring and user personalized self-customization to meet the personalized market and bring a safer and more enjoyable driving experience. Summary of the Invention

[0005] In view of the above problems of the existing technology, the present invention will consider the following several solution directions. First, establish a system for detecting the foot feeling, which can provide a basis for subsequent intelligent recognition through a database of pedal movement over time, so as to autonomously adjust the control strategy according to the actual driver's habits. Second, adopt diversified means to monitor the rotational speed and torque, so that the three are coordinated with the pedal movement. Once a failure occurs in the monitoring of one or two of the three, the remaining means can still be coordinated with the pedal movement to provide foot feeling detection.

[0006] Based on the above considerations, the present invention provides a control method for an electric vehicle drive motor, including providing a diversified intelligent control system, the system includes a pair of protrusions designed at the output end of the drive motor and symmetric with respect to the axis of the drive shaft of the drive motor, or a fastening kit sleeved on the drive shaft and having the pair of protrusions, a first rangefinder detachably installed near the drive shaft and aligned with the horizontal position of the protrusions for ranging, a drive motor torque sensor, a rotary encoder, a current compensation system for compensating the input current of the drive motor, a second rangefinder installed on the pedal for sensing distance change data according to the depth of pedal depression, a vehicle entry recognition system, and an in-vehicle computer and a remote server communicating with the in-vehicle computer, wherein,

[0007] The in-vehicle computer is electrically connected to the torque sensor, the rotary encoder, the first rangefinder, the current compensation system, and the second rangefinder, and by collecting the data of the second rangefinder, establishes a pedal motion function between the collected second rangefinder data and time. At the same time, it respectively collects the torque data, rotation data, and first rangefinder data of the torque sensor, the rotary encoder, and the first rangefinder, and synchronously (with the same time independent variable) makes a torque function, a rotation function, and a rangefinder function on the pedal motion function, and converts the rotation function and the rangefinder function into the real-time speed of the drive motor, sets a specified time period and a second rangefinder data threshold,

[0008] The in-vehicle computer calculates the number of times (i.e., the total number of peaks and valleys) that the second rangefinder data changes exceeding the second rangefinder data threshold in multiple of the specified time periods based on the pedal motion function, obtains corresponding multiple stepping frequencies, calculates the average value, and sets the driving styles from low to high according to the average value, which are divided into conservative type, normal type, sport type, and intense type. The in-vehicle computer identifies the current driving style according to the actual measured stepping frequency of the driver, and based on multiple subsequent current identified driving style results, determines the compensation strategy for the input current of the drive motor by the current compensation system, and continuously updates the compensation strategy in real time due to the accumulation of driving duration, and decides to use the updated compensation strategy to compensate the input current of the drive motor; when the real-time torque and the real-time speed do not match, a fault alarm is issued;

[0009] The specific method includes performing diversified intelligent control of the electric vehicle drive motor based on the diversified intelligent control system, and the steps are as follows:

[0010] In the first step, the vehicle entry recognition system is used to identify the driver. If it is a new driver, or the number of recorded driving styles is less than the predetermined number, then the driving strategy is determined according to the function of the stepping depth (i.e., measured by the difference between the initial distance D of zero pedal depression and the second rangefinder data)-standard current input I s and continuously records the driving style;

[0011] In the second step, after a preset number of records are made, the new driver is identified as a non-new driver, the non-new driver list is updated, and the driving is carried out using the compensation strategy; if it is identified as a non-new driver, the driving is carried out according to the previous driving strategy, and like the newly identified non-new driver, the driving style is continuously recorded and the driving strategy is continuously updated;

[0012] In the third step, during the execution of the driving strategy, a fault alarm is issued when the real-time torque and the real-time speed do not match.

[0013] It is easy to understand that when the drive motor rotates, the acquisition data of the first distance measuring instrument forms discrete pulse changes because the protrusion period passes through the first distance measuring instrument, and the speed of the drive shaft of the drive motor is calculated according to the frequency of the pulses. This is one of the means. Another speed measurement means can use a rotary encoder, so as to ensure the diversification of speed measurement means and ensure that one can continue to monitor when the other fails.

[0014] Optionally, the method for formulating the compensation strategy includes the following steps:

[0015] S1 The in-vehicle computer obtains the real-time stepping depth according to the pedal motion function, and calculates the change rate of the standard input current of the drive motor according to the stepping depth according to the pre-formulated stepping depth-standard current input I s function S2 The in-vehicle computer calculates the current actual input current change rate according to the measured torque function P r K is the torque constant, and the change rate difference is calculated S3 The in-vehicle computer determines the compensation current value given to the drive motor by the current compensation system according to the driver's current driving style where α is the current driving style coefficient, t c is the current moment, δt is a preset short time moment after the current moment t c and P r at t c to δt is obtained by artificial intelligence prediction for the function segment;

