Flat car high-precision control method and system under complex road surface condition

Abstract

The invention relates to the field of automatic driving, in particular to a high-precision control method and system for a flat car under the condition of a complex road surface. The method comprises the steps of collecting flat car sensor data, obtaining flat car state data according to a flat car moving speed and the flat car sensor data, and obtaining a deviation vector; constructing a flat car stability control model, and obtaining expected correction data according to the deviation vector; and according to the expected correction data and the flat car sensor data, performing flat car motion parameter optimization to perform optimal control on the flat car. The flat car state data is obtained according to the flat car moving speed and the flat car sensor data, the deviation vector is obtained according to the flat car state data, the deviation vector is used for representing the running safety of the flat car on the complex road surface, and running deviation calculation of the flat car is achieved. Meanwhile, expected correction data are obtained, motion parameters of the flat car are optimized and adjusted, and therefore high-precision control over the flat car under the complex road surface condition is achieved.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving, and more specifically, to a high-precision control method and system for a flatbed truck under complex road conditions. Background Art

[0002] With the rapid development of autonomous driving technology, flatbed trucks are often used to simulate various driving scenarios to evaluate and verify the performance of autonomous driving algorithms. Therefore, the research on high-precision control methods is particularly important. Under complex road conditions, such as slippery, rugged, muddy or snow-covered roads, the dynamic behavior of flatbed trucks will be significantly affected, and the vehicle's grip, stability and handling will be severely deformed. Traditional control methods and systems often have difficulty adapting to these changes.

[0003] The prior art discloses a control method, device and system for a flatbed vehicle, which includes: obtaining motion data representing the motion of the flatbed vehicle; determining whether the value of the motion data is equal to a preset threshold; when the value of the motion data is not equal to the preset threshold, generating difference data between the value of the motion data and the preset threshold; and generating a control signal for the flatbed vehicle according to the difference data. However, this method will still be significantly affected on complex roads, and the vehicle's grip, stability and handling will be seriously deformed. Summary of the invention

[0004] The purpose of the present invention is to disclose a high-precision control method and system for a flatbed truck under complex road conditions with better control effects.

[0005] In order to achieve the above object, the present invention provides a high-precision control method and system for a flatbed vehicle under complex road conditions, including:

[0006] S1: Collect sensor data of the flatbed truck and obtain the moving speed of the flatbed truck according to the sensor data of the flatbed truck;

[0007] S2: obtaining the state data of the flatbed vehicle according to the moving speed of the flatbed vehicle and the sensor data of the flatbed vehicle, and obtaining the deviation vector according to the state data of the flatbed vehicle;

[0008] S3: constructing a flatbed vehicle stability control model, which obtains expected correction data according to the deviation vector;

[0009] S4: Optimize the motion parameters of the flatbed vehicle according to the expected correction data and the sensor data of the flatbed vehicle to obtain the optimized motion parameters of the flatbed vehicle; and optimally control the flatbed vehicle according to the optimized motion parameters of the flatbed vehicle.

[0010] Furthermore, in step S1, collecting the flatbed vehicle sensor data includes:

[0011] A variety of sensors are deployed on the flatbed truck, and the sensors are used to collect the driving parameter data of the flatbed truck during driving to form the flatbed truck sensor data, where the sensors include a wheel speed sensor, a steering wheel angle sensor and an angular velocity sensor. The wheel speed sensor outputs a pulse signal between the wheel ring gear frequency and the wheel speed to obtain the wheel speed; the steering wheel angle sensor outputs the steering wheel angle of the flatbed truck, and the angular velocity outputs the yaw angular velocity of the flatbed truck; the flatbed truck sensor data includes: wheel speed, steering wheel angle and yaw angular velocity.

[0012] Further, in step S1, obtaining the moving speed of the flatbed vehicle according to the flatbed vehicle sensor data includes:

[0013] According to the sensor data of the flatbed truck, the wheel speed of different wheels in the flatbed truck at any time t is calculated, where the wheel speed of the nth wheel in the flatbed truck at any time t is

[0014] According to the wheel speeds of the N wheels of the flatbed truck at any time t, the speed of the flatbed truck at any time t is estimated.

[0015] Further, in step S1, estimating the speed of the flatbed vehicle at any time t according to the wheel speeds of N wheels of the flatbed vehicle at any time t includes:

[0016] S1.1: Based on the flatbed truck sensor data, calculate the wheel speed weights of different wheels at any time t

[0017] S1.2: Divide the N wheels into front wheels and rear wheels, and calculate the average turning angle of the front wheels of the flatbed truck at any time t:

[0018]

[0019] in:

[0020] represents the average front wheel turning angle of the flatbed truck at any time t, and ε represents the total steering transmission ratio;

[0021] α t represents the steering wheel angle of the flatbed truck at any time t;

[0022] S1.3: weighting the wheel speeds of the N wheels of the flatbed truck at any time t according to the wheel speed weights, and calculating the average weighted wheel speed of the front wheels and the average weighted wheel speed of the rear wheels;

[0023] Calculate the turning radius of the front wheel of the flatbed truck at any time t And the rear wheel turning radius

[0024] S1.4: Calculate the speed of the flatbed truck at any time t in represents the horizontal speed of the flatbed truck at any time t, It represents the vertical speed of the flatbed truck at any time t.

