Hydraulic excavator bucket track control method based on data driving

Abstract

The invention provides a hydraulic excavator bucket track control method based on data driving, which belongs to the technical field of complex engineering machinery control, and comprises the following steps: S1, measuring information of a bucket system under the condition of no load; s2, performing dynamic linearization processing on the input and output information of the bucket system in the S1; s3, designing a control law according to a cost function input by the bucket hydraulic system; and S4, the displacement of the bucket hydraulic cylinder is controlled through the PID controller. According to the invention, by introducing a data driving theory, the control method gets rid of model information constraints of a controlled system, and only depends on input / output measurement data of a controlled object to design a controller; meanwhile, the displacement of the bucket hydraulic cylinder is controlled through the PID controller, and the controller in a PID form is easy to design, so that high physical significance is achieved; meanwhile, the calculated amount is small, online control can be achieved, and robustness is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of complex engineering machinery control, relates to the technical field of hydraulic excavator bucket control, and specifically relates to a hydraulic excavator bucket trajectory control method based on data drive. Background Art

[0002] As a complex engineering machine, excavators are widely used to perform various tasks in various unknown and harsh construction environments. Hydraulic excavators usually require a driver with rich field work experience to manually operate the joystick to control the hydraulic mechanical arm to complete a series of specified actions to complete the task. However, the driver's safety in complex working environments and the decline in work efficiency due to long-term work are important issues that need to be solved urgently. The autonomous operation of hydraulic excavators is the core of solving the above problems.

[0003] An important prerequisite for the autonomous operation of hydraulic excavators is high control accuracy, which usually means that a reliable control method is needed for the autonomous operation of hydraulic excavators. On the one hand, as a complex engineering machinery, the operation of the internal hydraulic system of the hydraulic excavator is usually unknown, and it is difficult to accurately model the system. Therefore, the existing control methods based on mathematical models are not applicable; on the other hand, the control methods used by most hydraulic excavators for autonomous operation require a large amount of sensor measurement data support, with high computational costs and poor real-time performance. Therefore, it is urgent to propose a new, easy-to-implement and accurate online control method to meet the high control accuracy requirements of the autonomous operation of hydraulic excavators.

[0004] In view of the unknown and complexity of hydraulic systems, controllers built based on data-driven theory can provide better control effects. The core of data-driven theory is that it can design controllers based solely on real-time measured input and output data when the controlled system is unknown. The controller has the characteristics of low cost and strong robustness. PID control method is a mature technology widely used in practice. The PID controller has a simple structure, is easy to design, and has excellent control effect. However, in the face of unknown complex systems, traditional PID control methods often fail to show their own advantages. Model-free adaptive control is a typical data-driven control method. Model-free adaptive control methods based on tight format, partial format and full format dynamic linearization are common and easy to implement. Their control law structure is simple, but the convergence speed is slow. In addition, the adaptive control laws obtained based on these dynamic linearization methods lack physical meaning and have poor interpretability. In view of the advantages and disadvantages of the above methods, the present invention designs a data-driven PID controller based on incremental triangular dynamic linearization. Summary of the invention

[0005] In order to solve the current safety issues of drivers driving excavators in complex working environments and the problem of reduced work efficiency due to long-term operation, and to solve the problem that the adaptive control law obtained by the dynamic linearization method in the prior art lacks physical meaning and has poor interpretability, the present invention provides a data-driven hydraulic excavator bucket trajectory control method.

[0006] The technical solution of a data-driven hydraulic excavator bucket trajectory control method of the present invention is:

[0007] A data-driven hydraulic excavator bucket trajectory control method comprises the following steps:

[0008] S1. Measure the information of the bucket system under no-load conditions;

[0009] S2, dynamically linearize the input and output information of the bucket system in S1;

[0010] S3, designing the control law according to the cost function input by the bucket system;

[0011] S4. Control the displacement of the bucket hydraulic cylinder through the PID controller.

[0012] A further improvement of the technical solution of the present invention is that: the step S1 specifically includes:

[0013] The bucket is considered as an unknown second-order single-input single-output nonlinear discrete-time system:

[0014]

[0015] Where f represents an unknown second-order single-input single-output nonlinear discrete-time system equivalent to the bucket, y represents the system output, i.e., the displacement of the bucket hydraulic cylinder, and n u Represents the system input order, and u represents the system input, that is, the electrical signal.