[0016] S4 Similarly, according to S3, the torque function P r at t c before, the corresponding multiple compensation current values within k∈natural number of the specified time periods T α k is the corresponding driving style coefficient within the kth specified time period T before the current moment, and it is defined that α0 = α, and δt' is a preset short time moment after t c -kT, and k + 1 compensation current values are obtained;

[0017] ​The S5 remote server establishes an intelligent compensation prediction model. The in-vehicle computer sends k+1 compensation current values to the remote server. The remote server sorts all the compensation current values of each consecutive 24-hour period in chronological order in units of 24 hours to form an array, and assigns corresponding gray values or pseudo-colors to the current values, converting an array of compensation current values for a 24-hour period into a compensation image, thus obtaining multiple compensation images. These are divided into a training set and a validation set. The driving styles corresponding to the calculation of the compensation current values in each compensation image are statistically analyzed to obtain the average driving style coefficient corresponding to this compensation image. N is the number of the total specified time periods T within 24 hours minus 1; the training set is input into the intelligent compensation prediction model for training to predict the average driving style, the accuracy of the model is verified using the validation set, and the loss function value is calculated to optimize the model parameters. When the training reaches a stable accuracy, the training is stopped, and the intelligent compensation prediction model is obtained and sent to the in-vehicle computer for storage;

[0018] S6 The remote server generates the measured compensation image corresponding to the current moment t c or t c and any subsequent moment, and sends it to the in-vehicle computer. The in-vehicle computer substitutes it into the stored intelligent compensation prediction model to obtain the predicted average driving style coefficient α p , then it is determined that the input current magnitude of the current compensation system to the drive motor is where t cc is the current moment t c or t c and any subsequent moment, δt" is a preset short time moment after t cc , and P r in the function segment between t cc and δt" is also obtained by the above-mentioned artificial intelligence prediction. When I cc is negative, the in-vehicle computer directly controls the input current magnitude of the drive motor based on the real-time pedal depth to be I s | tcc -|I cc |, otherwise, the control intervenes in the current compensation system so that the input current magnitude of the drive motor is where the measured compensation image is the compensation image formed by the compensation current values calculated according to the S4 method between 24 hours backward with t cc as the end point, and are the standard current input and torque values at the moment of t cc respectively.

[0019] It should be understood that the above change rate is essentially the right derivative, which is the right derivative corresponding to the current moment of the pedal movement function and the measured torque function up to the current moment. Since the pedal movement function and the torque function are made synchronously, the derivative is the right derivative taken at each same current moment. Although t cc is probably not the time point corresponding to an integer number of 24 hours before and after t c , however, since the driving habits of drivers are generally stable, t cc can be regarded as the end point and regarded as t c (when t cc is t c excepted) and backward extrapolate 24 hours, which does not affect the prediction result.

[0020] In addition, although essentially the average driving style coefficient is calculated based on the existing compensation current in the past, the input current magnitude of the current compensation system to the drive motor determined based on the average driving style coefficient is for the compensation prediction of any future time period after the current moment t c or t c . Therefore, the calculated average driving style coefficient is actually regarded as the prediction result of the average driving style coefficient between t cc and δt" using past data.

[0021] Thus, the compensation strategy defines the driving style by the stepping frequency, synthesizes the torque output to represent the driver's evaluation of the pedal feel, and thus determines the final total input current to control the drive motor so that the relationship between the driving effect and the pedal feel meets the requirements of different driver driving styles.

[0022] Optionally, for conservative, normal, sporty, and intense driving styles, the value ranges of the driving style coefficients are 0.5 - 0.9, 0.9 - 1.1, 1.1 - 1.5, and 1.5 - 2 respectively, the preset short - time moment is 0.1 - 0.5 s, and the specified time period T = 10 s - 15 min.

[0023] Optionally, the method for obtaining the function segment of the P r between t c and δt by artificial intelligence prediction includes:

[0024] S3 - 1 Obtain consecutive multiple 24 - hour moments before t c to obtain multiple change rates corresponding to P r ; m is the natural number serial number of the number of 24 - hour moments from t c ; that is, the m - th 24 - hour moment represents the moment 24m hours before t c ; S3 - 2 Take the weighted average of multiple change rates βm is a weighting coefficient, and for any y ∈ m, x < y, there is always β x > β y ;

[0025] S3-3 is based on t c , to calculate the torque value at δt, and linearly connect the torque data points at t c and δt to obtain a predicted function segment.

[0026] Optionally, the current compensation system includes an energy storage device that stores the electric energy generated by the rotational motion of the wheels during the period from releasing the pedal to the next pedal step, which is converted into the electric energy of the drive motor, and a PWM speed regulation device or a voltage transformation device controllably connected in series with the energy storage device. The PWM speed regulation device or the voltage transformation device communicates with the vehicle-mounted computer, and the vehicle-mounted computer controls the PWM speed regulation device or the voltage transformation device to adjust the compensation current value according to the driving strategy.

