A hoisting control system and control method of a marine crane

The marine crane control system that automatically adjusts the lifting status through reinforcement learning algorithms solves the problem of cargo shaking in heavy wind and wave weather, and achieves stable lifting in different sea conditions and simplifies the control of multiple drives.

CN115072582BActive Publication Date: 2024-11-15JIANGSU UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210379097.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-12
Publication Date
2024-11-15
Estimated Expiration
2042-04-12

AI Technical Summary

Technical Problem

In the heavy wind and wave weather, existing ship cranes shake cargo, causing economic losses and safety risks, and intelligent lifting equipment requires complex parameters adjustment and high-demand operators to operate multiple drives, making it difficult to achieve stable lifting.

Method used

Adopt a marine crane control system that automatically adjusts the lifting status based on reinforcement learning, including data acquisition module, calculation module and execution module. The lifting drive is selected through the ε-greedy algorithm to perform actions, adjust the lifting status of the cargo, and simplify the control of multiple drives. logic.

Benefits of technology

The self-learning function under different sea conditions is realized, and the servo lifting capability is obtained through iterative learning in a short time, which simplifies the control of multiple drives, and improves the stability and operation simplicity of the crane.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115072582B_ABST
    Figure CN115072582B_ABST
Patent Text Reader

Abstract

The invention discloses a hoisting control system and control method of a ship crane, wherein the ship crane includes several hoisting drivers; the hoisting control system includes a data acquisition module to obtain the actual hoisting height and actual hoisting speed of the hoisted cargo; a calculation module to calculate the expected hoisting height and expected hoisting speed of the hoisted cargo; and a state equation of the hoisted cargo is constructed to obtain the hoisting state of the hoisted cargo; then an ε-greedy algorithm is used to select one of several hoisting drivers to perform an action; and an execution module is used to drive the hoisting driver selected in the calculation module to perform an action and adjust the hoisting state of the hoisted cargo. The hoisting control system has a self-learning function, and under different sea conditions, the hoisting state with the greatest value is selected to adjust the execution action, and the servo hoisting capability is obtained through iterative learning in a short period of time, and multiple driver control logics and methods are simplified.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to ship hoisting, and in particular to a hoisting control system of a ship crane and a control method thereof. Background Art

[0002] Ship cranes are widely used in marine research vessels, bulk carriers, special ships, etc. They are mainly used for the release and recovery of remotely operated unmanned vehicles (ROVs) and the loading and unloading of various types of cargo. Ordinary cranes are manually controlled. In the event of strong winds and waves, the shaking of the ship will cause the cargo to shake, and in severe cases, the cargo will hit the side of the ship or the crane equipment, causing economic losses and safety risks.

[0003] Some existing intelligent lifting equipment mostly uses conventional PID algorithms to offset the sway and heave of cargo. This algorithm requires targeted parameter adjustment for different sea conditions, and the parameter adjustment process is complicated and requires experienced engineers to board the ship to assist in the parameter adjustment, which consumes a lot of manpower and material resources. Ship cranes are very complex mechanical structures with multiple drives, such as multiple cylinders and winding devices. Operators are required to be proficient in operating multiple drives to achieve lifting, which places high demands on the operator's operating ability. Summary of the invention

[0004] Purpose of the invention: In view of the above shortcomings, the present invention provides a hoisting control system for a ship crane which automatically adjusts the hoisting state of the hoisted cargo based on reinforcement learning.

[0005] The invention also provides a control method for a ship crane hoisting control system.

[0006] Technical solution: To solve the above problems, the present invention adopts a hoisting control system for a ship crane, including a data acquisition module, a calculation module, and an execution module. The ship crane includes several hoisting drivers; the data acquisition module is used to obtain the actual hoisting height and the actual hoisting speed of the hoisted cargo; the calculation module is used to calculate the expected hoisting height and the expected hoisting speed of the hoisted cargo; and according to the actual hoisting height, the actual hoisting speed, the expected hoisting height and the expected hoisting speed of the hoisted cargo, a state equation of the hoisted cargo is constructed to obtain the hoisting state of the hoisted cargo; then the ε-greedy algorithm is used to select one of the several hoisting drivers to perform an action; the execution module is used to drive the hoisting driver selected in the calculation module to perform an action and adjust the hoisting state of the hoisted cargo.

