Electric Truck Steer-by-Wire System and Network Uncertainty Control Method Therefor

Abstract

An electric truck steer-by-wire system and a net-work uncertainty control method therefor. Said system comprises: a master control electric power module (14), a slave control electric power module (18), a road-sensing motor module (4), a steering wheel (1), an upper steering column (3), a lower steering column(10), a rack and pinion steering mechanism (12), wheels (13), a first steering angle sensor (2), a second steering angle sensor (11),a steering domain controller (8), and a vehicle-mounted CAN network (9). By means of the design of the master and slave controllers and actuating mechanisms, the system combines the advantages of steering angle tracking, torque tracking and current tracking, there-by satisfying the steering response requirements and power requirements of the electric truck, and further enhancing the robustness and fault-tolerant performance of the system under a random network time lag.

Claims

1. An electric truck steer-by-wire system, composed of a master control electric power module, a slave control electric power module, a road sensing motor module, a steering wheel, an upper steering column, a lower steering column, a rack-pinion steering gear, wheels, a first steering angle sensor, a second steering angle sensor, a steering domain controller and a vehicle-mounted CAN network; wherein the steering wheel is connected with a torque input end of the upper steering column by a spline, the first steering angle sensor is mounted between the steering wheel and the upper steering column, and the road sensing motor module is mounted at a lower end of the upper steering column; no mechanical connection exists between the upper steering column and the lower steering column, a gap exists between the upper steering column and the lower steering column in a vertical direction, the second steering angle sensor is mounted at an input end of the lower steering column, an output end of the lower steering column is connected with the rack-pinion steering gear, the master control electric power module and the slave control electric power module are mounted on a rack of the rack-pinion steering gear, and an end of the rack is connected with the wheels; the road sensing motor module consists of a road sensing controller, a first motor and a first deceleration mechanism; an output of the first motor is connected with an input of the first deceleration mechanism, and a torque is transmitted to the upper steering column after deceleration and torque is increased by the first deceleration mechanism; the road sensing controller controls the rotating speed of the first motor; the road sensing controller simulates a pavement steering torque according to an objective steering torque signal from the steering domain controller; the first steering angle sensor transmits a driver steering angle demand to the steering domain controller by the vehicle-mounted CAN network; the steering domain controller converts the driver steering angle demand into the objective steering torque signal and transmits the objective steering torque signal to the master control electric power module by the vehicle-mounted CAN network; the master control electric power module consists of a master controller, a second motor and a second deceleration mechanism; an output end of the second motor is connected with an input of the second deceleration mechanism, and a torque is transmitted to the rack-pinion steering gear after deceleration and torque is increased by the second deceleration mechanism; the master controller calculates an objective current according to a steering torque demand from the steering domain controller, then distributes the current, and further transmits the objective current to the slave control electric power module; the slave control electric power module consists of a slave controller, a third motor and a third deceleration mechanism; an output of the third motor is connected with an input of the third deceleration mechanism, and a torque is transmitted to the rack-pinion steering gear after deceleration and torque is increased by the third deceleration mechanism; and the master controller dynamically compares the first steering angle sensor with the second steering angle sensor according to the steering domain controller and the vehicle-mounted CAN network, so as to realize the tracking of a driving intention.

2. The electric truck steer-by-wire system according to claim 1, wherein the first motor, the second motor and the third motor all are permanent magnet synchronous motors.

3. The electric truck steer-by-wire system according to claim 1, wherein the first deceleration mechanism, the second deceleration mechanism and the third deceleration mechanism all are worm-gear deceleration mechanisms.

4. The electric truck steer-by-wire system according to claim 1, wherein the slave controller consists of a current acquisition circuit, so as to realize the acquisition of a present current signal of the third motor.

5. A network uncertainty control method for an electric truck steer-by-wire system, comprising the following steps: step 1): establishing an electric truck trajectory tracking linear error dynamic model with random network time lag; step 2): establishing an evaluation system of the trajectory tracking performance and dynamic model stability with random network time lag, and constructing an adaptive prediction model; and step 3): constructing a time-varying matrix polytope random network time lag uncertainty term representation method and proposing an adaptive model predictive control algorithm for a polytope time lag uncertain system.