[0025] Furthermore, in step S2, it includes:

[0026] The state data includes the flatbed vehicle speed, the average front wheel turning angle, and the wheel speed. The flatbed vehicle state data at any time t is S t :

[0027]

[0028] Calculate the center of mass deviation and yaw rate difference of the flatbed truck, where the direction of the flatbed truck includes the steering wheel angle and the center of mass deviation of the flatbed truck, where the center of mass deviation of the flatbed truck at any time t is β t ;

[0029] The calculation process of the center of mass angle deviation and yaw rate difference of the flatbed vehicle at any time t is as follows:

[0030] S2.1: Calculate the ideal yaw rate of the flatbed vehicle at any time t

[0031]

[0032] in:

[0033] represents the ideal yaw rate of the flatbed truck at any time t;

[0034] Len represents the body length of the flatbed truck;

[0035] g represents the mass of the flatbed truck;

[0036] C1 represents the cornering stiffness of the front axle of the flatbed truck, and C2 represents the cornering stiffness of the rear axle of the flatbed truck;

[0037] L2 represents the distance from the rear axle of the flatbed to the center of mass of the flatbed;

[0038] S2.2: Calculate the ideal center of mass deviation angle of the flatbed truck at any time t:

[0039]

[0040] in:

[0041] represents the ideal center of mass deviation angle of the flatbed truck at any time t;

[0042] S2.3: The mass center deviation angle deviation of the flatbed vehicle at any time t is calculated based on the ideal mass center deviation angle and the mass center deviation angle; the yaw rate difference is calculated based on the ideal yaw rate and the yaw rate, and the deviation vector is constructed based on the yaw rate difference and the mass center deviation angle deviation in

[0043] Furthermore, in step S3, the flatbed vehicle stability control model includes:

[0044] The flatbed vehicle stability control model takes the deviation vector and the flatbed vehicle state data as input and takes the expected correction data as output. The flatbed vehicle state data includes the flatbed vehicle speed, the average front wheel turning angle and the wheel speed. The flatbed vehicle stability control model is a deep neural network, including: an input layer, a state mapping layer, a deviation correction layer and an output layer;

[0045] The output layer is used to receive the deviation vector and the flatbed car status data;

[0046] The state mapping layer is a convolutional neural network structure, which is used to map the flatbed vehicle state data into state features;

[0047] The deviation correction layer is used to correct the state characteristics using the deviation vector to obtain the corrected state characteristics;

[0048] The output layer is used to map the corrected state features into corrected flatbed car state data as expected correction data;

[0049] Generate the expected correction data using the flatbed vehicle stability control model.

[0050] Furthermore, the flatbed vehicle stability control model is used to generate expected correction data, including:

[0051] The flatbed vehicle stability control model is used to generate the expected correction data, where the deviation vector f of the flatbed vehicle at any time t is t The expected calibration data generation process is:

[0052] S3.1: The output layer receives the flatbed car deviation vector f t And the flatbed status data S t ;

[0053] S3.2: The state mapping layer converts the flatbed vehicle state data S t Mapped to state feature G t :

[0054]

[0055] in:

[0056] exp(·) represents an exponential function with a natural constant as the base;

[0057] ||·|| represents the L1 norm;

[0058] represents the convolution matrix in the state mapping layer, and * represents the convolution operator;

[0059] S.33: The bias correction layer uses the bias vector f t Correct the state feature to obtain the corrected state feature g t , where the correction formula is:

[0060]

[0061] in:

[0062] W f represents the correction processing matrix;

[0063] dim represents the state feature G t The dimension of

[0064] ||·||2 represents the L2 norm;

[0065] g t represents the corrected state characteristics;

[0066] S3.4: The output layer will correct the state feature g t Mapped to the corrected flatbed status data As expected calibration data:

[0067]

[0068] in:

[0069] Represents the connection matrix in the output layer.