[0007] Furthermore, the marine crane includes a front small beam, a pulley, a parallel arm, and a seat body, one end of the parallel arm is hinged to the seat body, the other end of the parallel arm is hinged to one end of the front small beam, and a pulley is arranged on the other end of the front small beam; a plurality of lifting drives include a main hydraulic cylinder, an auxiliary hydraulic cylinder, and a capstan arranged on the seat body, a suspension line is wound around the capstan, and one end of the suspension line is fixedly connected to the capstan, and the other end of the suspension line is guided by the pulley, one end of the main hydraulic cylinder is hinged to the seat body, and the other end is hinged to the parallel arm body, one end of the auxiliary hydraulic cylinder is hinged to the parallel arm, and the other end is hinged to the front small beam body.

[0008] Furthermore, the ship crane also includes a force compensation device, which includes a counterweight, a center rod and a spring. One end of the center rod is connected to the counterweight, and the other end of the center rod is hinged to the parallel arm body. The counterweight and the parallel arm are respectively located on both sides of the seat body, and a limit block is fixedly provided on the seat body, and the center rod passes through the limit block; a spring is sleeved on the center rod, and the spring is located between the limit block and the counterweight.

[0009] Furthermore, a binocular camera is arranged on the front small beam, and the binocular camera is used to obtain a position image of the suspended cargo.

[0010] Furthermore, the parallel arm comprises an upper parallel arm and a lower parallel arm, the extension directions of the upper parallel arm and the lower parallel arm are always parallel, a wire sliding hole is provided between the upper parallel arm and the lower parallel arm, and the suspension wire passes through the wire sliding hole.

[0011] The present invention also adopts a control method of a marine crane hoisting control system, comprising the following steps:

[0012] Step 1: Set the crane working status;

[0013] Step 2: Obtain the actual lifting height and actual lifting speed of the lifted cargo; and calculate the expected lifting height and expected lifting speed of the lifted cargo at this time;

[0014] Step 3: According to the actual lifting height, actual lifting speed, expected lifting height and expected lifting speed of the lifted cargo, a state equation of the lifted cargo is constructed to obtain the lifting state of the lifted cargo;

[0015] Step 4: Use the ε-greedy algorithm to select one of several lifting drives to perform the action;

[0016] Step 5: Adjust the lifting state of the suspended cargo according to the lifting driver selected in step 4, and then return to step 2 until the suspended cargo is stably lifted.

[0017] Furthermore, the state equation of the suspended cargo in step 3 is:

[0018] S=a(dds )+b(vv s )+c

[0019] Among them, a is the hoisting height difference gain, b is the hoisting speed difference gain, c is the additional increment term, d is the actual hoisting height of the hoisted cargo, v is the actual hoisting speed of the hoisted cargo, d s is the expected lifting height of the cargo, v s It is the expected lifting speed of the cargo to be lifted.

[0020] Furthermore, the execution action of the hoisting driver belongs to the action set A, A = {CY p ,CY r ,DY p ,DY r ,RE p ,RE r}, where CY p Indicates the extension of the main hydraulic cylinder, CY r Indicates the shortening of the main hydraulic cylinder, DY p Indicates the extension of the auxiliary hydraulic cylinder, DY r Indicates the shortening of the auxiliary hydraulic cylinder, RE p Indicates the tightening of the hanging wire on the winch, RE r Indicates the loosening of the hanging wire on the capstan.

[0021] Furthermore, when the lifting driver performs an action in step 5, the lifted cargo enters a new lifting state S', and a reward R is obtained in the ε-greedy algorithm, and the value function of the execution action is updated:

[0022] Q(S,A)=Q(S,A)+α(R+γmaxQ(S',A)-Q(S,A))

[0023] Among them, α is the learning rate, γ is the reduction coefficient, and maxQ(S',A) is the maximum value when the new lifting state S' takes the execution action.