[0027] Optionally, the diversified intelligent control system further includes a mobile intelligent terminal held by the driver, which communicates with the remote server and / or the vehicle-mounted computer, and is used to view the pedal motion function, torque function, rotation function, ranging function, and the function curve of the pre-established pedal depth - standard current input I s . It can set the value range for different types corresponding to α, but the minimum value cannot be lower than 0.3, and the maximum value cannot exceed 2.3. It can also view and select to execute the compensation strategy used by the non-new driver in the most recent driving recorded by the vehicle-mounted computer. The non-new driver and / or the vehicle owner can set whether to allow viewing their own compensation strategy on the mobile intelligent terminal they hold. When the in-vehicle recognition system recognizes that the driver is not these personnel, the function of viewing the compensation strategy is correspondingly opened or closed.

[0028] Optionally, the vehicle-mounted computer and / or the mobile intelligent terminal can calibrate the initial distance D of zero pedal step, and the vehicle-mounted computer updates the function of the pedal depth - standard current input I s .

[0029] It is easy to understand that due to the different update situations of the floor carpets or floor mats in the vehicle, it is necessary to perform initial distance calibration.

[0030] Optionally, when the real-time torque and the real-time rotational speed do not match, a fault alarm is issued, which specifically includes: in the first case, when the difference between the real-time rotational speed converted from the rotation function and the ranging function is within the threshold range, it is alarmed that there may be a torque sensor and / or engine fault;

[0031] In the second case, if the converted real-time rotational speed difference of the driving motor from the rotation function and the distance measurement function is not within the threshold range, and the real-time torque does not match the real-time rotational speed measured by only one of the rotary encoder and the first distance measuring instrument, an alarm will be issued for the possible malfunction of the unmatched one;

[0032] In the third case, if the converted real-time rotational speed difference of the driving motor from the rotation function and the distance measurement function is not within the threshold range, and the real-time torque does not match the real-time rotational speeds measured by both the rotary encoder and the first distance measuring instrument, then the torque sensor, the rotary encoder, the first distance measuring instrument, and the engine may all malfunction. An alarm will suggest repair and further inspection of the torque sensor, the rotary encoder, the first distance measuring instrument, and the engine.

[0033] Beneficial effects

[0034] 1. Through the pedal motion function, the torque function, the rotation function, the distance measurement function, and the pre-established function curve of the stepping depth - standard current input I s and by training the artificial intelligence model for prediction based on the unique and artificially adjustable driving style coefficient, the comprehensive torque output represents the driver's evaluation of the pedal feel, realizing the input current compensation strategy for the driving motor to complete the personal intelligent customization of electric vehicle driving;

[0035] 2. Adopting diversified means to monitor the rotational speed and torque, ensuring a reduction in the overall failure rate of the monitoring;

[0036] 3. Based on the diversified means, a classified alarm for faults is realized. Description of the drawings

[0037] Figure 1 Schematic diagram of a diversified intelligent control system adopted in the electric vehicle drive motor control method of Embodiment 1 of the present invention in an electric vehicle,

[0038] Figure 2 Synchronous schematic diagram of the a pedal motion function, torque function, rotation function, and distance measurement function,

[0039] Figure 2 b is Figure 2 The measured torque function P r , time-varying function I s (t), the function partial of the stepping depth - standard current input I s , and the local enlarged view of the partial at time t c ,

[0040] Figure 3 Schematic diagram of obtaining the second ranging data and its threshold,

[0041] Figure 4 Schematic diagram of the app interface on the smartphone connected to the in-vehicle computer,

[0042] Figure 5 Schematic diagram of the structure and operating principle of the current compensation system

[0043] Figure 6 Flow chart of the specific steps of the intelligent control method

[0044] Figure 7 Composition structure diagram of the compensation image in Embodiment 2 of the present invention

[0045] Figure 8 Simplified diagram of the training process of the intelligent compensation prediction model

[0046] Wherein, 1 - drive motor, 11 - drive shaft, 12 - protrusion, 13 - first rangefinder, 14 - torque sensor, 2 - rear axle, 21 - rotary encoder, 3 - pedal, 4 - vehicle entry recognition system, 5 - vehicle-mounted computer, 6 - remote server, 7 - smart phone, 8 - rearview mirror, 9 - steering wheel. Specific implementation mode

[0047] Embodiment 1

[0048] This embodiment gives an overall description of a control method for an electric vehicle drive motor. Specifically, the method includes adopting a set of diversified intelligent control systems, such as Figure 1 The installation schematic of the drive motor and related system components in the vehicle and at the rear wheels is given.

[0049] The diversified intelligent control system includes: a pair of protrusions 12 designed at the output end of the drive motor 1 and symmetric with respect to the axis of the drive shaft 11 of the drive motor; a first rangefinder 13 detachably installed on the outer shell of the drive motor 1 near the drive shaft 11 and aligned with the horizontal position of the protrusion 12 for ranging; a drive motor torque sensor 14 installed on the drive shaft 11; a rotary encoder 21 installed on the rear axle 2 of the vehicle; a current compensation system for compensating the input current of the drive motor 1; a second rangefinder installed on the pedal 3; for sensing distance change data according to the depression depth of the pedal 3; a vehicle entry recognition system 4, a vehicle-mounted computer 5, and a remote server 6 communicating with the vehicle-mounted computer 5.