[0070] Further, in step S4, the flatbed vehicle motion parameters are optimized according to the expected correction data and the flatbed vehicle sensor data to obtain the flatbed vehicle motion optimized parameters, including:

[0071] Correct the data according to expectations With flatbed truck sensor data S t The flatbed car motion parameters are optimized, where the flatbed car motion parameters are the flatbed car's power parameters, including horizontal power, vertical power, and the average front wheel angle. The flatbed car motion parameter optimization process is as follows:

[0072] S4.1: Calculate the motion parameters of the flatbed at any time t

[0073] S4.2: Correct data based on expectations Calculate the expected motion parameters of the flatbed at any time t Among them, the expected flatbed car motion parameters During the calculation of the speed, the expected correction data is used. The velocity data in the figure are horizontal displacement. and vertical displacement Correction data for expectations The average turning angle of the front wheels in

[0074] S4.3: Multiple coefficients of the flatbed vehicle motion parameters are calculated using a group optimization method, including the proportional coefficient Integral term coefficient And the coefficient of the differential term

[0075] S4.4: Combine multiple coefficients to calculate the optimal vector B of the flatbed motion parameters at any time t t ;

[0076] S4.5: Optimize vector B using the motion parameters of the flatbed truck at any time t t The motion parameters of the flatbed truck are optimized and adjusted to achieve high-precision control of the flatbed truck.

[0077] Furthermore, in step S4.3, it includes:

[0078] S4.3.1: Initialize and generate D groups of polynomial coefficients, where the dth group of polynomial coefficients generated by initialization is d∈[1,d];

[0079] S4.3.2: Set the current iteration number of the polynomial coefficients to k, the initial value of k is 0, and the maximum value is K. Then the kth iteration result of the dth group of polynomial coefficients is P k (d);

[0080] S4.3.3: Calculate the wheel speed correction factor τ:

[0081]

[0082] in:

[0083] σ t Represents expected calibration data The vector representing the N wheel speeds and the flatbed vehicle state data S t The standard deviation of the difference vectors between the vectors representing the N wheel speeds in ;

[0084] μ t Represents expected calibration data The vector representing the N wheel speeds and the flatbed vehicle state data S t The mean of the difference vectors between the vectors representing the N wheel speeds in ;

[0085] S4.3.4: Calculate the adaptive iteration weight w(k) of the kth iteration based on the wheel speed correction factor;

[0086] S4.3.5: Get Q group training data and construct a multi-coefficient evaluation function:

[0087]

[0088] in:

[0089] Loss(·) represents a multi-coefficient evaluation function, Loss(P k (d)) represents the multinomial coefficient P k (d) The value of the multi-coefficient evaluation function;

[0090] B(q) represents the optimized vector of the actual flatbed vehicle motion parameters in the qth group of training data, It means that the expected correction data and the flatbed vehicle status data in the qth group of training data are taken as input, based on the multi-factor coefficient P k (d) Optimizing the motion parameters of the flatbed vehicle and obtaining an optimized vector of the motion parameters of the flatbed vehicle;

[0091] The multinomial coefficient with the lowest multinomial coefficient evaluation function value after the kth iteration is selected as the optimal multinomial coefficient P obtained by the kth iteration. k (best);

[0092] S4.3.6: Iterate the multinomial coefficients, where the iteration formula is:

[0093] P k+1 (d) = P k (best)+w(k)|P k (d)-P k (best)|cos(2π·rand(0,1))

[0094] S4.3.7: Let k = k + 1, return to step S436, until the maximum number of iterations is reached, and select the multinomial coefficient with the lowest multinomial coefficient evaluation function value at this time as the multinomial coefficient of the flatbed car motion parameter

[0095] In addition, the present invention provides a high-precision control system for a flatbed vehicle under complex road conditions, comprising:

[0096] Acquisition module: collects sensor data of the flatbed truck and obtains the moving speed of the flatbed truck based on the sensor data of the flatbed truck;

[0097] Vector module: obtain the status data of the flatbed vehicle according to the vehicle speed and sensor data of the flatbed vehicle, and obtain the deviation vector according to the status data of the flatbed vehicle;

[0098] Model module: construct a flatbed vehicle stability control model, which obtains expected correction data based on the deviation vector;

[0099] Control module: optimize the motion parameters of the flatbed vehicle according to the expected correction data and the sensor data of the flatbed vehicle to obtain the optimized motion parameters of the flatbed vehicle; and optimally control the flatbed vehicle according to the optimized motion parameters of the flatbed vehicle.

[0100] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0101] The present invention obtains the flatbed vehicle state data according to the flatbed vehicle movement speed and the flatbed vehicle sensor data, obtains the deviation vector according to the flatbed vehicle state data, uses the deviation vector as a representation of the flatbed vehicle's driving safety on a complex road surface, and realizes the calculation of the flatbed vehicle's driving deviation. At the same time, the present invention obtains the expected correction data and optimizes and adjusts the flatbed vehicle's movement parameters, thereby realizing high-precision control of the flatbed vehicle under complex road conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] Figure 1 This is a flow chart of the high-precision control method for a flatbed vehicle under complex road conditions described in Example 1;

[0103] Figure 2 This is a block diagram of a high-precision control system for a flatbed truck under complex road conditions as described in Example 3; DETAILED DESCRIPTION