[0024] Furthermore, if the status item value of the new lifting state of the lifted cargo is smaller than the status item value of the original lifting state, the reward R is a positive reward, and the smaller the status item value of the new lifting state is, the greater the reward R is; if the status item value of the new lifting state of the lifted cargo is greater than the status item value of the original lifting state, the reward R is a negative reward.

[0025] Beneficial effect: Compared with the prior art, the significant advantage of the present invention is that through reinforcement learning, the hoisting control system has the function of self-learning. Under different sea conditions, it selects the hoisting state with the greatest value to adjust the execution action, and obtains the ability of servo hoisting through iterative learning in a short time. The actions that can be executed by multiple hoisting drivers on the crane are summarized into an action set for the crane to choose during the learning process, which simplifies the control logic and method of multiple drivers. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 Shown is a schematic diagram of the overall structure of the marine crane of the present invention;

[0027] Figure 2 Shown is a side view of the ship crane of the present invention. DETAILED DESCRIPTION

[0028] Example 1

[0029] like Figure 1 and Figure 2 As shown, a lifting control system of a ship crane in this embodiment includes a data acquisition module, a calculation module, and an execution module. The ship crane includes a base 1, a winding device 2, a force compensation device 3, a parallel beam arm device 4, and a front small beam device 5. The lifting drive includes a main hydraulic cylinder 6, an auxiliary hydraulic cylinder 7, and a winch 202 arranged on a base body 101.

[0030] The base 1 includes a seat body 101, a hinge shaft 102, a hinge shaft 103, a limit block 104, a capstan fixing hole 105 and a hinge hole 106. The hinge shaft 102, the hinge shaft 103 and the limit block 104 are located in the same vertical plane, and the seat body 101 is arranged in sequence upward, and the capstan fixing hole 105 and the hinge hole 106 are respectively located on both sides of the hinge shaft 102.

[0031] The wire winding device 2 includes a suspension wire 201, a capstan 202, a wire sliding hole 404 and a pulley 502. The capstan 202 is arranged at the capstan fixing hole 105 on the seat body 101, and the capstan 202 rotates relative to the seat body 101. The suspension wire is wound around the capstan 202, and one end of the suspension wire 201 is fixedly connected to the capstan, and the other end of the suspension wire 201 passes through the wire sliding hole 404 and is guided and connected to the suspended goods through the pulley 502. The suspension wire is made of steel cable, and the capstan 202 is rotated to realize the retraction and release of the suspension wire.

[0032] The parallel beam arm device 4 includes an upper parallel arm 401, a lower parallel arm 402, and a vertical beam 403. The vertical beam 403 is provided with an upper hinge hole 405, a lower hinge hole 406, and a hinge hole 407. One end of the upper parallel arm 401 is hinged to the base body 101 through the hinge shaft 103, and the other end is hinged to the vertical beam 403 through the upper hinge hole 405. One end of the lower parallel arm 402 is hinged to the base body 101 through the hinge shaft 102, and the other end is hinged to the vertical beam 403 through the lower hinge hole 406. The extension directions of the upper parallel arm 401 and the lower parallel arm 402 are always parallel, and a plurality of wire sliding holes 404 are provided between the upper parallel arm 401 and the lower parallel arm 402. One end of the main hydraulic cylinder 6 is hinged to the seat body 101 through the hinge hole 106, and the other end is hinged to the arm body of the lower parallel arm 402. The extension and retraction of the main hydraulic cylinder 6 drives the upper parallel arm 401 and the lower parallel arm 402 to rotate up and down relative to the seat body 101.