[0050] The vehicle-mounted computer 5 is electrically connected to the torque sensor 14, the rotary encoder 21, the first rangefinder 13, the current compensation system, and the second rangefinder, and by collecting the data of the second rangefinder, a pedal motion function between the collected data of the second rangefinder and time is established. At the same time, the torque data, rotation data, and first ranging data of the torque sensor 14, the rotary encoder 21, and the first rangefinder 13 are respectively collected, and a torque function, a rotation function, and a ranging function ( Figure 2)These three types of functions. And according to the rotation function and the ranging function, the real-time rotational speed of the drive motor is converted, and a specified time period and a second ranging data threshold are set.

[0051] Figure 2 Figure a shows a synchronization diagram of the pedal movement function and the three types of functions within a specified time period T, where the second ranging data threshold is explained in Figure 3 . Figure 3 Figure shows the car floor with a floor mat laid on it. A second rangefinder is installed on the pedal. As shown in the figure, it is in the zero-step state, and the maximum detected distance, that is, the initial distance D, is measured. The second ranging data threshold is the distance when the detected distance is less than D, generally within 1 cm less than D. Therefore Figure 2 in Figure a, the pedal movement function only responds when it is below this threshold. The torque function is similar in shape to the pedal movement function. Because the input current of the pedal depth is proportional to the torque, inversely proportional to the torque, and proportional to the rotational speed, the larger the second ranging data, the greater the torque and the smaller the rotational speed.

[0052] The following explains the driving style. The in-vehicle computer 5 calculates the number of times the second ranging data exceeds the second ranging data threshold within multiple specified time periods based on the pedal movement function, and obtains corresponding multiple stepping frequencies. Figure 2 In Figure a, the number of changes within one T (1 minute) is 3, that is, the stepping frequency is 3.

[0053] Calculate the average value for multiple stepping frequencies, and set the driving style from low to high according to the average value. Still taking Figure 2 Figure a as an example within one T, it is divided into conservative type 0 times, normal type 1 time, sporty type 2 times, and intense type 3 times. That is Figure 2 Figure a shows that for the in-vehicle computer 5 for intense driving, it identifies the current driving style according to the actual stepping frequency of the driver, and based on multiple current identified driving style results thereafter, determines the compensation strategy for the input current of the drive motor by the current compensation system, and continuously updates the compensation strategy in real time due to the accumulation of driving duration, and decides to compensate the input current of the drive motor using the updated compensation strategy; when the real-time torque and the real-time rotational speed do not match, a fault alarm is issued.

[0054] The diversified intelligent control system further includes a smartphone 7 (or a tablet computer) held by the driver, which communicates with the remote server 6 and the in-vehicle computer 5, and is used to view the pedal movement function, torque function, rotation function, and ranging function in real time. Figure 4 Figure shows the app interface on the smartphone 7 connected to the in-vehicle computer 5, which schematically shows Figure 2Function synchronization diagram of a. The hidden synchronous function diagrams can be displayed by clicking the "View" button. Both the in-vehicle computer 5 and the mobile intelligent terminal can calibrate the initial distance D of zero pedal depression, and the in-vehicle computer 5 updates the pedal depression - standard current input I based on D. s function.

[0055] Figure 5 As shown, the specific current compensation system includes an energy storage device and a PWM speed control device controllably connected in series with the energy storage device through a control switch. The energy storage device stores the electric energy generated by the rotational motion of the wheel being converted into the driving motor during the period from releasing the pedal to the next pedal depression. The PWM speed control device and the control switch communicate with the in-vehicle computer 5, and the in-vehicle computer 5 controls the PWM speed control device or the voltage transformation device to adjust the compensation current value according to the driving strategy. Thus, when storing energy, the in-vehicle computer 5 controls the switch to disconnect the access of the battery and connect the energy storage device for energy storage; when the pedal is depressed, the control switch makes the energy storage device be connected in parallel with the battery for current compensation.

[0056] As Figure 6 shown, the specific method includes performing diversified intelligent control of the driving motor of the electric vehicle based on the above-mentioned diversified intelligent control system, and the steps are as follows:

[0057] In the first step, after the driver gets in the car, the in-car recognition system 4 set on the left rearview mirror 8 or the steering wheel 9, such as Figure 1 , recognizes the driver (for example, using face or fingerprint recognition). If it is a new driver, or the number of recorded driving styles is less than the predetermined number, then according to the pre-established pedal depression (measured by the difference obtained by subtracting the second ranging data from the initial distance D of zero pedal depression) - standard current input I s function to determine the driving strategy, and record the driving style every specified time period T.