[0104] The drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0105] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0106] Embodiment 1:

[0107] This embodiment provides Figure 1 The high-precision control method and system of a flatbed truck under complex road conditions shown in the figure include:

[0108] S1: Collect sensor data of the flatbed truck and obtain the moving speed of the flatbed truck according to the sensor data of the flatbed truck;

[0109] S2: obtaining the state data of the flatbed vehicle according to the moving speed of the flatbed vehicle and the sensor data of the flatbed vehicle, and obtaining the deviation vector according to the state data of the flatbed vehicle;

[0110] S3: constructing a flatbed vehicle stability control model, which obtains expected correction data according to the deviation vector;

[0111] S4: Optimize the motion parameters of the flatbed vehicle according to the expected correction data and the sensor data of the flatbed vehicle to obtain the optimized motion parameters of the flatbed vehicle; and optimally control the flatbed vehicle according to the optimized motion parameters of the flatbed vehicle.

[0112] This embodiment obtains the flatbed vehicle state data according to the flatbed vehicle movement speed and the flatbed vehicle sensor data, obtains the deviation vector according to the flatbed vehicle state data, and uses the deviation vector as a representation of the flatbed vehicle's driving safety on a complex road surface to realize the calculation of the flatbed vehicle's driving deviation. At the same time, the present invention obtains the expected correction data and optimizes and adjusts the flatbed vehicle's movement parameters, thereby realizing high-precision control of the flatbed vehicle under complex road conditions.

[0113] Embodiment 2:

[0114] This embodiment further discloses on the basis of the first embodiment:

[0115] Furthermore, in step S1, collecting the flatbed vehicle sensor data includes:

[0116] A variety of sensors are deployed on the flatbed truck, and the sensors are used to collect the driving parameter data of the flatbed truck during driving to form the flatbed truck sensor data, where the sensors include a wheel speed sensor, a steering wheel angle sensor and an angular velocity sensor. The wheel speed sensor outputs a pulse signal between the wheel ring gear frequency and the wheel speed to obtain the wheel speed; the steering wheel angle sensor outputs the steering wheel angle of the flatbed truck, and the angular velocity outputs the yaw angular velocity of the flatbed truck; the flatbed truck sensor data includes: wheel speed, steering wheel angle and yaw angular velocity.

[0117] Further, in step S1, obtaining the moving speed of the flatbed vehicle according to the flatbed vehicle sensor data includes:

[0118] According to the sensor data of the flatbed truck, the wheel speed of different wheels in the flatbed truck at any time t is calculated, where the wheel speed of the nth wheel in the flatbed truck at any time t is

[0119] According to the wheel speeds of the N wheels of the flatbed truck at any time t, the speed of the flatbed truck at any time t is estimated.

[0120] Further, in step S1, estimating the speed of the flatbed vehicle at any time t according to the wheel speeds of N wheels of the flatbed vehicle at any time t includes:

[0121] S1.1: Based on the flatbed truck sensor data, calculate the wheel speed weights of different wheels at any time t

[0122] S1.2: Divide the N wheels into front wheels and rear wheels, and calculate the average turning angle of the front wheels of the flatbed truck at any time t:

[0123]

[0124] in:

[0125] represents the average front wheel turning angle of the flatbed truck at any time t, and ε represents the total steering transmission ratio;

[0126] α t represents the steering wheel angle of the flatbed truck at any time t;

[0127] S1.3: weighting the wheel speeds of the N wheels of the flatbed truck at any time t according to the wheel speed weights, and calculating the average weighted wheel speed of the front wheels and the average weighted wheel speed of the rear wheels;

[0128] Calculate the turning radius of the front wheel of the flatbed truck at any time t And the rear wheel turning radius

[0129] S1.4: Calculate the speed of the flatbed truck at any time t in represents the horizontal speed of the flatbed truck at any time t, It represents the vertical speed of the flatbed truck at any time t.

[0130] Furthermore, in step S2, it includes:

[0131] The state data includes the flatbed vehicle speed, the average front wheel turning angle, and the wheel speed. The flatbed vehicle state data at any time t is S t :

[0132]

[0133] Calculate the center of mass deviation and yaw rate difference of the flatbed truck, where the direction of the flatbed truck includes the steering wheel angle and the center of mass deviation of the flatbed truck, where the center of mass deviation of the flatbed truck at any time t is β t ;

[0134] The calculation process of the center of mass angle deviation and yaw rate difference of the flatbed vehicle at any time t is as follows:

[0135] S2.1: Calculate the ideal yaw rate of the flatbed vehicle at any time t

[0136]

[0137] in:

[0138] represents the ideal yaw rate of the flatbed truck at any time t;

[0139] Len represents the body length of the flatbed truck;