6. The network uncertainty control method for the electric truck steer-by-wire system according to claim 5, wherein step 1) specifically comprises: 11) establishing an electric truck trajectory tracking linear error dynamic model containing random network time lag, as follows: $\stackrel{.}{\stackrel{~}{X}}=A_t\tilde{X}+B_t\tilde{u}$ $\tilde{X}={\left[\begin{array}{cccccc}\stackrel{.}{y}-{\stackrel{.}{y}}_r& \stackrel{.}{x}-{\stackrel{.}{x}}_r& \phi -{\phi }_r& \stackrel{.}{\phi }-{\stackrel{.}{\phi }}_r& y-y_r& x-x_r\end{array}\right]}^T$ $\tilde{u}=u_f-u_{\mathrm{fr}}$ wherein u.sub.f is a control quantity of a front wheel steering angle with random network time lag, x and y are respectively a longitudinal displacement and a transverse displacement of a vehicle, {dot over (x)} and {dot over (y)} are respectively a longitudinal speed and a transverse speed of the vehicle, φ and {dot over (φ)} are respectively a yaw angle and a yaw rate of the vehicle; X.sub.r=[{dot over (y)}.sub.r {dot over (x)}.sub.r φ.sub.r {dot over (φ)}.sub.r y.sub.r x.sub.r].sup.T is a state quantity at a reference point, and u.sub.fr is a reference control quantity; $A_t=\left[\begin{array}{cccccc}\frac{-2\left(C_{\mathrm{cf}}+C_{cr}\right)}{m\stackrel{.}{x}}& \frac{\partial f_{\stackrel{.}{y}}}{\partial \stackrel{.}{x}}& 0& -x+\frac{2\left(l_r{C}_{cr}-l_f{C}_{cf}\right)}{m\stackrel{.}{x}}& 0& 0\\ \stackrel{.}{\phi }-\frac{2C_{\mathrm{cf}}{\delta }_f}{m\stackrel{.}{x}}& \frac{\partial f_{\stackrel{.}{x}}}{\partial \stackrel{.}{x}}& 0& y-\frac{2l_f{C}_{\mathrm{cf}}{\delta }_f}{m\stackrel{.}{x}}& 0& 0\\ 0& 0& 0& 1& 0& 0\\ \frac{2\left(l_r{C}_{cr}-l_f{C}_{\mathrm{cf}}\right)}{I_z\stackrel{.}{x}}& \frac{\partial f_{\stackrel{.}{\phi }}}{\partial \stackrel{.}{x}}& 0& \frac{-2\left(l_f^2{C}_{\mathrm{cf}}+l_r^2{C}_{cr}\right)}{I_z\stackrel{.}{x}}& 0& 0\\ \cos \left(\phi \right)& \sin \left(\phi \right)& \stackrel{.}{x}\cos \left(\phi \right)-\stackrel{.}{y}\sin \left(\phi \right)& 0& 0& 0\\ -\sin \left(\phi \right)& \cos \left(\phi \right)& -\stackrel{.}{y}\cos \left(\phi \right)-\stackrel{.}{x}\sin \left(\phi \right)& 0& 0& 0\end{array}\right]$ $B_t={\left[\begin{array}{cccccc}\frac{2C_{\mathrm{cf}}}{m}& \frac{2C_{\mathrm{cf}}\left(2{\delta }_f-\frac{\stackrel{.}{y}+l_f\stackrel{.}{\phi }}{\stackrel{.}{x}}\right)}{m}& 0& \frac{2l_f{C}_{\mathrm{cf}}}{I_z}& 0& 0\end{array}\right]}^T$ $\mathrm{wherein}$ $\frac{\partial f_{\stackrel{.}{y}}}{\partial \stackrel{.}{x}}=\frac{2C_{\mathrm{cf}}\left({\stackrel{.}{y}}_r+l_f{\stackrel{.}{\phi }}_r\right)+2C_{cr}\left(\stackrel{.}{y}-l_r\stackrel{.}{\phi }\right)}{m{\stackrel{.}{x}}^2}-\stackrel{.}{\phi }$ $\frac{\partial f_x}{\partial \stackrel{.}{x}}=\frac{2C_{\mathrm{cf}}{\delta }_f\left(\stackrel{.}{y}+l_f\stackrel{.}{\phi }\right)}{m{\stackrel{.}{x}}^2}$ $\frac{\partial f_{\stackrel{.}{\phi }}}{\partial \stackrel{.}{x}}=\frac{2l_f{C}_{\mathrm{cf}}\left(\stackrel{.}{y}+l_f\stackrel{.}{\phi }\right)-2l_r{C}_{cr}\left(\stackrel{.}{y}-l_r\stackrel{.}{\phi }\right)}{I_z{\stackrel{.}{x}}^2}$ 12) performing discretization of the continuous model: performing discretization on the electric truck trajectory tracking linear error dynamic model with a ground fixed coordinate system OXY, as follows: $\tilde{X}\left(k+1\right)=A_d\tilde{X}\left(k\right)+B_d\tilde{u}\left(k\right)$ $\mathrm{wherein}$ $\{\begin{array}{c}{A}_d=e^{A_t{T}_s}={L^{-1}\left[\mathrm{sI}-A_t\right]}^{-1}\\ B_d={\int }_0^{T_s}{e}^{A_t\left(T_s-\theta \right)}{B}_{t}d\theta \end{array}$ where T.sub.s is a sampling time, L.sup.−1 is inverse Laplace transform, and s is a Laplace operator; and

13. performing augmentation on a system model: establishing a new state quantity based on the state quantity and the control quantity of the electric truck trajectory tracking linear error dynamic model with random network time lag established in step 11), as follows: $\xi \left(k\right)=\left[\begin{array}{c}\tilde{X}\left(k\right)\\ \tilde{u}\left(k-1\right)\end{array}\right]$ to obtain a state space expression of a discrete ideal trajectory tracking system, as follows: $\xi \left(k+1\right)={\stackrel{\_}{A}}_d\xi \left(k\right)+{\stackrel{\_}{B}}_d\Delta u\left(k\right)$ $\mathrm{wherein}$ ${\stackrel{\_}{A}}_d=\left[\begin{array}{cc}{A}_d& B_d\\ 0& I\end{array}\right],B_d=\left[\begin{array}{c}{B}_d\\ I\end{array}\right].$