[0058] In the second step, after the preset number is recorded, then this new driver is recognized as a non-new driver, and the non-new driver list is updated, and the compensation strategy is adopted for driving; if it is recognized as a non-new driver, then drive according to the previous driving strategy, and like the newly recognized non-new driver, continue to continuously record the driving style and continuously update the driving strategy;

[0059] In the third step, during the execution of the driving strategy, when the real-time torque and the real-time speed do not match, a fault alarm is issued, otherwise Null does nothing.

[0060] Figure 6 If the recognition fails in the first step of s , then the input current is controlled according to the function of the pedal depression - standard current input I throughout the driving process. Specifically, the pre-established pedal depression - standard current input Is The method for formulating the function of s includes: recruiting multiple drivers with a driving experience of more than ten years, having these drivers drive to obtain multiple current recognized driving style results, screening out multiple drivers with normal driving styles based on the results, retrieving multiple function graphs of the corresponding pedal depression - current input during driving of these drivers, and averaging the multiple function graphs to obtain the pedal depression - standard current input I s function.

[0061] Embodiment 2

[0062] This embodiment will illustrate that the method for formulating the compensation strategy includes the following steps:

[0063] S1 The in - vehicle computer obtains the real - time pedal depression according to the pedal movement function as shown in Figure 2 Figure a, and calculates the change rate of the standard drive motor input current according to the pedal depression and the pre - formulated pedal depression - standard current input I Figure 2 as shown in s Figure b. Since the pedal depression is a function of time, the standard current input I s is also Figure 2 a time - varying function I s (t) shown in s Figure b, and the independent variable range of the time - varying function I r (t) takes a sufficiently long time to ensure the time synchronization when calculating the change rate difference in step S2.

[0064] S2 The in - vehicle computer calculates the current actual input current change rate according to the measured torque function P where K is the torque constant, and calculates the change rate difference Figure 2 The current moment t is circled in c Figure b, and is the right derivative at this moment. S3 The in - vehicle computer determines the compensation current value given by the current compensation system to the drive motor according to the driver's current driving style where α is the current driving style coefficient, δt is a preset short - time moment after the current moment t c , and P r in the function segment between t c and δt is obtained by artificial intelligence prediction; Figure 2 The enlarged view of the circled moment t where the right derivative exists is also given in c Figure b, where the I c between t s (t) and the predicted P r are given to find the right derivative of the moment t c .

[0065] Similarly, according to S3, the torque function P is obtained. r At time t c Before that, corresponding to k ∈ natural numbers of the specified time periods T, a plurality of compensation current values α k Is the driving style coefficient corresponding to the k-th specified time period T before the current moment. Define α0 = α, and δt' is the preset short time moment after t c - kT. k + 1 compensation current values are obtained;

[0066] S5 The remote server 6 establishes an intelligent compensation prediction model. The in-vehicle computer sends the k + 1 compensation current values to the remote server 6. The intelligent compensation prediction model includes a convolutional neural network CNN, and the output end is connected to a fully connected function and then output to the softmax function for classification.

[0067] The remote server 6 takes 24 hours as a unit, sorts all the compensation current values of each continuous 24-hour period in chronological order. As Figure 7 Shown, arranged from left to right first and then from top to bottom in the direction of the arrow to form an array, and assign current values and corresponding gray values. Convert the compensation current value array of a 24-hour period into as Figure 7 Shown compensation image (the horizontal and vertical three points indicate omitting other multiple 6×6 pixel sub-arrays), thus obtaining as Figure 8 Shown multiple compensation images.

[0068] As Figure 8 In, it is divided into a training set and a validation set. The driving styles corresponding to the compensation current values in each compensation image are counted to obtain the average driving style coefficient corresponding to the compensation image N is the number of the total specified time periods T within 24 hours minus 1; the training set is substituted into the intelligent compensation prediction model for training, and the predicted average driving style classification is output by the softmax function. The validation set is used to verify the accuracy of the model, and the loss function value is calculated to optimize the model parameters. When the training makes the accuracy stable, stop the training, obtain the intelligent compensation prediction model, and send it to the in-vehicle computer 5 for storage;

[0069] S6 The remote server 6 generates the measured compensation image corresponding to the current moment t c Or t c After any moment, and send it to the in-vehicle computer. The in-vehicle computer substitutes it into the stored intelligent compensation prediction model to obtain the predicted average driving style coefficient α p , then determine that the input current magnitude of the current compensation system to the drive motor is Where t cc Is the current moment t c Or tc At any time after that, δt" is t cc At a preset short time after that, and P r At t cc The function segment between and δt" is also obtained by the artificial intelligence prediction. When I cc Is negative, the on-vehicle computer directly controls the input current of the drive motor based on the real-time pedal depth to be I s | tcc -|I cc |, otherwise, control intervenes in the current compensation system so that the input current of the drive motor is Among them, the measured compensation image is the compensation image formed by the compensation current values calculated according to the S4 method between t cc And the end point in the reverse 24 hours, And Are respectively the standard current input and torque value at the moment of t cc Negative values are negative compensations, which are caused by current driving such as conservative driving. Controlling the input current of the drive motor to be lower than the standard input current, otherwise a positive compensation is formed.

