Intelligent multi-wheel drive device and control method in large pipelines

Abstract

An intelligent multi-wheel drive device and control method in a large pipeline, which includes an operation cabin set in the main structure of the mobile device, an operation device controller located at one end of the main structure of the mobile device, multiple mounting seats set on the main structure of the mobile device and connected to a telescopic hydraulic rod, a hub motor drive wheel is set at the telescopic end of the telescopic hydraulic rod, and a shock-absorbing compensation spring is set between the telescopic hydraulic rod and the hub motor drive wheel; multiple support mechanisms are set at one end of the main structure of the mobile device. The bearing capacity of the entire platform and the total traction of the platform are improved by adopting a multi-drive wheel drive method; the vertical load of the pipeline robot tire is calculated by force analysis of the pipeline machine wheel and the adhesion coefficient of the pipeline inner wall is estimated by using a double unscented Kalman filter algorithm, and the pipeline welding robot is driven and anti-skid controlled by using fuzzy PID control, which effectively improves the driving ability and positioning accuracy of the pipeline robot on the smooth inner wall of the pipeline.

Description

Technical Field

[0001] The invention belongs to the field of robots and control technology, and relates to an intelligent multi-wheel driving device in a large pipeline and a control method. Background Art

[0002] In the pipeline construction industry, especially in the construction of long-distance oil and gas pipelines, welding technology plays a vital role. However, with the continuous advancement of science and technology, traditional manual welding can no longer meet the needs of modern machinery manufacturing. The emergence of intelligent welding technology makes the welding process more precise, efficient and reliable. When welding pressure steel pipes, the construction progress is often slow due to the complexity of the construction environment and the shortage of welding workers. Traditional pipeline welding generally uses large scaffolding, which is difficult to construct and has a heavy labor burden, low efficiency and low intelligence.

[0003] Currently, there is a lack of mobile equipment that supports pressure pipe welding in scenarios involving large steel pipes (radial length of 6 meters or more). The welding of large pressure steel pipes still relies on manual welding. At the same time, in the field of hydropower station construction, the installation and construction of inclined shaft pressure steel pipes has always adopted traditional construction methods. Traditional construction methods are restricted by factors such as construction site conditions and construction period, and the degree of digital and intelligent construction is relatively low.

[0004] The driving force of the existing pipeline robot is insufficient, especially when heavy operations (such as grinding) are required, the existing pipeline robot often cannot provide sufficient traction or carrying capacity. For example, when some robots pass through a bend, the contact point between the driving wheel and the inner wall of the pipe is inconsistent, causing the driving wheel to slip, thereby affecting the stability of the robot and increasing the motor load. In addition, when operating in a complex pipeline environment, its positioning accuracy is often not accurate enough. This is mainly because the control method of the driving device is complex and the real-time performance is poor, resulting in a decrease in the robot's positioning and obstacle-crossing capabilities when encountering irregular pipe walls or turns. In addition, the existing pipeline robot driving device has a complex structure and a cumbersome control method. For example, some robots use a gear differential to achieve drive adjustment, but this design is not only large in size, but also has a high manufacturing cost. Finally, many pipeline robots fail to fully consider the carrying capacity during design, resulting in poor performance when performing heavy tasks. For example, although some robots can adapt to pipes with larger diameters, their carrying capacity is obviously insufficient when they need to carry a large load for operation. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an intelligent multi-wheel drive device and a control method in a large pipeline, which can adapt to the changeable and complex pipeline environment and complete various tasks performed in the pipeline.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: an intelligent multi-wheel drive device in a large pipeline, which includes an operating cabin arranged in the main structure of the mobile device, the operating device controller is located at one end of the main structure of the mobile device, a plurality of mounting seats are arranged on the main structure of the mobile device and connected to the retractable hydraulic rod, a hub motor drive wheel is arranged at the retractable end of the retractable hydraulic rod, and a shock-absorbing compensation spring is arranged between the retractable hydraulic rod and the hub motor drive wheel; a plurality of supporting mechanisms are arranged at one end of the main structure of the mobile device.

[0007] At least two hub motor driving wheels are arranged in front and behind the main structure of the mobile device, and are symmetrically distributed at 60 degrees; at least two symmetrically distributed auxiliary wheels are arranged in the middle of the main structure of the mobile device.