[0016] A further improvement of the above technical solution of the present invention is that: the step S2 specifically includes the following steps:

[0017] S2.1. Introducing the extended control input ω in the incremental triangular dynamic linearization model

[0018]

[0019] Among them, parameters m and N are two positive integers greater than zero, let m≥2, N>m;

[0020] S2.2. Define the system input vector

[0021] U t =[y t ,Δy t,ω t ] T #(3)

[0022] The system in formula (2) is dynamically linearized to obtain a virtual data model equivalent to the unknown system (1):

[0023]

[0024] Among them, ψ t =[ψ 1t ,ψ 2t ψ 3t ] T represents the pseudo partial derivative vector.

[0025] A further improvement of the above technical solution of the present invention is that: the step S2.1 is specifically,

[0026] The incremental triangular dynamic linearization model is a special case of the partial format dynamic linearization model in form. The incremental triangular dynamic linearization model can express the system (1) as:

[0027] Δy t+1 =h t Δω t #(13)

[0028] Among them, h t is a bounded time-varying parameter, ω t For extended control input.

[0029] A further improvement of the above technical solution of the present invention is that the cost function in step S3 is expressed as J(u t ), and step S3 specifically includes the following steps:

[0030] S3.1,

[0031] Where r∈R is the bucket reference trajectory signal, let equation (5) be t The partial derivative of is zero, and the adaptive control law is obtained;

[0032]

[0033] S3.2. Define the redundant extended control input in formula (6) as

[0034]

[0035] At the same time, define redundant extended control input with adjustable parameters

[0036]

[0037] in, is a custom weight parameter, δ t is a free parameter greater than 0;

[0038] Defining the learning rate

[0039]

[0040] A further improvement of the above technical solution of the present invention is that: the step S4 specifically includes the following steps:

[0041] S4.1, Substituting formula (8) and (9) into formula (6), the adaptive control law can be converted into PID form;

[0042]

[0043] The above formula is the ideal adaptive PID controller expression, where: is t The predicted value, L t =diag(l Pt ,l Dt ,l It ), θ t =[Δy t ,Δ 2 y t ,e t -Δω' t ] T ; Define PID gains;

[0044] K(t)=[K P (t),K D (t),K I (t)] T #(11)

[0045] At this time, formula (10) can be equivalently written as follows

[0046] Δu t =K(t)Ωθ t #(12)

[0047] Where, Ω = diag(-1,-1,1);

[0048] S4.2. Calculate the adaptive update law Θ for the unknown parameters in the PID-type adaptive control law (10) A ;

[0049] S4.3, calculate the control law for PID parameter adaptation;

[0050] S4.4, PID controller controls the displacement of the bucket hydraulic cylinder according to the control law in S4.3.

[0051] A further improvement of the technical solution of the present invention is that: the step S4.2 is specifically,

[0052] Define the adaptive update law

[0053]

[0054] Give targeted The cost function

[0055]

[0056] Similar to formula (5), let formula (14) be The partial derivative of is zero, and the adaptive update law is specifically

[0057]

[0058] Among them, λ>0 is the weight parameter, η t-1 ∈(0,1.5] is a custom parameter.

[0059] A further improvement of the above technical solution of the present invention is that: the step S4.3 is specifically, A (L t-1 ,θ t-1 ) is substituted into the adaptive control law (10), and we get

[0060]

[0061] Right now

[0062]

[0063] in, Furthermore, define and As a time-invariant parameter,

[0064]

[0065] Thus, the control law of PID parameter adaptation is obtained, that is,

[0066]

[0067] Due to the adoption of the above technical solution, the technical advances achieved by the present invention include:

[0068] 1. The present invention provides a data-driven hydraulic excavator bucket trajectory control method. By introducing data-driven theory, the control method of the present invention is freed from the model information constraints of the controlled system and only relies on the input / output measurement data of the controlled object to design the controller.

[0069] 2. The present invention controls the displacement of the bucket hydraulic cylinder through a PID controller. Since a controller in the form of a PID is easy to design, it has a strong physical meaning; at the same time, the amount of calculation is small, and online control can be achieved with strong robustness.