[0140] g represents the mass of the flatbed truck;

[0141] C1 represents the cornering stiffness of the front axle of the flatbed truck, and C2 represents the cornering stiffness of the rear axle of the flatbed truck;

[0142] L2 represents the distance from the rear axle of the flatbed to the center of mass of the flatbed;

[0143] S2.2: Calculate the ideal center of mass deviation angle of the flatbed truck at any time t:

[0144]

[0145] in:

[0146] represents the ideal center of mass deviation angle of the flatbed truck at any time t;

[0147] S2.3: The mass center deviation angle deviation of the flatbed vehicle at any time t is calculated based on the ideal mass center deviation angle and the mass center deviation angle; the yaw rate difference is calculated based on the ideal yaw rate and the yaw rate, and the deviation vector is constructed based on the yaw rate difference and the mass center deviation angle deviation in

[0148] Furthermore, in step S3, the flatbed vehicle stability control model includes:

[0149] The flatbed vehicle stability control model takes the deviation vector and the flatbed vehicle state data as input and takes the expected correction data as output. The flatbed vehicle state data includes the flatbed vehicle speed, the average front wheel turning angle and the wheel speed. The flatbed vehicle stability control model is a deep neural network, including: an input layer, a state mapping layer, a deviation correction layer and an output layer;

[0150] The output layer is used to receive the deviation vector and the flatbed car status data;

[0151] The state mapping layer is a convolutional neural network structure, which is used to map the flatbed vehicle state data into state features;

[0152] The deviation correction layer is used to correct the state characteristics using the deviation vector to obtain the corrected state characteristics;

[0153] The output layer is used to map the corrected state features into corrected flatbed car state data as expected correction data;

[0154] Generate the expected correction data using the flatbed vehicle stability control model.

[0155] Furthermore, the flatbed vehicle stability control model is used to generate expected correction data, including:

[0156] The flatbed vehicle stability control model is used to generate the expected correction data, where the deviation vector f of the flatbed vehicle at any time t is t The expected calibration data generation process is:

[0157] S3.1: The output layer receives the flatbed car deviation vector f t And the flatbed status data S t ;

[0158] S3.2: The state mapping layer converts the flatbed vehicle state data S t Mapped to state feature G t :

[0159]

[0160] in:

[0161] exp(·) represents an exponential function with a natural constant as the base;

[0162] ||·|| represents the L1 norm;

[0163] represents the convolution matrix in the state mapping layer, and * represents the convolution operator;

[0164] S.33: The bias correction layer uses the bias vector f t Correct the state feature to obtain the corrected state feature g t , where the correction formula is:

[0165]

[0166] in:

[0167] W f represents the correction processing matrix;

[0168] dim represents the state feature G t The dimension of

[0169] ||·||2 represents the L2 norm;

[0170] g t represents the corrected state characteristics;

[0171] S3.4: The output layer will correct the state feature g t Mapped to the corrected flatbed status data As expected calibration data:

[0172]

[0173] in:

[0174] Represents the connection matrix in the output layer.

[0175] Further, in step S4, the flatbed vehicle motion parameters are optimized according to the expected correction data and the flatbed vehicle sensor data to obtain the flatbed vehicle motion optimized parameters, including:

[0176] Correct the data according to expectations With flatbed truck sensor data S t The flatbed car motion parameters are optimized, where the flatbed car motion parameters are the flatbed car's power parameters, including horizontal power, vertical power, and the average front wheel angle. The flatbed car motion parameter optimization process is as follows:

[0177] S4.1: Calculate the motion parameters of the flatbed at any time t

[0178] S4.2: Correct data based on expectations Calculate the expected motion parameters of the flatbed at any time t Among them, the expected flatbed car motion parameters During the calculation of the speed, the expected correction data is used. The velocity data in the figure are horizontal displacement. and vertical displacement Correction data for expectations The average turning angle of the front wheels in

[0179] S4.3: Multiple coefficients of the flatbed vehicle motion parameters are calculated using a group optimization method, including the proportional coefficient Integral term coefficient And the coefficient of the differential term

[0180] S4.4: Combine multiple coefficients to calculate the optimal vector B of the flatbed motion parameters at any time t t ;

[0181] S4.5: Optimize vector B using the motion parameters of the flatbed truck at any time t t The motion parameters of the flatbed truck are optimized and adjusted to achieve high-precision control of the flatbed truck.