[0008] A motor connected to a reducer is arranged in the housing of the hub motor driving wheel, a brake is arranged at the output end of the motor, an encoder located in the housing is electrically connected to the driver and the motor; and the driver is electrically connected to the working device controller.

[0009] In the driving wheel control method of the intelligent multi-wheel driving device in a large pipeline as described above, the wheel radius of the hub motor driving wheel (1) is r. Assuming that the received driving torque is T, the wheel will generate a force in the opposite direction of driving on the ground. According to Newton's third law, at this time, the ground will generate a reaction force F on the wheel X , i.e. the longitudinal force of the wheel; The dynamic model of the driving wheel is derived by analyzing the force on each wheel. Considering the low driving speed and ignoring the air resistance, the following is obtained based on the balance of torque: ; ; In the formula, is the wheel longitudinal force, N; is the motor output torque, N·m; is the wheel speed, J w is the equivalent moment of inertia of the rotating part of the wheel hub, kg·m 2 ; is the vertical interference term; According to the kinetic energy theorem of theoretical mechanics, when various energy losses are ignored, the total kinetic energy of the equivalent component is equal to the sum of the kinetic energy of each part, which can be expressed as follows: ; Where J s is the moment of inertia of each rotating component in the wheel rim, kg·m 2 ; is the speed of each rotating component; The tire is the part that directly contacts the inner wall of the pipe with the mobile platform. It mainly transmits the force between the welding robot and the inner wall of the pipe, which is of great significance for stable driving and accurate positioning. Compared with the magic tire model, the Dugoff tire model has the advantages of fewer parameters and less calculation. Therefore, the Dugoff tire model is used to characterize the interaction between the tire and the inner wall of the pipe. The model expression is: ; ; in is the longitudinal force of the wheel; is the wheel lateral force; is the longitudinal stiffness of the tire; is the lateral stiffness of the tire; is the friction coefficient between the wheel and the inner wall of the pipe; is the vertical load on the tire; is the longitudinal slip rate; is the slip angle (the angle between the actual direction of movement of the tire and the direction in which it is pointed); is the correction function; the tire model is transformed into a normalized form: ; The input of the tire model includes the slip rate, sideslip angle and vertical load of each tire. The sideslip angle and slip rate of each wheel are calculated as follows: ; ; The vertical load of each wheel can be calculated from the longitudinal and lateral accelerations as follows: ; Where h is the height of the vehicle's center of mass; The double unscented Kalman filter (DUKF) algorithm is used to realize the real-time estimation of the adhesion coefficient of the pipeline welding robot. Combined with the estimated values ​​of the velocity and acceleration of the pipeline welding robot, the UKF road adhesion coefficient observer is used to realize the real-time estimation of the adhesion coefficient of each wheel of the pipeline welding robot. Construct the adhesion coefficient observer of the inner wall of the welded pipe and select is the system state vector, is the system observation vector, is the system control quantity, and the nonlinear discrete system expression of the adhesion coefficient of the inner wall of the welded pipe is: ; ; ; Where: ; ; ; .

[0010] The unscented Kalman filter uses sampling methods to construct a point set with constant mean and covariance to approximate a nonlinear system. In essence, it replaces the nonlinear system itself with an approximation of the probability density of the nonlinear system, making it suitable for the standard Kalman filter structure under the linear assumption, including: UT transformation, Based on the traceless transformation principle, dimensional state vector , using the symmetric sampling method, construct indivual Point Set: ; In the above formula, for Sampling point state quantity, , are the mean and covariance of the input state quantities at past moments, express The matrix List, is the scaling factor, usually represents the proportionality coefficient, Representation and The scale factor related to the sampling point distribution is taken here ; Calculate each Sampling point weights: ; In the formula, , For the The mean and covariance weights of the sampling points; Represents a non-negative weight value, here we take .

[0011] It also includes time updates; based on The state prediction update of the sampling point at the moment Moment state mean and covariance: ; ; ; Use UT transformation again to construct a new Sampling point set: .

[0012] Also included are measurement updates; The newly constructed Substituting the point set into the system observation equation, we can get the prior prediction value of the observation vector: ; Further calculate the mean and covariance matrix of the prior prediction value: ; ; Compute the cross-correlation covariance matrix: ; Calculate the Kalman gain, complete the a posteriori estimate of the system's current state and update the covariance matrix: ; ; .