[0033] The front beam device 5 includes a front beam 501, one end of which is hinged to the vertical beam 403 through the lower hinge hole 406, and the other end is provided with a pulley 502. One end of the auxiliary hydraulic cylinder 7 is hinged to the vertical beam 403 through the hinge hole 407, and the other end is hinged to the beam body of the front beam 501. The extension and contraction of the auxiliary hydraulic cylinder 7 drives the front beam 501 to rotate up and down relative to the parallel beam arm device 4. A binocular camera 504 is provided on the front beam device 5, and the binocular camera obtains the position image of the hoisted cargo, and calculates the hoisting height and hoisting speed of the hoisted cargo according to the image.

[0034] The force compensation device 3 includes a counterweight 303, a center rod 301 and a spring 302. One end of the center rod 301 is connected to the counterweight 303, and the other end of the center rod 301 is hinged to the parallel arm body. The counterweight 303 and the parallel arm are respectively located on both sides of the seat body 101, and the center rod 301 passes through the limit block 104; the spring 302 is pre-stressed on the center rod 301, and the spring 302 is located between the limit block 104 and the counterweight 303.

[0035] The data acquisition module obtains the actual lifting height and actual lifting speed of the lifted goods through the images obtained by the binocular camera 504;

[0036] The calculation module calculates the expected lifting height and expected lifting speed of the goods being lifted; and constructs the state equation of the goods being lifted according to the actual lifting height, actual lifting speed, expected lifting height and expected lifting speed of the goods being lifted, and obtains the lifting state of the goods being lifted; and then uses the ε-greedy algorithm to select one of several lifting drivers to perform the action;

[0037] The execution module is used to drive the lifting driver selected in the calculation module to perform actions, adjust the lifting state of the lifted goods, and then use the ε-greedy algorithm to select and adjust the new lifting state to achieve stable lifting of the goods.

[0038] Example 2

[0039] In this embodiment, a control method for a marine crane hoisting control system is provided. The marine crane structure adopts the structure in Embodiment 1, which will not be described in detail herein. The control method comprises the following steps:

[0040] Step 1: Set the working state of the ship crane, such as keeping the cargo stable relative to the lifting reference surface (such as the sea surface or the deck) or rising or falling steadily at a certain speed. During lifting, the crane arm extends out of the side of the ship to lift the cargo;

[0041] Step 2: Calculate the distance between the suspended cargo and the camera through the binocular camera arranged on the front small beam, obtain the distance D between the suspended cargo and the front small beam, and the distance L between the front small beam and the lifting reference surface (such as the sea surface or deck), calculate the lifting height d=LD of the suspended cargo, and calculate the vertical movement speed of the cargo v=△d / △t, where △d is the difference in distance between time points t1 and t2, and △t is the length between time points t1 and t2; calculate the expected lifting height and expected lifting speed of the suspended cargo at this time according to the set crane working state;

[0042] Step 3: According to the actual lifting height, actual lifting speed, expected lifting height and expected lifting speed of the lifted cargo, a state equation of the lifted cargo is constructed to obtain the lifting state of the lifted cargo;

[0043] The state equation of the suspended cargo is:

[0044] S=a(dd s )+b(vv s )+c

[0045] Among them, a is the hoisting height difference gain, b is the hoisting speed difference gain, c is the additional increment term, d is the actual hoisting height of the hoisted cargo, v is the actual hoisting speed of the hoisted cargo, d s is the expected lifting height of the cargo, v s It is the expected lifting speed of the cargo to be lifted.

[0046] Step 4: Calculate the initial state S according to the distance and speed of the cargo, and use the ε-greedy algorithm to select an execution action of the crane driver; the execution action of the crane driver belongs to the action set A, A = {CY p ,CY r ,DY p ,DY r ,RE p ,RE r}, where CY p Indicates the extension of the main hydraulic cylinder, CY r Indicates the shortening of the main hydraulic cylinder, DY pIndicates the extension of the auxiliary hydraulic cylinder, DY r Indicates the shortening of the auxiliary hydraulic cylinder, RE p Indicates the tightening of the hanging wire on the winch, RE r Indicates the loosening of the hanging wire on the capstan.