[0070] 3. The dynamic linearization method of the present invention provides good rapidity and stability for the autonomous operation of the hydraulic excavator. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 It is a control flow chart of a data-driven hydraulic excavator bucket trajectory control method of the present invention;

[0072] Figure 2 It is the block diagram of the hydraulic system of the excavator;

[0073] Figure 3 It is the PID gain adaptive update diagram;

[0074] Figure 4 This is a method control effect diagram of a data-driven hydraulic excavator bucket trajectory control method of the present invention. DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solution and advantages of the present invention more clear, the present invention is further described in detail below in combination with specific implementation methods and with reference to the accompanying drawings. In the following description, the description of the public structure and technology is omitted to avoid unnecessary confusion of the concept of the present invention.

[0076] First reference Figure 2 It can be seen that the hydraulic system of the excavator mechanical arm is mainly composed of an electronic control system, a diesel engine, a hydraulic pump, a distribution valve, and a boom, a dipper rod, and a bucket cylinder. Taking the hydraulic excavator bucket as an example, the reference signal (i.e., a given electrical signal) is provided by the joystick. The electronic control system drives the motor according to the reference signal and distributes electricity to the hydraulic pump. Under the premise of the diesel engine providing power, the hydraulic pump converts mechanical energy into hydraulic energy, and the distribution valve distributes the hydraulic energy to the corresponding bucket hydraulic cylinder, driving the bucket hydraulic cylinder to convert the hydraulic energy into mechanical energy to realize the rotation of the bucket.

[0077] like Figure 2 As shown, the control method of the present invention comprises the following steps:

[0078] S1. Measure the information of the bucket system under no-load conditions, and regard the bucket as an unknown second-order single-input single-output nonlinear discrete-time system:

[0079]

[0080] Where f represents an unknown second-order single-input single-output nonlinear discrete-time system equivalent to the bucket, y represents the output of the bucket system, i.e., the displacement of the bucket hydraulic cylinder, and n u Represents the system input order, and u represents the bucket system input, that is, the electrical signal.

[0081] S2, dynamically linearize the input and output information of the bucket system in S1; specifically includes the following steps:

[0082] S2.1. Introducing the extended control input ω in the incremental triangular dynamic linearization model t

[0083]

[0084] Among them, parameters m and N are two positive integers greater than zero, and let m≥2, N>m.

[0085] Specifically, the incremental triangular dynamic linearization model is a formal special case of the partial format dynamic linearization model. The incremental triangular dynamic linearization model can express system (1) as:

[0086] Δy t+1 =h t Δω t #(13)

[0087] Among them, h t is a bounded time-varying parameter, ω t For extended control input.

[0088] S2.2. Define the system input vector

[0089] U t =[y t ,Δy t ,ω t ] T #(3)

[0090] The system in formula (2) is dynamically linearized to obtain a virtual data model equivalent to the unknown system (1):

[0091]

[0092] Among them, ψ t =[ψ 1t ,ψ 2t ψ 3t ] T represents the pseudo partial derivative vector.

[0093] S3. Design the control law based on the cost function input by the bucket system. The cost function is expressed as J(u t ), and step S3 specifically includes the following steps:

[0094] S3.1. The cost function of the bucket hydraulic system input u is:

[0095]

[0096] Where r∈R is the bucket reference trajectory signal, let equation (5) be t The partial derivative of is zero, and the adaptive control law is obtained;

[0097]

[0098] S3.2. Define the redundant extended control input in formula (6) as

[0099]

[0100] At the same time, define redundant extended control input with adjustable parameters

[0101]

[0102] in, is a custom weight parameter, δ t is a free parameter greater than 0;

[0103] Defining the learning rate

[0104]

[0105] S4, controlling the displacement of the bucket hydraulic cylinder through a PID controller. Specifically, it includes the following steps:

[0106] S4.1, Substituting formula (8) and (9) into formula (6), the adaptive control law can be converted into PID form;

[0107]

[0108] The above formula is the ideal adaptive PID controller expression, where: is t The predicted value, L t =diag(l Pt ,l Dt ,l It ), θ t =[Δy t ,Δ 2 y t ,e t -Δω' t ] T ; Define PID gains;

[0109] K(t)=[KP (t),K D (t),K I (t)] T #(11)

[0110] At this time, formula (10) can be equivalently written as follows

[0111] Δu t =K(r)Ωθ t #(12)

[0112] Where, Ω = diag(-1,-1,1);

[0113] S4.2. Calculate the adaptive update law Θ for the unknown parameters in the PID-type adaptive control law (10) A .