[0013] The anti-skid control of the wheel hub motor drive wheel adopts fuzzy PID control, taking the optimal slip rate of each wheel currently attached to the road surface as the control target to achieve anti-skid control at the single wheel level; the actual slip rate of each wheel is calculated based on the wheel corrected wheel speed and the longitudinal vehicle speed estimated by the vehicle speed estimation module: ; The slip rate deviation of each wheel is calculated based on the optimal slip rate estimated by the pipeline welding robot and the actual slip rate: ; In the above formula, is the slip rate deviation of each wheel; is the optimal slip rate corresponding to each wheel; the target value of the driving torque of each wheel in the driving anti-skid control system is: ; In the above formula, is the target torque of each wheel; , and They are proportional coefficient, integral coefficient and differential coefficient respectively; is the slip rate deviation of this cycle; is the slip rate deviation of the previous cycle.

[0014] Combining fuzzy control with PID control, the parameter adjustment experience of fuzzy inference rules is used to adjust the PID control parameters and improve the robustness of the control system.

[0015] The error between the actual slip rate and the optimal slip rate and its rate of change As the input of the fuzzy controller, after fuzzification, fuzzy rule reasoning and defuzzification operations, the control parameters of the PID controller are output. , and

[0016] The main beneficial effects of the present invention are: By adopting a multi-drive wheel drive method, the carrying capacity of the entire platform and the total traction of the platform are improved. This method not only effectively improves the driving ability of the pipeline robot, but also effectively realizes the shape requirements of the welding pipeline, and improves its obstacle crossing ability and passability.

[0017] The problem of positioning accuracy was solved by adjusting the local driving wheels and the retractable hydraulic rod.

[0018] The vertical load of the pipeline robot tire is calculated by analyzing the force on the pipeline robot wheel, and the double unscented Kalman filter algorithm is used to estimate the adhesion coefficient of the pipeline inner wall. Finally, fuzzy PID control is used to drive the pipeline welding robot for anti-skid control, which effectively improves the driving ability and positioning accuracy of the pipeline robot on the smooth inner wall of the pipeline. The stability of the mobile platform is improved by adjusting the retractable hydraulic rod. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0020] Figure 1 It is a structural schematic diagram of the present invention.

[0021] Figure 2 It is a schematic diagram of the hub motor structure of the present invention.

[0022] Figure 3 This is a force analysis diagram of the wheel of the present invention.

[0023] Figure 4 This is the driving anti-skid control diagram of the present invention.

[0024] Markings in the figure: hub motor driving wheel 1; shock-absorbing compensation spring 2; retractable hydraulic rod 3; mounting seat 4; main structure of mobile device 5; operating compartment 6; supporting mechanism 7; operating device controller 8; pressure steel pipe 9; reducer 10; motor 11; brake 12; encoder 13; driver 14. DETAILED DESCRIPTION

[0025] like Figure 1~Figure 4In the invention, an intelligent multi-wheel drive device in a large pipeline is disclosed, which includes an operating cabin 6 arranged in a main structure 5 of the mobile device, an operating device controller 8 is located at one end of the main structure 5 of the mobile device, a plurality of mounting seats 4 are arranged on the main structure 5 of the mobile device and connected with a telescopic hydraulic rod 3, a hub motor driving wheel 1 is arranged at the telescopic end of the telescopic hydraulic rod 3, a shock-absorbing compensation spring 2 is arranged between the telescopic hydraulic rod 3 and the hub motor driving wheel 1; and a plurality of supporting mechanisms 7 are arranged at one end of the main structure 5 of the mobile device.

[0026] At least two hub motor driving wheels 1 are arranged in front and behind the main structure 5 of the mobile device, and are symmetrically distributed at 60 degrees; at least two symmetrically distributed auxiliary wheels are set in the middle of the main structure 5 of the mobile device.

[0027] Preferably, the two auxiliary wheels mainly function to absorb radial vibration and interference of the pipe wall during welding to ensure the stability of the device during welding.

[0028] A motor 11 connected to a reducer 10 is arranged in the housing of the hub motor driving wheel 1, a brake 12 is arranged at the output end of the motor 11, an encoder 13 located in the housing is electrically connected to a driver 14 and the motor 11; the driver 14 is electrically connected to the working device controller 8.