7. The network uncertainty control method for the electric truck steer-by-wire system according to claim 5, wherein step 2) specifically consists of: 21) establishing a relationship between a prediction time domain and a control time domain, wherein the relationship between the control time domain and the prediction time domain is as follows: $N_c=\mathrm{round}\left(\frac{N_p}{n}\right)$ where N.sub.p is a prediction time domain, N.sub.c is a control time domain, round(.Math.) is a rounding function, n is an undetermined coefficient, and n={1,2 . . . N.sub.p}; 22) establishing an adaptive prediction time domain function: proposing an adaptive prediction time domain function, as follows:
N.sub.p=ƒ.sub.N.sub.p(V.sub.xk,C.sub.k,τ.sub.k)=k.sub.1V.sub.xk+b.sub.1+k.sub.2C.sub.k+b.sub.2+k.sub.3τ.sub.k+b.sub.3 where V.sub.xk, C.sub.k and τ.sub.k are respectively a longitudinal speed of the vehicle, a trajectory curvature and a network delay obtained by a time stamp at time k, k.sub.i and b.sub.i are undetermined coefficients, and i=1, 2, 3, obtained by offline optimization; 23) establishing a parameter offline optimization model, wherein the parameter offline optimization model is as follows: $\hspace{1em}\{\begin{array}{cc}\min & f_{f_{\mathrm{itness}}}=\frac{k_1}{s_1}{f}_{Time}\left(N_P,N_c\right)+\frac{k_2}{s_2}{f}_{Error}\left(N_P,N_c\right)\\ s.t.& \max \left\{f_{Time}\left(N_P,N_c\right)\right\}\le T_s\end{array}$ where k.sub.1 and k.sub.2 are weight factors, and s.sub.1 and s.sub.2 are scale factors used to convert two objectives into the same order of magnitude; an objective function ƒ.sub.Time(.Math.) is a CPU calculation time required for a rolling optimization process, and an objective function ƒ.sub.Error(.Math.) is an average trajectory tracking error in a whole process; 24) updating the prediction model, wherein after an undermined parameter is obtained by solving offline optimization, a system prediction equation expression is obtained as follows:
Y(k)=ψξ(k)+ΘΔu(k) when the prediction time domain N.sub.p and the control time domain N.sub.c change adaptively according to working conditions, coefficient matrices of the state quantity and the control quantity in the prediction model also change accordingly, where Y(k) is a prediction output at time k, ξ(k) is a state vector at time k, Δu(k) is a control increment vector at time k, ψ and Θ are respectively coefficient matrices of the state quantity and the control quantity, and the expressions are as follows: $\psi =\left[\begin{array}{c}{\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^2\\ \mathrm{.Math.}\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p}\end{array}\right],\Theta =\left[\begin{array}{cccc}{\stackrel{\_}{C}}_d{\stackrel{\_}{B}}_d& 0& \mathrm{.Math.}& 0\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d{\stackrel{\_}{B}}_d& {\stackrel{\_}{C}}_d{\stackrel{\_}{B}}_d& & \mathrm{.Math.}\\ & & \ddots & 0\\ & & & {\stackrel{\_}{C}}_d{\stackrel{\_}{B}}_d\\ \mathrm{.Math.}& \mathrm{.Math.}& & \mathrm{.Math.}\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p-1}{\stackrel{\_}{B}}_d& {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p-2}{\stackrel{\_}{B}}_d& \mathrm{.Math.}& {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p-N_c-1}{\stackrel{\_}{B}}_d\end{array}\right];$ and 25) determining a next sampling time: if the next time still exists in an instruction generated by a counting module of a steering domain controller, returning to step 23), otherwise ending the whole solution process.