[0114] The above step S4.2 is specifically as follows:

[0115] Define the adaptive update law

[0116]

[0117] Give targeted The cost function

[0118]

[0119] Similar to formula (5), let formula (14) be The partial derivative of is zero, and the adaptive update law is specifically

[0120]

[0121] Among them, λ>0 is the weight parameter, η t-1 ∈(0,1.5] is a custom parameter.

[0122] S4.3. Calculate the control law for PID parameter adaptation.

[0123] The above S4.3 is specifically:

[0124] Θ A (L t-1 ,θ t-1 ) is substituted into the adaptive control law (10), and we get

[0125]

[0126] Right now

[0127]

[0128] in, Furthermore, define and As a time-invariant parameter,

[0129]

[0130] Thus, the control law of PID parameter adaptation is obtained, that is,

[0131]

[0132] S4.4, PID controller controls the displacement of the bucket hydraulic cylinder according to the control law in S4.3.

[0133] Simulation verification:

[0134] Consider the bucket as a nonlinear time-invariant system as follows:

[0135]

[0136] Among them, y represents the system output, that is, the displacement of the bucket hydraulic cylinder, u represents the system input, that is, the electrical signal, and ξ t The mean is 0 and the variance is 0.1 2 Gaussian white noise, that is, disturbance in an unknown working environment. Control the hydraulic cylinder of the excavator bucket to track the reference signal Initialize system parameters:

[0137] y(0)=0,u(0)=0.

[0138] Introducing the Extended Control Input in the Incremental Triangular Dynamic Linearization Model

[0139]

[0140] And redundant expansion control input with adjustable parameters

[0141]

[0142] Among them, φ t is a custom weight parameter, δ t is a free parameter greater than 0. t Can be converted to a constant value greater than 0 without changing the parameter In this example, select N=4,m=2.

[0143] In order to ensure the accuracy and stability of the hydraulic cylinder tracking reference signal of the hydraulic excavator bucket, a controller with adaptive PID parameters is established:

[0144]

[0145] Where K(t)=[KP (t),K D (t),K I (t)] T is the adaptive updated PID gain, To customize the time-invariant parameters, is the learning rate of the corresponding PID gain, λ is a weight parameter greater than 0, and θ t =[Δy t ,Δ 2 y t ,e t -Δω' t ] T And Ω=diag(-1,-1,1). Initialize PID controller parameters:

[0146] K(0)=[0,0,0] T , λ=1,

[0147] The PID controller is designed. The hydraulic cylinder control program of the hydraulic excavator bucket is written by MATLAB. The simulation results and analysis are as follows:

[0148] like Figure 3 As shown in the figure, the PID gain K is adaptively updated as the reference signal changes, ensuring the rapidity, accuracy and stability of the control. The algorithm control effect is as follows: Figure 4 As shown, in the case of taking into account Gaussian white noise, the control method proposed in the present invention can still maintain good tracking accuracy.

[0149] In the above embodiment, the present invention provides a data-driven hydraulic excavator bucket trajectory control method. By introducing data-driven theory, the control method of the present invention is freed from the model information constraints of the controlled system, and only relies on the input / output measurement data of the controlled object to design the controller; at the same time, the present invention controls the displacement of the bucket hydraulic cylinder through a PID controller. Since the controller in the form of PID is easy to design, it has a strong physical meaning; at the same time, the amount of calculation is small, and online control can be realized with strong robustness; at the same time, the triangular dynamic linearization method of the present invention provides good rapidity and stability for the autonomous operation of the hydraulic excavator.

[0150] The above-described embodiments are merely descriptions of preferred implementations of the present invention, and are not intended to limit the concept and scope of the present invention. Without departing from the design concept of the present invention, various modifications and improvements made by ordinary persons in the art to the technical solution of the present invention should fall within the protection scope of the present invention, and the technical contents for which protection is sought in the present invention have been fully recorded in the claims.