[0029] Preferably, the outer tire of the hub motor is made of polyurethane material, which can effectively resist particle wear and extend the service life. It has good grip, remains stable on slopes and uneven roads, and reduces the possibility of sliding. This feature enables the drive wheel to provide stable support under various ground conditions. In addition, the rated load-bearing mass of a single drive wheel of the drive device is 5t, and an auxiliary support device is designed to improve the stability of the entire device during automatic welding and ensure the welding quality.

[0030] Preferably, the device is a multi-wheel cooperatively driven intelligent mobile platform device, which enters the pipeline and actively adjusts the retractable hydraulic rod 3 to apply hydraulic pressure to make the driving wheel contact the inner wall of the pipeline. And provide sufficient support force for the driving wheel to prevent the wheel from slipping, so that it can move smoothly. After the mobile platform moves to the welding position, after adjustment to meet the positioning accuracy requirements, the retractable hydraulic rod 3 can be further adjusted to improve the overall stability of the mobile platform, and the support mechanism 7 is unfolded to further improve the stability. It is convenient to carry out subsequent welding work. When the welding work is completed, the support mechanism 7 is retracted, the retractable rod 3 is adjusted to apply pressure, and the hub motor drives the device to move to the next working point.

[0031] Preferably, the device is moved forward or backward in the pipeline by driving the wheel hub motor. By adjusting the retractable hydraulic rod 3, the mobile platform has a certain ability to adapt to changes in pipe diameter; the device also includes a sensor module, such as a pressure sensor, which cooperates with other mechanisms to achieve the adjustment of traction and the control of the clamping mechanism; by adjusting the speed of the wheel hub motor, the traction of the device is adjusted to adapt to different pipeline conditions and improve the ability to pass obstacles; the drive wheel unit adopts a modular design, which is easy to maintain and replace. At the same time, it is also convenient to upgrade and improve as needed; the controller of the wheel hub motor is connected to the main control system of the device, and receives control signals to adjust the motion state, such as speed and direction. The device can adjust its positioning according to the information received by various sensors carried on the vehicle body to meet the construction requirements. It can also be manually sent through the central control platform in the operation cabin 6 according to the actual situation to make local adjustments and improve the positioning accuracy. In summary, the drive module is able to adapt to the changing and complex pipeline environment and complete various tasks performed in the pipeline.

[0032] Embodiment 1, The distributed drive pipeline welding robot installs the hub motor in the wheel rim, which has a high degree of integration. The wheel radius of the hub motor driving wheel (1) is r. Assuming that the driving torque received is T, the wheel will generate a force in the opposite direction of driving on the ground. According to Newton's third law, the ground will generate a reaction force F on the wheel at this time. X , that is, the longitudinal force of the wheel. The force on the wheel can be seen in Figure 3 shown.

[0033] Figure 3 middle, is the wheel longitudinal force, N; is the vertical load of the wheel, N; T is the motor output torque, N·m; is the wheel speed, .

[0034] The dynamic model of the driving wheel is derived by analyzing the force on each wheel. Considering the low driving speed and ignoring the air resistance, the following is obtained based on the balance of torque: ; ; In the formula, is the wheel longitudinal force, N; is the motor output torque, N·m; is the wheel speed, ; J w is the equivalent moment of inertia of the rotating part of the wheel hub, kg·m 2 ; is the vertical interference term; According to the kinetic energy theorem of theoretical mechanics, when various energy losses are ignored, the total kinetic energy of the equivalent component is equal to the sum of the kinetic energy of each part, which can be expressed as follows: ; Where J s is the moment of inertia of each rotating component in the wheel rim, kg·m 2 ; is the speed of each rotating component. The tire is the part that directly contacts the inner wall of the pipe with the mobile platform. It mainly transmits the force between the welding robot and the inner wall of the pipe, which is of great significance for stable driving and accurate positioning. Compared with the magic tire model, the Dugoff tire model has the advantages of fewer parameters and less calculation. Therefore, the Dugoff tire model is used to characterize the interaction between the tire and the inner wall of the pipe. The model expression is: ; ; Transform the tire model into normalized form: ; in is the longitudinal force of the wheel; is the wheel lateral force; is the longitudinal stiffness of the tire; is the lateral stiffness of the tire; is the friction coefficient between the wheel and the inner wall of the pipe; is the vertical load on the tire; is the longitudinal slip rate; is the slip angle (the angle between the actual direction of movement of the tire and the direction in which it is pointed); is a correction function; the input of the tire model includes the slip rate, sideslip angle and vertical load of each tire. The calculation formula of the sideslip angle and slip rate of each wheel is as follows: ; ; The vertical load of each wheel can be calculated from the longitudinal and lateral accelerations as follows: ; Where h is the height of the vehicle's center of mass.