8. The network uncertainty control method for the electric truck steer-by-wire system according to claim 5, wherein step 3) specifically comprises: 31) receiving an upper layer trajectory signal input and setting a corresponding filter of a steering domain controller to a high level; 32) establishing a time lag augmented system model: establishing a generalized state vector of the system with random network time lag as follows:
ξ.sub.τ.sub.k(k)=[{tilde over (X)}(k)u(k−1) . . . u(k−γ−1)].sup.T establishing a discrete system state equation considering random network time lag as follows: ${\xi }_{{\tau }_k}\left(k+1\right)={\stackrel{\_}{A}}_{d{\tau }_k}{\xi }_{{\tau }_k}\left(k\right)+{\stackrel{\_}{B}}_{d{\tau }_k}\Delta u\left(k\right)+{\Delta }_k\Delta U\left(k\right)$ $\mathrm{wherein}$ ${\overline{A}}_{d{\tau }_k}=\left[\begin{array}{ccccc}{A}_d& {\Delta }_{1,k}& \mathrm{.Math.}& {\Delta }_{\gamma -1,k}& {\Delta }_{\gamma ,k}\\ 0& I& \mathrm{.Math.}& 0& 0\\ 0& 0& I& & 0\\ \mathrm{.Math.}& \mathrm{.Math.}& & \ddots & \mathrm{.Math.}\\ 0& 0& \mathrm{.Math.}& 0& I\end{array}\right],{\stackrel{\_}{B}}_{d{\tau }_k}=\left[\begin{array}{c}{B}_{d{\tau }_k}\\ I\\ 0\\ \mathrm{.Math.}\\ 0\end{array}\right],{\stackrel{\_}{B}}_{d{\tau }_k}=\left[\begin{array}{c}{B}_{d{\tau }_k}\\ I\\ 0\\ \mathrm{.Math.}\\ 0\end{array}\right]$ ${\Delta }_k=[\begin{array}{cccc}{\Delta }_{1,k}& {\Delta }_{2,k}& \mathrm{.Math.}& {\Delta }_{\gamma ,k}\\ I& & & \\ & I& & \\ & & \ddots & \\ & & & I\end{array}],\Delta U\left(k\right)=\left[\begin{array}{c}\Delta u\left(k-1\right)\\ \Delta u(k-2)\\ \mathrm{.Math.}\\ \Delta u(k-\gamma )\end{array}\right]$ 33) establishing a time lag uncertainty coefficient linear representation method and a random network time lag polytope model, wherein Δ.sub.i,k is an uncertainty coefficient caused by random network delay as follows: ${\Delta }_{i,k}={\int }_0^{\Delta {\Gamma }_{k-i}}{e}^{A_t\left(T_s-\theta \right)}{B}_{t}d\theta$ $i=\left\{1,2\mathrm{.Math.}\gamma \right\}$ $\{\begin{array}{c}{\mathrm{\Delta \Gamma }}_{k-i}={\Gamma }_{k-i}-{\Gamma }_{k-i-1}\\ {\Gamma }_{k-i}=\min \left\{\max \left\{0,{\tau }_{k-i}-i{T}_s\right\},T_s\right\}\end{array}$ where ΔΓ.sub.i is an integral upper limit which is an actual action time length of a control signal Δu(k−i) in a corresponding interval, and τ.sub.k-1 is a total delay length of signals transmitted at time (k−i)T.sub.s; and establishing a random network time lag uncertainty coefficient matrix, which can be expressed by a polytope model as follows: $\left[\begin{array}{c}{\Delta }_{0,k}\\ {\Delta }_{1,k}\\ \mathrm{.Math.}\\ {\Delta }_{\gamma ,k}\end{array}\right]=\left[\begin{array}{c}{\eta }_{0,0}\left(k\right)\\ {\eta }_{1,0}\left(k\right)\\ \mathrm{.Math.}\\ {\eta }_{\gamma ,0}\left(k\right)\end{array}\right]{\hat{\Delta }}_0+\left[\begin{array}{c}{\eta }_{0,1}\left(k\right)\\ {\eta }_{1,1}\left(k\right)\\ \mathrm{.Math.}\\ {\eta }_{\gamma ,2}\left(k\right)\end{array}\right]{\hat{\Delta }}_1+\mathrm{.Math.}+\left[\begin{array}{c}{\eta }_{0,h}\left(k\right)\\ {\eta }_{1,h}\left(k\right)\\ \mathrm{.Math.}\\ {\eta }_{\gamma ,h}\left(k\right)\end{array}\right]{\hat{\Delta }}_h\left[\begin{array}{c}{\hat{\Delta }}_1\\ {\hat{\Delta }}_2\\ \mathrm{.Math.}\\ {\hat{\Delta }}_h\end{array}\right]=e^{A_t{T}_s}\left[\begin{array}{cccc}\frac{{\left(-1\right)}^2}{1!}{A}_t^0& \frac{{\left(-1\right)}^3}{2!}{A}_t^1& \mathrm{.Math.}& \frac{{\left(-1\right)}^{h+1}}{h!}{A}_t^{h-1}\end{array}\right]\left[\begin{array}{c}{x}_{\tau k,1}^T\\ x_{\tau k,2}^T\\ \mathrm{.Math.}\\ X_{\tau k,h}^T\end{array}\right]B_t\left[\begin{array}{c}{x}_{\tau k,1}\\ x_{\tau k,2}\\ \mathrm{.Math.}\\ x_{\tau k,h}\end{array}\right]=\left[\begin{array}{cccc}{\stackrel{\_}{x}}_{\tau k}& 0& \mathrm{.Math.}& 0\\ {\stackrel{\_}{x}}_{\tau k}& {\stackrel{\_}{x}}_{\tau k}^2& & \mathrm{.Math.}\\ \mathrm{.Math.}& & \ddots & 0\\ {\stackrel{\_}{x}}_{\tau k}& {\stackrel{\_}{x}}_{\tau k}^2& \mathrm{.Math.}& {\stackrel{\_}{x}}_{\tau k}^h\end{array}\right]$ ${\stackrel{\_}{x}}_{\tau k}=v{T}_s$ where η.sub.i,h(k) is a time-varying coefficient of polyhedron vertices related to network time lag k, which satisfies a relationship as follows: $\underset{n=0}{\overset{h}{.Math.}}{\eta }_{i,n}\left(k\right)=1i=\left\{1,2\mathrm{.Math.}\gamma \right\}$ v∈[0,1) represents a time length where predictive control should account for the effective control signal transmitted at time (k−i)T.sub.s with the control quantity at time k; 34) establishing an adaptive model predictive control algorithm for an uncertainty system: establishing a trajectory tracking dynamic optimization index based on the dynamic model as follows: $J\left(\xi \left(k\right),U\left(k-1\right),\Delta u\left(k\right)\right)=\underset{i=1}{\overset{N_p}{.Math.}}{.Math.{\xi }_{{\tau }_k}\left(k+i\right)-{\xi }_{{\tau }_k}\left(k+i\right).Math.}^{2}Q+\underset{i=1}{\overset{N_c}{.Math.}}{.Math.\Delta u\left(k+i\right).Math.}^{2}R+{\mathrm{\rho .Math.}}^2$ where N.sub.p is a prediction time domain, N.sub.c is a control time domain, Q and R are weight matrices, ρ is a relaxation factor weight coefficient, and s is a relaxation factor; and establishing a dynamic optimization model of a driverless vehicle trajectory tracking controller with network time lag in each control cycle as follows: ${\min \left[\Delta {u\left(t\right)}^T,\mathrm{.Math.}\right]}^T{H}_t\left[\Delta {u\left(t\right)}^T,\mathrm{.Math.}\right]+G_t\left[\Delta {u\left(t\right)}^T,\mathrm{.Math.}\right]$ $s.t.\left[\begin{array}{c}\Delta U_{\min }\\ y_{\mathrm{hc}\min }\\ y_{\mathrm{sc}\min }-\mathrm{.Math.}\end{array}\right]\le \left[\begin{array}{c}\Delta U\\ y_{hc}\\ y_{sc}\end{array}\right]\le \left[\begin{array}{c}\Delta U_{\max }\\ y_{\mathrm{hc}\max }\\ +y_{\mathrm{hc}\max }+\mathrm{.Math.}\end{array}\right]H_t=\left[\begin{array}{cc}{\Theta }^{T}Q\Theta +R& 0\\ 0& \rho \end{array}\right],G_t=\left[\begin{array}{cc}2e^{T}Q\Theta & 0\end{array}\right]$ where ΔU is a control incremental constraint, y.sub.hc is a hard constraint output, y.sub.sc is a soft constraint output, the range of the soft constraint output can be adjusted by the relaxation factor ε, and e is a tracking error in the prediction time domain; and 35) determining a next sampling time: if a next time still exists in an instruction generated by a counting module of the steering domain controller, returning to step 33), otherwise ending a whole solution process.

Description

BRIEF DESCRIPTION OF FIGURES

[0066] FIG. 1 is a schematic diagram of an electric truck steer-by-wire system according to the disclosure.

[0067] FIG. 2 is a logic diagram of master and slave control strategies according to the disclosure.

[0068] FIG. 3 is a flowchart of an adaptive prediction model according to the disclosure.

[0069] FIG. 4 is a flowchart of an adaptive model predictive control algorithm for a time lag uncertainty system according to the disclosure.

[0070] In order to facilitate the understanding for someone skilled in the art, the disclosure is further described in detail with an embodiment and accompanying drawings below. The content mentioned in the implementation is not the limitation to the disclosure.