[0070] For conservative, normal, sporty, and intense driving styles, the value ranges of the driving style coefficients are 0.7, 1.0, 1.3, and 1.7 respectively, and the preset short time is 0.2s. Such as Figure 4 , the smartphone can also use the "α setting" button as an entry to set the value range for different types corresponding to α, but the minimum value cannot be lower than 0.3, and the maximum value cannot exceed 2.3. It can use the "compensation strategy" button as an entry to view and select the compensation strategy used by the on-vehicle computer to record the last drive of a non-new driver. Non-new drivers and / or car owners can set whether to allow or not allow viewing their own compensation strategies on their mobile intelligent terminals. When the in-vehicle recognition system identifies that the driver is not these people, the function of viewing the compensation strategy is correspondingly opened or closed.

[0071] In step S3, the P r At t c The method for obtaining the function segment between and δt by artificial intelligence prediction includes:

[0072] S3-1 Obtain consecutive multiple 24-hour moments before t c To obtain multiple change rates corresponding to P r On m is the natural number serial number of the number of 24-hour moments from t c ; that is, the mth 24-hour moment represents the moment 24m hours before t c ; S3-2 Take the weighted average of multiple change rates J is the total number of 24-hour moments at a distance t c , and M = 1 + 2 + … + J;

[0073] S3-3 is based on t c 、 to calculate the torque value at δt. For example Figure 2 b, connect the torque data points at t c and δt with a straight line to obtain the predicted function segment.

[0074] Finally, for fault alarm, when the real-time torque and real-time speed do not match, a fault alarm will be issued. Specifically, it includes: In the first case, when the difference in the real-time speed of the drive motor converted from the rotation function and the ranging function (derived from the power transmission differential) is within the threshold range, an alarm may indicate a possible fault in the torque sensor 14 and / or the engine;

[0075] In the second case, when the difference in the real-time speed of the drive motor converted from the rotation function and the ranging function is not within the threshold range, and the real-time torque only does not match the real-time speed measured by either the rotary encoder 21 or the first ranging instrument 13, an alarm may indicate a possible fault in the non-matching device;

[0076] In the third case, when the difference in the real-time speed of the drive motor converted from the rotation function and the ranging function is not within the threshold range, and the real-time torque does not match the real-time speeds measured by both the rotary encoder 21 and the first ranging instrument 13, then the torque sensor 14, the rotary encoder 21, the first ranging instrument 13, and the engine may all be faulty, and the alarm recommends repair and further inspection of the torque sensor 14, the rotary encoder, the first ranging instrument 13, and the engine.

Claims

1. A control method for an electric vehicle drive motor, characterized in that, Including providing a set of diversified intelligent control systems, the system includes: a pair of protrusions designed at the output end of the drive motor and symmetric with respect to the axis of the drive shaft of the drive motor, or a fastening kit sleeved on the drive shaft and having the pair of protrusions, a first rangefinder detachably installed near the drive shaft and aligned with the horizontal position of the protrusions for distance measurement, a drive motor torque sensor, a rotary encoder, a current compensation system for compensating the input current of the drive motor, a second rangefinder installed on the pedal for sensing distance change data according to the depth of the pedal being stepped on, an in-vehicle identification system, an in-vehicle computer, and a remote server communicating with the in-vehicle computer, wherein, The in-vehicle computer is electrically connected to the torque sensor, the rotary encoder, the first rangefinder, the current compensation system, and the second rangefinder, and by collecting the data of the second rangefinder, establishes a pedal motion function between the collected second rangefinder data and time. At the same time, it respectively collects the torque data, rotation data of the torque sensor, rotary encoder, and first rangefinder, and the first rangefinder data, and synchronously makes a torque function, a rotation function, and a rangefinder function on the pedal motion function, and converts them into the real-time speed of the drive motor according to the rotation function and the rangefinder function. Sets a specified time period and a second rangefinder data threshold, The in-vehicle computer calculates the number of times the second rangefinder data changes and exceeds the second rangefinder data threshold in multiple said specified time periods based on the pedal motion function, obtains corresponding multiple stepping frequencies, calculates the average value, and sets the driving style from low to high according to the average value, which is divided into conservative type, normal type, sporty type, and intense type. The in-vehicle computer identifies the current driving style according to the actual stepping frequency of the driver, and based on multiple subsequent current identified driving style results, determines the compensation strategy for the input current of the drive motor by the current compensation system, and continuously updates the compensation strategy in real time due to the accumulation of driving duration, and decides to use the updated compensation strategy to compensate the input current of the drive motor; when the real-time torque and the real-time speed do not match, a fault alarm is issued; The specific method includes performing diversified intelligent control of the drive motor of the electric vehicle based on the diversified intelligent control system, and the steps are as follows: First step, identify the driver through the vehicle entry recognition system. If the driver is a new driver or the number of recorded driving styles is less than the predetermined number, determine the driving strategy according to the function of the pre-established pedal depth - standard current input, and continuously record the driving style; if the recognition fails, the entire driving process will input the current according to the function of the pedal depth - standard current input and continuously record the driving style; if the recognition fails, the entire driving process will input the current according to the function of the pedal depth - standard current input to control the input current; In the second step, when the preset number is recorded, the new driver is recognized as a non-new driver, and the non-new driver list is updated, and the drive is performed using the compensation strategy; if it is recognized as a non-new driver, the drive is performed according to the previous drive strategy, and like the newly recognized non-new driver, continuously records the driving style and continuously updates the drive strategy; In the third step, during the execution of the drive strategy, when the real-time torque and the real-time speed do not match, a fault alarm is issued.