[0035] Embodiment 2, This patent adopts the double unscented Kalman filter (DUKF) algorithm to realize the real-time estimation of the adhesion coefficient of the pipeline welding robot's driving pipeline. Combined with the estimated values ​​of the speed and acceleration of the pipeline welding robot, the UKF road adhesion coefficient observer is used to realize the real-time estimation of the adhesion coefficient of each wheel of the pipeline welding robot.

[0036] Construct the adhesion coefficient observer of the inner wall of the welded pipe and select is the system state vector, is the system observation vector, is the system control quantity, and the nonlinear discrete system expression of the adhesion coefficient of the inner wall of the welded pipe is: ; ; ; Where: ; ; ; ; The unscented Kalman filter uses sampling methods to construct a point set with constant mean and covariance to approximate the nonlinear system. In essence, it replaces the nonlinear system itself by approximating the probability density of the nonlinear system, making it suitable for the standard Kalman filter structure under the linear assumption. The specific steps are as follows: (1) UT transformation Based on the traceless transformation principle, dimensional state vector , using the symmetric sampling method, construct indivual Point set: ; In the above formula, for Sampling point state quantity, , are the mean and covariance of the input state quantities at past moments, express The matrix List, is the scaling factor, usually represents the proportionality coefficient, Representation and The scale factor related to the sampling point distribution is taken here .

[0037] Calculate each Sampling point weights: ; In the formula, , For the The mean and covariance weights of the sampling points; Represents a non-negative weight value, here we take .

[0038] (2) Time update based on The state prediction update of the sampling point at the moment Moment state mean and covariance: ; ; ; Use UT transformation again to construct a new Sampling point set: ; (3) Measurement update The newly constructed Substituting the point set into the system observation equation, we can get the prior prediction value of the observation vector: ; Further calculate the mean and covariance matrix of the prior prediction value: ; ; Compute the cross-correlation covariance matrix: ; Calculate the Kalman gain, complete the a posteriori estimate of the system's current state and update the covariance matrix: ; ; ; Embodiment 3, The drive anti-skid control of this patent adopts fuzzy PID control, taking the optimal slip rate of each wheel's current adhesion road surface as the control target to achieve drive anti-skid control at the single-wheel level.

[0039] The PID control module of the target torque in the drive anti-slip control system is mainly affected by the proportional coefficient K p , integral coefficient K I , differential coefficient K D The adjustment of these three control parameters will significantly change the effect of the actual wheel slip rate following the slip rate target value in the drive anti-skid control. p The size of K determines the speed at which the actual wheel slip rate approaches the optimal slip rate. p When it is larger, it can significantly increase the response speed of the system, but it will also cause the slip rate to oscillate around the optimal slip rate, reducing the longitudinal adhesion performance of the tire. At the same time, the oscillation of the slip rate will cause the driving torque to tremble, reducing the smoothness of the vehicle's driving. Integral coefficient K IThe size of K determines the steady-state error between the actual wheel slip rate and the optimal slip rate. I A larger value will cause the system to respond more slowly. Differential coefficient K D The size of determines the overshoot of the actual slip rate. The target value of the driving torque is adjusted according to the change rate of the slip rate error to reduce the oscillation of the actual slip rate.

[0040] The actual slip rate of each wheel is calculated based on the wheel corrected speed and the longitudinal speed estimated by the vehicle speed estimation module: ; The slip rate deviation of each wheel is calculated based on the optimal slip rate estimated by the pipeline welding robot and the actual slip rate: ; In the above formula, is the slip rate deviation of each wheel; is the optimal slip rate corresponding to each wheel; the target value of the driving torque of each wheel in the driving anti-skid control system is: ; In the above formula, is the target torque of each wheel; , and They are proportional coefficient, integral coefficient and differential coefficient respectively; is the slip rate deviation of this cycle; is the slip rate deviation of the previous cycle.