[0071] Referring to FIG. 1 and FIG. 2, the disclosure provides an electric truck steer-by-wire system, which includes a master control electric power module 14, a slave control electric power module 18, a road sensing motor module 4, a steering wheel 1, an upper steering column 3, a lower steering column 10, a rack-pinion steering gear 12, wheels 13, a first steering angle sensor 2, a second steering angle sensor 11, a steering domain controller 8 and a vehicle-mounted CAN network 9.

[0072] The steering wheel 1 is connected with a torque input of the upper steering column 3 by a spline. The first steering angle sensor 2 is mounted between the steering wheel 1 and the upper steering column 3. The road sensing motor module 4 is mounted at a lower end of the upper steering column 3. No mechanical connection exists between the upper steering column 3 and the lower steering column 10. A gap exists between the upper steering column and the lower steering column in a vertical direction. The second steering angle sensor 11 is mounted at an input of the lower steering column 10. An output of the lower steering column 10 is connected with the rack-pinion steering gear 12. The master control electric power module 14 and the slave control electric power module 18 are mounted on a rack of the rack-pinion steering gear 12. An end of the rack is connected with the wheels 13, so as to realize the electric truck steer-by-wire operation with a large front axle load.

[0073] The road sensing motor module 4 includes a road sensing controller 5, a first motor 6 and a first deceleration mechanism 7. An output of the first motor 6 is connected with an input of the first deceleration mechanism 7. A torque is transmitted to the upper steering column 3 after deceleration and torque is increased by the first deceleration mechanism 7. The road sensing controller 5 controls the rotating speed of the first motor 6. The road sensing controller 5 simulates a pavement steering torque according to an objective steering torque signal transmitted by the steering domain controller 8. The first steering angle sensor 2 transmits a driver steering angle demand to the steering domain controller 8 by the vehicle-mounted CAN network 9. The steering domain controller 8 converts the driver steering angle demand into the objective steering torque signal and transmits the objective steering torque signal to the master control electric power module 14 by the vehicle-mounted CAN network 9. The road sensing controller 5 supplies power to the first motor 6 and the first steering angle sensor 2.

[0074] The master control electric power module 14 includes a master controller 15, a second motor 16 and a second deceleration mechanism 17. An output of the second motor 16 is connected with an input of the second deceleration mechanism 17. A torque is transmitted to the rack-pinion steering gear 12 after deceleration and torque is increased by the second deceleration mechanism 17. The master controller 15 calculates an objective current according to the steering torque demand from the steering domain controller 8, and then distributes the current, and further transmits the objective current to the slave control electric power module 18. The master controller 15 supplies power to the second motor 16 and the second steering angle sensor 11.

[0075] The slave control electric power module 18 includes a slave controller 19, a third motor 20 and a third deceleration mechanism 21. An output end of the third motor 20 is connected with an input of the third deceleration mechanism 21. A torque is transmitted to the rack-pinion steering gear 12 after deceleration and torque is increased by the third deceleration mechanism 21. The slave controller includes a current acquisition circuit, so as to realize the acquisition of a present current signal from the third motor 20. The slave controller 19 supplies power to the third motor 20.

[0076] The master controller 15 dynamically compares the first steering angle sensor 2 with the second steering angle sensor 11 according to the steering domain controller 8 and the vehicle-mounted CAN network 9, so as to realize the tracking of a driving intention.

[0077] The first motor, the second motor and the third motor all are permanent magnet synchronous motors.

[0078] The first deceleration mechanism, the second deceleration mechanism and the third deceleration mechanism all are worm-gear deceleration mechanisms.

[0079] The disclosure provides a network uncertain control method for an electric truck steer-by-wire system, which, based on the above system, includes the following steps:

[0080] step 1): establishing an electric truck trajectory tracking linear error dynamic model with random network time lag;

[0081] step 2): establishing a trajectory tracking performance evaluation system and dynamic model stability evaluation system with random network time lag, and constructing an adaptive prediction model; and

[0082] step 3): constructing a time-varying matrix polytope random network time lag uncertainty term representation method and proposing an adaptive model predictive control algorithm for a polytope time lag uncertain system.

[0083] Step 1) specifically includes:

[0084] 11) establishing an electric truck trajectory tracking linear error dynamic model with random network time lag, as follows (using a ground fixed coordinate system OXY):

[00015]$\stackrel{.}{\stackrel{~}{X}}=A_t\tilde{X}+B_t\tilde{u}$$\tilde{X}={\left[\begin{array}{cccccc}\stackrel{.}{y}-{\stackrel{.}{y}}_r& \stackrel{.}{x}-{\stackrel{.}{x}}_r& \phi -{\phi }_r& \stackrel{.}{\phi }-{\stackrel{.}{\phi }}_r& y-y_r& x-x_r\end{array}\right]}^T$$\tilde{u}=u_f-u_{\mathrm{fr}}$

[0085] where u.sub.f is a control quantity of a front wheel steering angle with random network time lag, x and y are respectively a longitudinal displacement and a transverse displacement of a vehicle, {dot over (x)} and {dot over (y)} are respectively a longitudinal speed and a transverse speed of the vehicle, φ and {dot over (φ)} are respectively a yaw angle and a yaw rate of the vehicle; X.sub.r=[{dot over (y)}.sub.r {dot over (x)}.sub.r φ.sub.r {dot over (φ)}.sub.r y.sub.r x.sub.r].sup.T is a state quantity at a reference point, and u.sub.fr is a reference control quantity;