2. The method according to claim 1, wherein The method for formulating the compensation strategy includes the following steps: S1. The onboard computer obtains the real-time pedaling depth according to the pedal motion function, and inputs the pedaling depth-standard current according to the pre-set pedaling depth. The function of calculating the standard rate of change of the input current of the drive motor , S2. The in-vehicle computer calculates the current actual input current change rate according to the measured torque function and , where is the torque constant, and calculates the change rate difference ; S3. The in-vehicle computer determines the compensation current value for the drive motor by the current compensation system according to the driver's current driving style , where is the current driving style coefficient, is the current moment, is at the current moment the preset short-time moment after that, and at to the function segment between is obtained by artificial intelligence prediction; S4. Similarly, obtain the torque function according to S3 At before each of the said specified time periods corresponding multiple compensation current values , is the driving style coefficient corresponding to the th specified time period before the current moment, and define , , is the preset short time moment after to obtain multiple compensation current values; S5. The remote server establishes an intelligent compensation prediction model, and the in-vehicle computer sends compensation current values to the remote server. The remote server sorts all the compensation current values of each continuous 24-hour period in chronological order in units of 24 hours to form an array, assigns corresponding gray values or pseudo-colors to the current values, converts the array of compensation current values for a 24-hour period into a compensation image, thereby obtaining multiple compensation images, divides them into a training set and a validation set, counts the driving styles corresponding to the calculation of the compensation current values in each compensation image, and obtains the average driving style coefficient corresponding to this compensation image , is the total specified time period within 24 hours is the number of minus 1; substitute the training set into the intelligent compensation prediction model for training, predict the average driving style, use the validation set to verify the model accuracy, and calculate the loss function value for optimizing the model parameters. When the training reaches a stable accuracy, stop the training, obtain the intelligent compensation prediction model, and send it to the in-vehicle computer for storage; S6. The remote server generates the current moment or any moment after that, and sends the measured compensation image corresponding to it to the in-vehicle computer. The in-vehicle computer substitutes it into the saved intelligent compensation prediction model to obtain the predicted average driving style coefficient , and then determines that the input current magnitude of the drive motor by the current compensation system is , where is the current moment or any moment after that, is a preset short time moment after that, and at to the function segment between them is also obtained by the artificial intelligence prediction. When is negative, the in-vehicle computer directly controls the input current magnitude of the drive motor based on the real-time pedal depth to be , otherwise, the control intervenes in the current compensation system so that the input current magnitude of the drive motor is , where the measured compensation image is the compensation image formed by the compensation current values calculated according to the S4 method between 24 hours back from the end point, and are respectively the standard current input and torque values at the moment 3. The method according to claim 2, wherein The said Between And The function segment obtained by artificial intelligence prediction includes: S3-1. Obtain The consecutive multiple 24-hour moments before, and obtain the corresponding Multiple rates of change thereon , Is the natural number serial number of the number of 24-hour moments from ; That is, the th 24-hour moment represents the moment 24 hours before from ; S3-2. Take the weighted average of multiple rates of change , is the weighting coefficient, , and , always have ; S3-3. Based on , , calculate the torque value at . Connect the torque data points at and with a straight line to obtain the predicted function segment.

4. The method according to claim 3, characterized in that, The current compensation system includes an energy storage device that stores the electric energy generated by the rotational motion of the wheels and converted into the driving motor during the period from releasing the pedal to the next pedal step, and a PWM speed control device or a voltage transformation device controllably connected in series with the energy storage device. The PWM speed control device or the voltage transformation device communicates with the vehicle-mounted computer, and the vehicle-mounted computer controls the PWM speed control device or the voltage transformation device to adjust the compensation current value according to the driving strategy.

5. The method according to claim 4, wherein The diversified intelligent control system further includes a mobile intelligent terminal held by the driver and communicating with the remote server and / or the in-vehicle computer, which is used to view in real time the pedal motion function, torque function, rotation function, ranging function, and the pre-established function curve of pedal depth - standard current input and can set the value range for different types corresponding to it. However, the minimum value cannot be lower than 0.3, and the maximum value cannot exceed 2.