[0041] Differential coefficient; is the slip rate deviation of this cycle; is the slip rate deviation of the previous cycle. Fuzzy control is mainly based on control system adjustment experience rules, suitable for nonlinear control systems, and has good stability. Combining fuzzy control with PID control and using the parameter adjustment experience of fuzzy inference rules to adjust PID control parameters is conducive to improving the robustness of the control system. and its rate of change As the input of the fuzzy controller, after fuzzification, fuzzy rule reasoning and defuzzification operations, the control parameters of the PID controller are output. , and The fuzzification and defuzzification use the triangular membership function, K , and The fuzzy rule table is as follows: Fuzzy rule table

[0042]

[0043]

[0044] The present invention improves the carrying capacity of the entire platform and the total traction of the platform by adopting a multi-drive wheel drive method. This method not only effectively improves the driving ability of the pipeline robot, but also effectively meets the shape requirements of the welding pipeline, and improves its obstacle crossing ability and passability. The problem of positioning accuracy is solved by adjusting the local drive wheel and the retractable hydraulic rod.

[0045] The vertical load of the pipeline robot tire is calculated by analyzing the force on the pipeline robot wheel, and the double unscented Kalman filter algorithm is used to estimate the adhesion coefficient of the pipeline inner wall. Finally, fuzzy PID control is used to drive the pipeline welding robot for anti-skid control, which effectively improves the driving ability and positioning accuracy of the pipeline robot on the smooth inner wall of the pipeline. The stability of the mobile platform is improved by adjusting the retractable hydraulic rod.

[0046] The above-mentioned embodiments are only preferred technical solutions of the present invention and should not be regarded as limitations of the present invention. The embodiments and features in the embodiments of the present application can be arbitrarily combined with each other without conflict. The protection scope of the present invention shall be the technical solutions recorded in the claims, including the equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. An intelligent multi-wheel drive device in a large pipeline, characterized by: The invention comprises an operating cabin (6) arranged in a main structure (5) of a mobile device, an operating device controller (8) located at one end of the main structure (5) of the mobile device, a plurality of mounting seats (4) arranged on the main structure (5) of the mobile device and connected to a telescopic hydraulic rod (3), a hub motor driving wheel (1) arranged at the telescopic end of the telescopic hydraulic rod (3), a shock absorbing compensation spring (2) arranged between the telescopic hydraulic rod (3) and the hub motor driving wheel (1); and a plurality of supporting mechanisms (7) arranged at one end of the main structure (5) of the mobile device.

2. The intelligent multi-wheel driving device in a large pipeline according to claim 1 is characterized in that: At least two of the hub motor driving wheels (1) are arranged at the front and rear of the main structure (5) of the mobile device, and are symmetrically distributed at 60 degrees; at least two symmetrically distributed auxiliary wheels are arranged in the middle of the main structure (5) of the mobile device.

3. The intelligent multi-wheel driving device in a large pipeline according to claim 1 is characterized in that: A motor (11) connected to a reducer (10) is arranged in the housing of the hub motor driving wheel (1); a brake (12) is arranged at the output end of the motor (11); an encoder (13) in the housing is electrically connected to a driver (14) and the motor (11); and the driver (14) is electrically connected to a working device controller (8).