[00016]$A_t=\left[\begin{array}{cccccc}\frac{-2\left(C_{\mathrm{cf}}+C_{\mathrm{cr}}\right)}{m\stackrel{.}{x}}& \frac{\partial f_{\stackrel{.}{y}}}{\partial \stackrel{.}{x}}& 0& -x+\frac{2\left(l_r{C}_{\mathrm{cr}}-l_f{C}_{\mathrm{cf}}\right)}{m\stackrel{.}{x}}& 0& 0\\ \stackrel{.}{\phi }-\frac{2C_{\mathrm{cf}}{\delta }_f}{m\stackrel{.}{x}}& \frac{\partial f_{\stackrel{.}{x}}}{\partial \stackrel{.}{x}}& 0& y-\frac{2l_f{C}_{\mathrm{cf}}{\delta }_f}{m\stackrel{.}{x}}& 0& 0\\ 0& 0& 0& 1& 0& 0\\ \frac{2\left(l_r{C}_{\mathrm{cr}}-l_f{C}_{\mathrm{cf}}\right)}{I_z\stackrel{.}{x}}& \frac{\partial f_{\stackrel{.}{\phi }}}{\partial \stackrel{.}{x}}& 0& \frac{-2\left(l_{f^2}{C}_{\mathrm{cf}}+l_r^2{C}_{\mathrm{cr}}\right)}{I_z\stackrel{.}{x}}& 0& 0\\ \cos \left(\phi \right)& \sin \left(\phi \right)& \stackrel{.}{x}\cos \left(\phi \right)-\stackrel{.}{y}\sin \left(\phi \right)& 0& 0& 0\\ -\sin \left(\phi \right)& \cos \left(\phi \right)& -\stackrel{.}{y}\cos \left(\phi \right)-\stackrel{.}{x}\sin \left(\phi \right)& 0& 0& 0\end{array}\right]$$B_t={\left[\begin{array}{cc}\frac{2C_{\mathrm{cf}}}{m}& \frac{2C_{\mathrm{cf}}\left(2{\delta }_f-\frac{\dot{y}+l_f\dot{\phi }}{\dot{x}}\right)}{m}\end{array}\begin{array}{cccc}0& \frac{2l_f{C}_{\mathrm{cf}}}{I_z}& 0& 0\end{array}\right]}^T$$\mathrm{where}$$\frac{\partial f_{\stackrel{.}{y}}}{\partial \stackrel{.}{x}}=\frac{2C_{\mathrm{cf}}\left({\dot{y}}_r+l_f{\dot{\phi }}_r\right)+2C_{\mathrm{cr}}\left(\dot{y}-l_r\dot{\phi }\right)}{m{\dot{x}}^2}-\dot{\phi }$$\frac{\partial f_{\stackrel{.}{x}}}{\partial \stackrel{.}{x}}=\frac{2C_{\mathrm{cf}}{\delta }_f\left(\dot{y}+l_f\dot{\phi }\right)}{m{\dot{x}}^2}$$\frac{\partial f_{\stackrel{.}{\phi }}}{\partial \stackrel{.}{x}}=\frac{2l_f{C}_{\mathrm{cf}}\left(\dot{y}+l_f\dot{\phi }\right)-2l_r{C}_{\mathrm{cr}}\left(\dot{y}-l_r\dot{\phi }\right)}{I_z{\dot{x}}^2}$

[0086] 12) performing discretization of the continuous model:

[0087] performing discretization on the electric truck trajectory tracking linear error dynamic model with a ground fixed coordinate system OXY, as follows:

[00017]$\tilde{X}\left(k+1\right)=A_d\tilde{X}\left(k\right)+B_d\tilde{u}\left(k\right)$$\mathrm{wherein}$$\{\begin{array}{c}{A}_d=e^{A_t{T}_s}={L^{-1}\left[\mathrm{sI}-A_t\right]}^{-1}\\ B_d={\int }_0^{T_s}{e}^{A_1\left(T_s-\theta \right)}{B}_{t}d\theta \end{array}$

[0088] where T.sub.s is a sampling time, L.sup.−1 is inverse Laplace transform, and s is a Laplace operator; and

[0089] 13) performing augmentation on a system model:

[0090] establishing a new state quantity based on the state quantity and the control quantity of the electric truck trajectory tracking linear error dynamic model with random network time lag, established in step 11), as follows:

[00018]$\xi \left(k\right)=\left[\begin{array}{c}\tilde{X}\left(k\right)\\ \tilde{u}\left(k-1\right)\end{array}\right]$

[0091] to obtain a state space expression of a discrete ideal trajectory tracking system, as follows:

[00019]$\xi \left(k+1\right)={\overline{A}}_d\xi \left(k\right)+{\overline{B}}_d\Delta u\left(k\right)$$\mathrm{wherein}$${\overline{A}}_d=\left[\begin{array}{cc}{A}_d& B_d\\ 0& I\end{array}\right],{\overline{B}}_d=\left[\begin{array}{c}{B}_d\\ I\end{array}\right].$

[0092] Referring to FIG. 3, step 2) specifically includes:

[0093] 21) establishing a relationship between a prediction time domain and a control time domain, wherein

[0094] at each sampling time, different values of the prediction time domain and the control time domain change the dimensions of a prediction equation coefficient matrix, thus affecting the solution direction of rolling optimization; the relationship between the value of the control time domain and the prediction time domain is as follows:

[00020]$N_c=\mathrm{round}\left(\frac{N_p}{n}\right)$

[0095] where N.sub.p is a prediction time domain, N.sub.c is a control time domain, round(.Math.) is a rounding function, n is an undetermined coefficient, and n={1,2 . . . N.sub.p};

[0096] 22) establishing an adaptive prediction time domain function, wherein

[0097] the values of the prediction time domain and the control time domain are dynamically determined by comprehensively considering environmental parameters such as a road curvature, working condition parameters such as a vehicle speed and network conditions, and an adaptive prediction time domain function is accordingly proposed as follows:

N.sub.p=ƒ.sub.N.sub.p(V.sub.xk, C.sub.k, τ.sub.k)=k.sub.1V.sub.xk+b.sub.1+k.sub.2C.sub.k+b.sub.2+k.sub.3τ.sub.k+b.sub.3

[0098] where V.sub.xk, C.sub.k and τ.sub.k are respectively longitudinal speed of the vehicle, a trajectory curvature and a network delay obtained by a time stamp at time k, k.sub.i and b.sub.i are undetermined coefficients, and i=1, 2, 3, obtained by offline optimization;