Claims

1. A data-driven hydraulic excavator bucket trajectory control method, characterized in that: The following steps are included: S1. Measure the information of the bucket system under no-load conditions; S2, dynamically linearize the input and output information of the bucket system in S1; S3, designing the control law according to the cost function input by the bucket system; S4. Control the displacement of the bucket hydraulic cylinder through the PID controller.

2. The data-driven hydraulic excavator bucket trajectory control method according to claim 1 is characterized in that: The step S1 specifically includes: The bucket is considered as an unknown second-order single-input single-output nonlinear discrete-time system: Where f represents an unknown second-order single-input single-output nonlinear discrete-time system equivalent to the bucket, y represents the system output, i.e., the displacement of the bucket hydraulic cylinder, and n u Represents the system input order, and u represents the system input, that is, the electrical signal.

3. A data-driven hydraulic excavator bucket trajectory control method according to claim 2, characterized in that: The step S2 specifically includes the following steps: S2.

1. Introducing the extended control input ω in the incremental triangular dynamic linearization model t Among them, parameters m and N are two positive integers greater than zero, let m≥2, N>m; S2.

2. Define the system input vector U t =[y t ,Δy t ,ω t ] T #(3) The system in formula (2) is dynamically linearized to obtain a virtual data model equivalent to the unknown system (1): Among them, ψ t =[ψ 1t ,ψ 2t ψ 3t ] T represents the pseudo partial derivative vector.

4. The data-driven hydraulic excavator bucket trajectory control method according to claim 3 is characterized in that: The step S2.1 is specifically as follows: The incremental triangular dynamic linearization model is a formal special case of the partial format dynamic linearization model; the incremental triangular dynamic linearization model can express the system (1) as: Δy t+1 =h t Give t #(13) Among them, h t is a bounded time-varying parameter, ω t For extended control input.

5. The data-driven hydraulic excavator bucket trajectory control method according to claim 3 is characterized in that: The cost function in step S3 is J(u t ), and step S3 specifically includes the following steps: S3.

1. The cost function of the bucket hydraulic system input u is: Where r∈R is the bucket reference trajectory signal, let equation (5) be t The partial derivative of is zero, and the adaptive control law is obtained; S3.

2. Define the redundant extended control input in formula (6) as At the same time, define redundant extended control input with adjustable parameters in, is a custom weight parameter, δ t is a free parameter greater than 0; Defining the learning rate 6. The data-driven hydraulic excavator bucket trajectory control method according to claim 5, characterized in that: The step S4 specifically includes the following steps: S4.1, Substituting formula (8) and (9) into formula (6), the adaptive control law can be converted into PID form; The above formula is the ideal adaptive PID controller expression, where: is t The predicted value, L t =diag(l Pt ,l Dt ,l It ), θ t =[Δy t ,Δ 2 y t ,e t -Δω' t ] T ; Define PID gains; K(t)=[K P (t),K D (t),K I (t)] T #(11) At this time, formula (10) can be equivalently written as follows Thu t =K(t)Ωθ t #(12) Where, Ω = diag(-1,-1,1); S4.

2. Calculate the adaptive update law Θ for the unknown parameters in the PID-type adaptive control law (10) A ; S4.3, calculate the control law for PID parameter adaptation; S4.4, PID controller controls the displacement of the bucket hydraulic cylinder according to the control law in S4.

3.

7. A data-driven hydraulic excavator bucket trajectory control method according to claim 6, characterized in that: The step S4.2 is specifically as follows: Defining the adaptive update law Give targeted The cost function Similar to formula (5), let formula (14) be The partial derivative of is zero, and the adaptive update law is specifically Among them, λ>0 is the weight parameter, η t-1 ∈(0,1.5] is a custom parameter.

8. The data-driven hydraulic excavator bucket trajectory control method according to claim 7 is characterized in that: The step S4.3 is specifically as follows: Θ A (L t-1 ,θ t-1 ) is substituted into the adaptive control law (10), and we get Right now in, Furthermore, define and As a time-invariant parameter, Thus, the control law of PID parameter adaptation is obtained, that is,

Images