[0182] Furthermore, in step S4.3, it includes:

[0183] S4.3.1: Initialize and generate D groups of polynomial coefficients, where the dth group of polynomial coefficients generated by initialization is d∈[1,d];

[0184] S4.3.2: Set the current iteration number of the polynomial coefficients to k, the initial value of k is 0, and the maximum value is K. Then the kth iteration result of the dth group of polynomial coefficients is P k (d);

[0185] S4.3.3: Calculate the wheel speed correction factor τ:

[0186]

[0187] in:

[0188] σ t Represents expected calibration data The vector representing the N wheel speeds and the flatbed vehicle state data S t The standard deviation of the difference vectors between the vectors representing the N wheel speeds in ;

[0189] μ t Represents expected calibration data The vector representing the N wheel speeds and the flatbed vehicle state data S t The mean of the difference vectors between the vectors representing the N wheel speeds in ;

[0190] S4.3.4: Calculate the adaptive iteration weight w(k) of the kth iteration based on the wheel speed correction factor;

[0191] S4.3.5: Get Q group training data and construct a multi-coefficient evaluation function:

[0192]

[0193] in:

[0194] Loss(·) represents a multi-coefficient evaluation function, Loss(P k (d)) represents the multinomial coefficient P k (d) The value of the multi-coefficient evaluation function;

[0195] B(q) represents the optimized vector of the actual flatbed vehicle motion parameters in the qth group of training data, It means that the expected correction data and the flatbed vehicle status data in the qth group of training data are taken as input, based on the multi-factor coefficient P k (d) Optimizing the motion parameters of the flatbed vehicle and obtaining an optimized vector of the motion parameters of the flatbed vehicle;

[0196] The multinomial coefficient with the lowest multinomial coefficient evaluation function value after the kth iteration is selected as the optimal multinomial coefficient P obtained by the kth iteration. k (best);

[0197] S4.3.6: Iterate the multinomial coefficients, where the iteration formula is:

[0198] P k+1 (d) = P k (best)+w(k)|P k (d)-P k (best)|cos(2π·rand(0,1))

[0199] S4.3.7: Let k = k + 1, return to step S436, until the maximum number of iterations is reached, and select the multinomial coefficient with the lowest multinomial coefficient evaluation function value at this time as the multinomial coefficient of the flatbed car motion parameter

[0200] This embodiment obtains the flatbed vehicle state data according to the flatbed vehicle movement speed and the flatbed vehicle sensor data, obtains the deviation vector according to the flatbed vehicle state data, and uses the deviation vector as a representation of the flatbed vehicle's driving safety on a complex road surface to realize the calculation of the flatbed vehicle's driving deviation. At the same time, the present invention obtains the expected correction data and optimizes and adjusts the flatbed vehicle's movement parameters, thereby realizing high-precision control of the flatbed vehicle under complex road conditions.

[0201] Embodiment three:

[0202] This embodiment provides Figure 2 The high-precision control system for flatbed vehicles under complex road conditions shown in the figure includes:

[0203] Acquisition module: collects sensor data of the flatbed truck and obtains the moving speed of the flatbed truck based on the sensor data of the flatbed truck;

[0204] Vector module: obtain the status data of the flatbed vehicle according to the vehicle speed and sensor data of the flatbed vehicle, and obtain the deviation vector according to the status data of the flatbed vehicle;

[0205] Model module: construct a flatbed vehicle stability control model, which obtains expected correction data based on the deviation vector;

[0206] Control module: optimize the motion parameters of the flatbed vehicle according to the expected correction data and the sensor data of the flatbed vehicle to obtain the optimized motion parameters of the flatbed vehicle; and optimally control the flatbed vehicle according to the optimized motion parameters of the flatbed vehicle.

[0207] This embodiment obtains the flatbed vehicle state data according to the flatbed vehicle movement speed and the flatbed vehicle sensor data, obtains the deviation vector according to the flatbed vehicle state data, and uses the deviation vector as a representation of the flatbed vehicle's driving safety on a complex road surface to realize the calculation of the flatbed vehicle's driving deviation. At the same time, the present invention obtains the expected correction data and optimizes and adjusts the flatbed vehicle's movement parameters, thereby realizing high-precision control of the flatbed vehicle under complex road conditions.

[0208] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A high-precision control method for a flatbed truck under complex road conditions, characterized in that: include: S1: Collect sensor data of the flatbed truck and obtain the moving speed of the flatbed truck according to the sensor data of the flatbed truck; S2: obtaining the state data of the flatbed vehicle according to the moving speed of the flatbed vehicle and the sensor data of the flatbed vehicle, and obtaining the deviation vector according to the state data of the flatbed vehicle; S3: constructing a flatbed vehicle stability control model, which obtains expected correction data according to the deviation vector; S4: Optimize the motion parameters of the flatbed vehicle according to the expected correction data and the sensor data of the flatbed vehicle to obtain the optimized motion parameters of the flatbed vehicle; and optimally control the flatbed vehicle according to the optimized motion parameters of the flatbed vehicle.

2. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 1 is characterized in that: In step S1, collecting flatbed vehicle sensor data includes: A variety of sensors are deployed on the flatbed truck, and the sensors are used to collect the driving parameter data of the flatbed truck during driving to form the flatbed truck sensor data, where the sensors include a wheel speed sensor, a steering wheel angle sensor and an angular velocity sensor. The wheel speed sensor outputs a pulse signal between the wheel ring gear frequency and the wheel speed to obtain the wheel speed; the steering wheel angle sensor outputs the steering wheel angle of the flatbed truck, and the angular velocity outputs the yaw angular velocity of the flatbed truck; the flatbed truck sensor data includes: wheel speed, steering wheel angle and yaw angular velocity.

3. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 1 is characterized in that: In step S1, obtaining the moving speed of the flatbed vehicle according to the sensor data of the flatbed vehicle includes: According to the sensor data of the flatbed truck, the wheel speed of different wheels in the flatbed truck at any time t is calculated, where the wheel speed of the nth wheel in the flatbed truck at any time t is According to the wheel speeds of the N wheels of the flatbed truck at any time t, the speed of the flatbed truck at any time t is estimated.

4. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 3 is characterized in that: In step S1, estimating the speed of the flatbed vehicle at any time t according to the wheel speeds of the N wheels of the flatbed vehicle at any time t includes: S1.1: Based on the flatbed truck sensor data, calculate the wheel speed weights of different wheels at any time t S1.2: Divide the N wheels into front wheels and rear wheels, and calculate the average turning angle of the front wheels of the flatbed truck at any time t: in: represents the average front wheel turning angle of the flatbed truck at any time t, and ε represents the total steering transmission ratio; α t represents the steering wheel angle of the flatbed truck at any time t; S1.3: weighting the wheel speeds of the N wheels of the flatbed truck at any time t according to the wheel speed weights, and calculating the average weighted wheel speed of the front wheels and the average weighted wheel speed of the rear wheels; Calculate the turning radius of the front wheel of the flatbed truck at any time t And the rear wheel turning radius S1.4: Calculate the speed of the flatbed truck at any time t in represents the horizontal speed of the flatbed truck at any time t, It represents the vertical speed of the flatbed truck at any time t.

5. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 2 is characterized in that: In step S2, it includes: The state data includes the flatbed vehicle speed, the average front wheel turning angle, and the wheel speed. The flatbed vehicle state data at any time t is S t : Calculate the center of mass deviation and yaw rate difference of the flatbed truck, where the direction of the flatbed truck includes the steering wheel angle and the center of mass deviation of the flatbed truck, where the center of mass deviation of the flatbed truck at any time t is β t ; The calculation process of the center of mass angle deviation and yaw rate difference of the flatbed vehicle at any time t is as follows: S2.1: Calculate the ideal yaw rate of the flatbed vehicle at any time t in: represents the ideal yaw rate of the flatbed truck at any time t; Len represents the body length of the flatbed truck; g represents the mass of the flatbed truck; C1 represents the cornering stiffness of the front axle of the flatbed truck, and C2 represents the cornering stiffness of the rear axle of the flatbed truck; L2 represents the distance from the rear axle of the flatbed to the center of mass of the flatbed; S2.2: Calculate the ideal center of mass deviation angle of the flatbed truck at any time t: in: represents the ideal center of mass deviation angle of the flatbed truck at any time t; S2.3: The mass center deviation angle deviation of the flatbed vehicle at any time t is calculated based on the ideal mass center deviation angle and the mass center deviation angle; the yaw rate difference is calculated based on the ideal yaw rate and the yaw rate, and the deviation vector is constructed based on the yaw rate difference and the mass center deviation angle deviation in 6. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 1 is characterized in that: In step S3, the flatbed vehicle stability control model includes: The flatbed vehicle stability control model takes the deviation vector and the flatbed vehicle state data as input and takes the expected correction data as output. The flatbed vehicle state data includes the flatbed vehicle speed, the average front wheel turning angle and the wheel speed. The flatbed vehicle stability control model is a deep neural network, including: an input layer, a state mapping layer, a deviation correction layer and an output layer; The output layer is used to receive the deviation vector and the flatbed car status data; The state mapping layer is a convolutional neural network structure, which is used to map the flatbed vehicle state data into state features; The deviation correction layer is used to correct the state features using the deviation vector to obtain the corrected state features; The output layer is used to map the corrected state features into corrected flatbed car state data as expected correction data; Generate the expected correction data using the flatbed vehicle stability control model.

7. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 6 is characterized in that: Generate the expected correction data using the flatbed vehicle stability control model, including: The flatbed vehicle stability control model is used to generate the expected correction data, where the deviation vector f of the flatbed vehicle at any time t is t The expected calibration data generation process is: S3.1: The output layer receives the flatbed car deviation vector f t And the flatbed status data S t ; S3.2: The state mapping layer converts the flatbed vehicle state data S t Mapped to state feature G t : in: exp(·) represents an exponential function with a natural constant as the base; ||·|| represents the L1 norm; represents the convolution matrix in the state mapping layer, and * represents the convolution operator; S.33: The bias correction layer uses the bias vector f t Correct the state feature to obtain the corrected state feature g t , where the correction formula is: in: W t represents the correction processing matrix; dim represents the state feature G t The dimension of ||·||2 represents the L2 norm; g t represents the corrected state characteristics; S3.4: The output layer will correct the state feature g t Mapped to the corrected flatbed status data As expected calibration data: in: Represents the connection matrix in the output layer.

8. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 1 is characterized in that: In step S4, the flatbed vehicle motion parameters are optimized according to the expected correction data and the flatbed vehicle sensor data to obtain the optimized parameters of the flatbed vehicle motion, including: Correct the data according to expectations With flatbed truck sensor data S t The flatbed car motion parameters are optimized, where the flatbed car motion parameters are the flatbed car's power parameters, including horizontal power, vertical power, and the average front wheel angle. The flatbed car motion parameter optimization process is as follows: S4.1: Calculate the motion parameters of the flatbed at any time t S4.2: Correct data based on expectations Calculate the expected motion parameters of the flatbed at any time t Among them, the expected flatbed car motion parameters During the calculation of the speed, the expected correction data is used. The velocity data in the figure are horizontal displacement. and vertical displacement Correction data for expectations The average turning angle of the front wheels in S4.3: Multiple coefficients of the flatbed vehicle motion parameters are calculated using a group optimization method, including the proportional coefficient Integral term coefficient And the differential coefficient S4.4: Combining multiple coefficients to calculate the flatbed motion parameters optimization vector B of the flatbed car at any time t t ; S4.5: Optimize vector B using the motion parameters of the flatbed truck at any time t t The motion parameters of the flatbed truck are optimized and adjusted to achieve high-precision control of the flatbed truck.

9. The high-precision control method for a flatbed vehicle under complex road conditions according to claim 8, characterized in that: In step S4.3, it includes: S4.3.1: Initialize and generate D groups of polynomial coefficients, where the dth group of polynomial coefficients generated by initialization is d∈[1,d]; S4.3.2: Set the current iteration number of the polynomial coefficients to k, the initial value of k is 0, and the maximum value is K. Then the k-th iteration result of the d-th group of polynomial coefficients is P k (d); S4.3.3: Calculate the wheel speed correction factor τ: in: σ t Represents expected calibration data The vector representing the N wheel speeds and the flatbed vehicle state data S t The standard deviation of the difference vectors between the vectors representing the N wheel speeds in ; μ t Represents expected calibration data The vector representing the N wheel speeds and the flatbed vehicle state data S t The mean of the difference vectors between the vectors representing the N wheel speeds in ; S4.3.4: Calculate the adaptive iteration weight w(k) of the kth iteration based on the wheel speed correction factor; S4.3.5: Get Q group training data and construct a multi-coefficient evaluation function: in: Loss(·) represents a multi-coefficient evaluation function, Loss(P k (d)) represents the multinomial coefficient P k (d) The value of the multi-coefficient evaluation function; B(q) represents the optimized vector of the actual flatbed vehicle motion parameters in the qth group of training data, It means that the expected correction data and the flatbed vehicle status data in the qth group of training data are taken as input, based on the multi-factor coefficient P k (d) Optimizing the motion parameters of the flatbed vehicle and obtaining an optimized vector of the motion parameters of the flatbed vehicle; The multinomial coefficient with the lowest multinomial coefficient evaluation function value after the kth iteration is selected as the optimal multinomial coefficient P obtained by the kth iteration. k (best); S4.3.6: Iterate the multinomial coefficients, where the iteration formula is: P k+1 (d)=P k (best)+w(k)|P k (d)-P k (best)|cos(2π·rand(0,1)) S4.3.7: Let k = k + 1, return to step S436, until the maximum number of iterations is reached, and select the multinomial coefficient with the lowest multinomial coefficient evaluation function value at this time as the multinomial coefficient of the flatbed car motion parameter 10. The high-precision control system for flatbed vehicles under complex road conditions is characterized by: include: Acquisition module: collects sensor data of the flatbed truck and obtains the moving speed of the flatbed truck based on the sensor data of the flatbed truck; Vector module: obtain the status data of the flatbed vehicle according to the vehicle speed and sensor data of the flatbed vehicle, and obtain the deviation vector according to the status data of the flatbed vehicle; Model module: construct a flatbed vehicle stability control model, which obtains expected correction data based on the deviation vector; Control module: optimize the motion parameters of the flatbed vehicle according to the expected correction data and the sensor data of the flatbed vehicle to obtain the optimized motion parameters of the flatbed vehicle; and optimally control the flatbed vehicle according to the optimized motion parameters of the flatbed vehicle.

Images