[0099] 23) establishing a parameter offline optimization model, wherein

[0100] the parameter offline optimization model is as follows:

[00021]$\{\begin{array}{cc}\min & f_{\mathrm{fitness}}=\frac{k_1}{s_1}{f}_{\mathrm{Time}}\left(N_P,N_C\right)+\frac{k_2}{s_2}{f}_{\mathrm{Error}}\left(N_P,N_C\right)\\ s.t.& \max \left\{f_{Time}\left(N_P,N_c\right)\right\}\le T_s\end{array}$

[0101] the values of the prediction time domain and the control time domain are obtained by adjusting the undetermined coefficients in multiple iterations, so as to obtain a set of undetermined coefficients that minimize the calculation time and tracking error, where k.sub.1 and k.sub.2 are weight factors, and s.sub.1 and s.sub.2 are scale factors used to convert two objectives into the same order of magnitude; an objective function ƒ.sub.Time(.Math.) is a CPU calculation time required for a rolling optimization, and an objective function ƒ.sub.Error(.Math.) is an average trajectory tracking error in a whole process; the constraint is that the CPU calculation time required for a rolling optimization process does not exceed the sampling time;

[0102] 24) updating the prediction model, wherein

[0103] after an undermined parameter is obtained by solving by offline optimization, a system prediction equation expression is obtained as follows:

Y(k)=ψξ(k)+ΘΔu(k)

[0104] when the prediction time domain N.sub.p and the control time domain N.sub.c change according to working conditions, coefficient matrices of the state quantity and the control quantity in the prediction model also change adaptively, where Y(k) is a prediction output at time k, ξ(k) is a state vector at time k, Δu(k) is a control increment vector at time k, ψ and Θ are respectively coefficient matrices of the state quantity and the control quantity, and the expressions are as follows:

[00022]$\psi =\left[\begin{array}{c}{\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^2\\ .Math.\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p}\end{array}\right],\Theta =\left[\begin{array}{cccc}{\stackrel{\_}{C}}_d{\stackrel{\_}{B}}_d& 0& .Math.& 0\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d{\stackrel{\_}{B}}_d& {\stackrel{\_}{C}}_d{\stackrel{\_}{B}}_d& & .Math.\\ & & \ddots & 0\\ .Math.& .Math.& & {\stackrel{\_}{C}}_d{\stackrel{\_}{B}}_d\\ & & & .Math.\\ {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p-1}{\stackrel{\_}{B}}_d& {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p-2}{\stackrel{\_}{B}}_d& .Math.& {\stackrel{\_}{C}}_d{\stackrel{\_}{A}}_d^{N_p-N_c}{\stackrel{\_}{B}}_d\end{array}\right];$

and

[0105] 25) determining a next sampling time:

[0106] if a next time still exists in an instruction generated by a counting module of a steering domain controller, returning to step 22), otherwise ending the whole solution process.

[0107] Referring to FIG. 4, step 3) specifically includes:

[0108] 31) receiving an upper trajectory signal input and setting a corresponding filter of a steering domain controller to a high level;

[0109] 32) establishing a time lag augmented system model:

[0110] establishing a generalized state vector of the system under with random network time lag as follows:

ξ.sub.τ.sub.k(k)=[{tilde over (X)}(k)u(k−1) . . . u(k−γ−1)].sup.T

[0111] establishing a discrete system state equation considering random network time lag as follows:

[00023]${\xi }_{{\tau }_k}\left(k+1\right)={\overline{A}}_{d{\tau }_k}{\xi }_{{\tau }_k}\left(k\right)+{\overline{B}}_{d{\tau }_k}\Delta u\left(k\right)+{\Delta }_k\Delta U\left(k\right)$$\mathrm{wherein}$${\overline{A}}_{d{\tau }_k}=\left[\begin{array}{ccccc}{A}_d& {\Delta }_{1,k}& .Math.& {\Delta }_{\gamma -1,k}& {\Delta }_{\gamma ,k}\\ 0& I& .Math.& 0& 0\\ 0& 0& I& & 0\\ .Math.& .Math.& & \ddots & .Math.\\ 0& 0& .Math.& 0& I\end{array}\right],{\overline{B}}_{d{\tau }_k}=\left[\begin{array}{c}{B}_{d{\tau }_k}\\ I\\ 0\\ .Math.\\ 0\end{array}\right],{\overline{B}}_{d{\tau }_k}=\left[\begin{array}{c}{B}_{d{\tau }_k}\\ I\\ 0\\ .Math.\\ 0\end{array}\right]$${\Delta }_k=\left[\begin{array}{cccc}{\Delta }_{1,k}& {\Delta }_{2,k}& .Math.& {\Delta }_{\gamma ,k}\\ I& & & \\ & I& & \\ & & \ddots & \\ & & & I\end{array}\right],\Delta U\left(k\right)=\left[\begin{array}{c}\Delta u\left(k-1\right)\\ \Delta u\left(k-2\right)\\ .Math.\\ \Delta u\left(k-\gamma \right)\end{array}\right]$

[0112] 33) establishing a time lag uncertainty coefficient linear representation method and a random network time lag polytope model, wherein

[0113] Δ.sub.i,k is an uncertainty coefficient caused by random network delay as follows:

[00024]${\Delta }_{i,k}={\int }_0^{\Delta {\Gamma }_{k-i}}{e}^{A_t\left(T_s-\theta \right)}{B}_{t}d\theta$$i=\left\{1,2.Math.\gamma \right\}$$\{\begin{array}{c}\Delta {\Gamma }_{k-i}={\Gamma }_{k-i}-{\Gamma }_{k-i-1}\\ {\Gamma }_{k-i}=\min \left\{\max \left\{0,{\tau }_{k-i}-{\mathrm{iT}}_s\right\},T_s\right\}\end{array}$