3. It can also view and select the compensation strategy used in the last driving of the non-new driver recorded by the specified in-vehicle computer. Non-new drivers and / or vehicle owners can set on or off the viewing of their own compensation strategies on the mobile intelligent terminals they hold. When the in-vehicle recognition system identifies that the driver is not one of these people, the function of viewing the compensation strategy is correspondingly turned on or off.

6. The method according to claim 5, wherein The in-vehicle computer and / or the mobile intelligent terminal can calibrate the initial distance D of zero pedal depression, and the in-vehicle computer updates the function of pedal depression - standard current input based on D. The method for formulating the function of pedal depression - standard current input includes: recruiting multiple drivers with a driving experience of more than ten years, allowing these drivers to drive to obtain multiple current recognized driving style results, screening out multiple drivers with normal driving styles based on the results, retrieving multiple function graphs of pedal depression - current input corresponding to these drivers during driving, and averaging the multiple function graphs to obtain the function of pedal depression - standard current input function.

7. The method according to claim 6, wherein When the real-time torque and the real-time rotational speed do not match, a fault alarm is issued. Specifically, in the first case, if the difference in the real-time rotational speed of the driving motor converted by the rotation function and the ranging function is within the threshold range, an alarm may indicate a possible torque sensor and / or engine fault. In the second case, if the difference in the real-time rotational speed of the driving motor converted by the rotation function and the ranging function is not within the threshold range, and the real-time torque does not match the real-time rotational speed measured by only one of the rotary encoder and the first rangefinder, an alarm may indicate a possible fault of the non-matching device. In the third case, if the difference in the real-time rotational speed of the driving motor converted by the rotation function and the ranging function is not within the threshold range, and the real-time torque does not match the real-time rotational speeds measured by both the rotary encoder and the first rangefinder, then the torque sensor, the rotary encoder, the first rangefinder, and the engine may all have faults, and the alarm suggests repairing and further checking the torque sensor, the rotary encoder, the first rangefinder, and the engine.

8. A diversified intelligent control system for an electric vehicle drive motor, characterized in that, The system includes a pair of protrusions designed at the output end of the driving motor and symmetric with respect to the axis of the driving shaft of the driving motor, or a fastening kit sleeved on the driving shaft and having the pair of protrusions. A first rangefinder that is detachably installed near the driving shaft and aligned with the horizontal position of the protrusions for ranging, a driving motor torque sensor, a rotary encoder, a current compensation system for compensating the input current of the driving motor, a second rangefinder installed on the pedal for sensing distance change data according to the depth of the pedal step, an in-vehicle recognition system, a vehicle-mounted computer, and a remote server communicating with the vehicle-mounted computer, and a mobile intelligent terminal held by the driver and communicating with the remote server and / or the vehicle-mounted computer. Among them, The vehicle-mounted computer is electrically connected to the torque sensor, the rotary encoder, the first rangefinder, the current compensation system, and the second rangefinder. By collecting the data of the second rangefinder, a pedal motion function between the collected second ranging data and time is established. At the same time, the torque data, rotation data, and first ranging data of the torque sensor, the rotary encoder, and the first rangefinder are respectively collected, and a torque function, a rotation function, and a ranging function are respectively made synchronously on the pedal motion function. The real-time rotational speed of the driving motor is calculated according to the rotation function and the ranging function, and a specified time period and a second ranging data threshold are set. The on-board computer calculates the number of times the second distance measurement data changes due to exceeding the second distance measurement data threshold value in the plurality of prescribed time periods based on the pedal motion function, obtains a plurality of corresponding pedaling frequencies, calculates an average value, and sets the driving style from low to high according to the average value, which is divided into conservative, normal, sporty, and intense. The on-board computer identifies the current driving style of the driver according to the actual pedaling frequency measured by the driver, and determines the compensation strategy of the current compensation system for the input current of the drive motor according to the method described in any one of claims 2 to 7 based on the results of the multiple current identified driving styles performed thereafter, and continuously updates the compensation strategy in real time due to the accumulation of driving time, and decides to use the updated compensation strategy to compensate the input current of the drive motor; when the real-time torque and the real-time speed do not match, a fault alarm is issued; The mobile intelligent terminal is used to view the pedal motion function, torque function, rotation function, distance measurement function, and the pre-set pedal depth-standard current input in real time. The function curve can set the value range for different types of driving style coefficients, but the minimum value cannot be lower than 0.3 and the maximum value cannot exceed 2.

3. It can also view and select the compensation strategy used by the non-new driver in the most recent driving recorded by the designated on-board computer. The non-new driver and / or the car owner can set on the mobile smart terminal they hold to allow or not to view the compensation strategy of the non-new driver and / or the car owner. When the vehicle entry recognition system identifies the driver as non-these persons, the function of viewing the compensation strategy is turned on or off accordingly.

9. The diversified intelligent control system according to claim 8, wherein, The mobile intelligent terminal is a smart phone or a tablet computer.

Images