4. The driving wheel control method of the intelligent multi-wheel driving device in a large pipeline according to claim 3 is characterized by: The wheel radius of the hub motor drive wheel (1) is r. Assuming the received driving torque is T, the wheel will generate a force in the opposite direction of the driving on the ground. According to Newton's third law, the ground will generate a reaction force F on the wheel at this time. X , i.e. the longitudinal force of the wheel; The dynamic model of the driving wheel is derived by analyzing the force on each wheel. Considering the low driving speed and ignoring the air resistance, the following is obtained based on the balance of torque: ； ； In the formula, is the wheel longitudinal force, N; is the motor output torque, N·m; is the wheel speed, ; J w is the equivalent moment of inertia of the rotating part of the wheel hub, kg·m 2 ; is the vertical interference term; According to the kinetic energy theorem of theoretical mechanics, when various energy losses are ignored, the total kinetic energy of the equivalent component is equal to the sum of the kinetic energy of each part, which is expressed as follows: ； Where J s is the moment of inertia of each rotating component in the wheel rim, kg·m 2 ; is the speed of each rotating component; The tire is the part that directly contacts the inner wall of the pipe with the mobile platform, and transmits the force between the welding robot and the inner wall of the pipe. The Dugoff tire model is used to characterize the interaction between the tire and the inner wall of the pipe. The model expression is: ； ； in is the longitudinal force of the wheel; is the wheel lateral force; is the longitudinal stiffness of the tire; is the lateral stiffness of the tire; is the friction coefficient between the wheel and the inner wall of the pipe; is the vertical load on the tire; is the longitudinal slip rate; is the slip angle (the angle between the actual direction of movement of the tire and the direction in which it is pointed); is the correction function; the tire model is transformed into a normalized form: ； The input of the tire model includes the slip rate, sideslip angle and vertical load of each tire. The sideslip angle and slip rate of each wheel are calculated as follows: ； ； The vertical load of each wheel can be calculated from the longitudinal and lateral accelerations as follows: ； Where h is the height of the vehicle's center of mass; The double unscented Kalman filter (DUKF) algorithm is used to realize the real-time estimation of the adhesion coefficient of the pipeline welding robot. Combined with the estimated values ​​of the velocity and acceleration of the pipeline welding robot, the UKF road adhesion coefficient observer is used to realize the real-time estimation of the adhesion coefficient of each wheel of the pipeline welding robot. Construct the adhesion coefficient observer of the inner wall of the welded pipe and select is the system state vector, is the system observation vector, is the system control quantity, and the nonlinear discrete system expression of the adhesion coefficient of the inner wall of the welded pipe is: ； ； ； Where: ； ； ； 。 5. The driving wheel control method of the intelligent multi-wheel driving device in a large pipeline according to claim 4 is characterized by: The unscented Kalman filter uses sampling methods to construct a point set with constant mean and covariance to approximate a nonlinear system. In essence, it replaces the nonlinear system itself with an approximation of the probability density of the nonlinear system, making it suitable for the standard Kalman filter structure under the linear assumption, including: UT transformation, Based on the traceless transformation principle, dimensional state vector , using the symmetric sampling method, construct indivual Point set: ； In the above formula, for Sampling point state quantity, , are the mean and covariance of the input state quantities at past moments, express The matrix List, is the scaling factor, usually represents the proportionality coefficient, Representation and The scale factor related to the sampling point distribution is taken here ; Calculate each Sampling point weights: ； In the formula, , For the The mean and covariance weights of the sampling points; Represents a non-negative weight value, here we take .

6. The driving wheel control method of the intelligent multi-wheel driving device in a large pipeline according to claim 5 is characterized by: It also includes time updates; based on The state prediction update of the sampling point at the moment Moment state mean and covariance: ； ； ； Use UT transformation again to construct a new Sampling point set: 。 7. The driving wheel control method of the intelligent multi-wheel driving device in a large pipeline according to claim 5 is characterized by: Also included are measurement updates; The newly constructed Substituting the point set into the system observation equation, we can get the prior prediction value of the observation vector: ； Further calculate the mean and covariance matrix of the prior prediction value: ； ； Compute the cross-correlation covariance matrix: ； Calculate the Kalman gain, complete the a posteriori estimate of the system's current state and update the covariance matrix: ； ； 。 8. The driving wheel control method of the intelligent multi-wheel driving device in a large pipeline according to any one of claims 1 to 7, characterized in that: The anti-slip control of the wheel hub motor driving wheel (1) adopts fuzzy PID control, taking the optimal slip rate of each wheel currently adhering to the road surface as the control target, and realizing the anti-slip control of the single wheel level; the actual slip rate of each wheel is calculated based on the wheel corrected wheel speed and the longitudinal vehicle speed estimated by the vehicle speed estimation module: ； The slip rate deviation of each wheel is calculated based on the optimal slip rate estimated by the pipeline welding robot and the actual slip rate: ； In the above formula, is the slip rate deviation of each wheel; is the optimal slip rate corresponding to each wheel; the target value of the driving torque of each wheel in the driving anti-skid control system is: ； In the above formula, is the target torque of each wheel; , and They are proportional coefficient, integral coefficient and differential coefficient respectively; is the slip rate deviation of this cycle; is the slip rate deviation of the previous cycle.

9. The driving wheel control method of the intelligent multi-wheel driving device in a large pipeline according to claim 8 is characterized by: Combining fuzzy control with PID control, the parameter adjustment experience of fuzzy inference rules is used to adjust the PID control parameters and improve the robustness of the control system.

10. The driving wheel control method of the intelligent multi-wheel driving device in a large pipeline according to claim 8, characterized in that: The error between the actual slip rate and the optimal slip rate and its rate of change As the input of the fuzzy controller, after fuzzification, fuzzy rule reasoning and defuzzification operations, the control parameters of the PID controller are output. , and .

Images