[0114] wherein ΔΓ.sub.i is an integral upper limit, which is an actual action time length of a control signal Δu(k−i) in a corresponding interval, and τ.sub.k-i is a total delay length of signals transmitted at time (k−i)T.sub.s; and

[0115] establishing a random network time lag uncertainty coefficient matrix, which is capable of being expressed by a polytope model as follows:

[00025]$\left[\begin{array}{c}{\Delta }_{0,k}\\ {\Delta }_{1,k}\\ .Math.\\ {\Delta }_{\gamma ,k}\end{array}\right]=\left[\begin{array}{c}{\eta }_{0,0}\left(k\right)\\ {\eta }_{1,0}\left(k\right)\\ .Math.\\ {\eta }_{\gamma ,0}\left(k\right)\end{array}\right]{\hat{\Delta }}_0+\left[\begin{array}{c}{\eta }_{0,1}\left(k\right)\\ {\eta }_{1,1}\left(k\right)\\ .Math.\\ {\eta }_{\gamma ,2}\left(k\right)\end{array}\right]{\hat{\Delta }}_1+.Math.+\left[\begin{array}{c}{\eta }_{0,h}\left(k\right)\\ {\eta }_{1,h}\left(k\right)\\ .Math.\\ {\eta }_{\gamma ,h}\left(k\right)\end{array}\right]{\hat{\Delta }}_h$$\left[\begin{array}{c}{\hat{\Delta }}_1\\ {\hat{\Delta }}_2\\ .Math.\\ {\hat{\Delta }}_h\end{array}\right]=e^{A_t{T}_s}\left[\frac{{\left(-1\right)}^2}{1!}{A}_t^0\frac{{\left(-1\right)}^3}{2!}{A}_t^1.Math.\frac{{\left(-1\right)}^{h+1}}{h!}{A}_t^{h-1}\right]\left[\begin{array}{c}{x}_{\tau k,1}^T\\ x_{\tau k,2}^T\\ .Math.\\ x_{\tau k,h}^T\end{array}\right]B_t$$\left[\begin{array}{c}{x}_{\tau k,1}^T\\ x_{\tau k,2}^T\\ .Math.\\ x_{\tau k,h}^T\end{array}\right]=[\left[\begin{array}{cccc}{\stackrel{\_}{x}}_{\tau k}& 0& .Math.& 0\\ {\stackrel{\_}{x}}_{\tau k}& {\stackrel{\_}{x}}_{\tau k}^2& & .Math.\\ .Math.& & \ddots & 0\\ {\stackrel{\_}{x}}_{\tau k}& {\stackrel{\_}{x}}_{\tau k}^2& .Math.& {\stackrel{\_}{x}}_{\tau k}^h\end{array}\right]{\stackrel{\_}{x}}_{\tau k}={\mathrm{vT}}_s$

[0116] wherein η.sub.i,h(k) is a time-varying coefficient of polyhedron vertices related to network time lag k, which satisfies a relationship as follows:

[00026]$\underset{n=0}{\overset{h}{.Math.}}{\eta }_{i,n}\left(k\right)=1$$i=\left\{1,2.Math.\gamma \right\}$

[0117] all signals act in a sampling cycle; v∈[0,1) represents a time length where predictive control should account for the effective control signal transmitted at time (k−1) T.sub.s with control quantity at time k;

[0118] 34) establishing an adaptive model predictive control algorithm for an uncertain system:

[0119] establishing a trajectory tracking dynamic optimization index based on the dynamic model as follows:

[00027]$J\left(\xi \left(k\right),U\left(k-1\right),\Delta u\left(k\right)\right)=\underset{i=1}{\overset{N_p}{.Math.}}{.Math.{\xi }_{{\tau }_k}\left(k+i\right)-{\xi }_{{\tau }_k}\left(k+i\right).Math.}^{2}Q+\underset{i=1}{\overset{N_c}{.Math.}}{.Math.\Delta u\left(k+i\right).Math.}^{2}R+{\mathrm{\rho \epsilon }}^2$

[0120] where N.sub.p is a prediction time domain, N.sub.c is a control time domain, Q and R are weight matrices, ρ is a relaxation factor weight coefficient, and ε is a relaxation factor; and

[0121] establishing a dynamic optimization model of a driverless vehicle trajectory tracking controller with network time lag in each control cycle as follows:

[00028]${\min \left[\Delta {u\left(t\right)}^T,\mathrm{.Math.}\right]}^T{H}_t\left[\Delta {u\left(t\right)}^T,\mathrm{.Math.}\right]+G_t\left[\Delta {u\left(t\right)}^T,\mathrm{.Math.}\right]$$s.t.\left[\begin{array}{c}\Delta U_{\min }\\ y_{\mathrm{hc}\min }\\ y_{\mathrm{sc}\min }-\mathrm{.Math.}\end{array}\right]\le \left[\begin{array}{c}\Delta U\\ y_{hc}\\ y_{sc}\end{array}\right]\le \left[\begin{array}{c}\Delta U_{\max }\\ y_{\mathrm{hc}\max }\\ +y_{\mathrm{hc}\max }+\mathrm{.Math.}\end{array}\right]H_t=\left[\begin{array}{cc}{\Theta }^{T}Q\Theta +R& 0\\ 0& \rho \end{array}\right],G_t=\left[\begin{array}{cc}2e^{T}Q\Theta & 0\end{array}\right]$

[0122] wherein ΔU is a control incremental constraint, y.sub.hc is a hard constraint output, y.sub.sc is a soft constraint output, the range of the soft constraint can be adjusted by the relaxation factor ε, and e is a tracking error in the prediction time domain; and

[0123] 35) determining a next sampling time:

[0124] if a next time still exists in an instruction generated by a counting module of the steering domain controller, returning to step 33), otherwise ending a whole solution process.

[0125] There exist many specific applications of this disclosure, and the above example is only a preferred implementation of the disclosure. It should be noted that, for those of ordinary skill in the art, without deviating from the principle of the disclosure, several improvements can be made, and these improvements should also be regarded as the protection scope of this disclosure.