[0047] Step 5: Adjust the lifting state of the lifted cargo according to the lifting driver selected in step 4. When the lifting driver executes the action, the lifted cargo enters a new lifting state S', and the reward R is obtained in the ε-greedy algorithm. The state item value of the new lifting state of the lifted cargo is less than the state item value of the original lifting state. The reward R is a positive reward, and the smaller the state item value of the new lifting state, the greater the reward R; the state item value of the new lifting state of the lifted cargo is greater than the state item value of the original lifting state, and the reward R is a negative reward. Update the lifting state and the value function of the execution action:

[0048] Q(S,A)=Q(S,A)+α(R+γmaxQ(S',A)-Q(S,A))

[0049] Among them, α is the learning rate, γ is the reduction coefficient, maxQ(S',A) is the maximum value when the new lifting state S' takes the execution action, and then returns to step 2. After a certain number of initial training times, the crane control can learn the choice of the maximum value in each state, and by looping each execution of the action with the maximum value, the servo lifting effect of the cargo being stabilized at the expected lifting height or lifted at the expected speed is achieved.

Claims

1. A hoisting control system for a marine crane, characterized in that: It includes a data acquisition module, a calculation module, an execution module, and the marine crane includes several hoisting drivers; The data acquisition module is used to obtain the actual hoisting height and actual hoisting speed of the hoisted goods; The calculation module is used to calculate the expected lifting height and expected lifting speed of the goods to be lifted; and according to the actual lifting height, actual lifting speed, expected lifting height and expected lifting speed of the goods to be lifted, the state equation of the goods to be lifted is constructed to obtain the lifting state of the goods to be lifted; and then use The algorithm selects one of several lifting actuators to perform the action; The execution module is used to drive the hoisting driver selected in the calculation module to perform actions and adjust the hoisting state of the hoisted goods; The marine crane comprises a front small beam (501), a pulley (502), a parallel arm, and a seat (101); one end of the parallel arm is hinged to the seat (101), the other end of the parallel arm is hinged to one end of the front small beam (501), and the pulley (502) is arranged at the other end of the front small beam (501); a plurality of lifting drivers comprise a main hydraulic cylinder (6), an auxiliary hydraulic cylinder (7), and a winch (202) arranged on the seat (101); a suspension line is wound around the winch (202), one end of the suspension line is fixedly connected to the winch, and the other end of the suspension line is guided by the pulley (502); one end of the main hydraulic cylinder (6) is hinged to the seat (101), and the other end is hinged to the parallel arm; one end of the auxiliary hydraulic cylinder (7) is hinged to the parallel arm, and the other end is hinged to the front small beam (501); The marine crane further comprises a force compensation device (3), the force compensation device (3) comprising a counterweight (303), a center rod (301) and a spring (302), one end of the center rod (301) being connected to the counterweight (303), the other end of the center rod (301) being hinged to the parallel arm, the counterweight (303) and the parallel arm being respectively located on two sides of the seat body (101), and a limit block (104) being fixedly arranged on the seat body (101), and the center rod (301) passing through the limit block (104); the spring (302) being sleeved on the center rod (301), and the spring (302) being located between the limit block (104) and the counterweight (303); The parallel arm comprises an upper parallel arm (401) and a lower parallel arm (402); the extension directions of the upper parallel arm (401) and the lower parallel arm (402) are always parallel; a wire-passing sliding hole (404) is provided between the upper parallel arm (401) and the lower parallel arm (402); and the suspension wire passes through the wire-passing sliding hole (404).

2. The hoisting control system according to claim 1, characterized in that: A binocular camera (504) is arranged on the front small beam (501), and the binocular camera (504) is used to obtain a position image of the suspended cargo.

3. A control method for a hoisting control system according to claim 1 or 2, characterized in that: The following steps are involved: Step 1: Set the crane working status; Step 2: Obtain the actual lifting height and actual lifting speed of the lifted cargo; and calculate the expected lifting height and expected lifting speed of the lifted cargo at this time; Step 3: According to the actual lifting height, actual lifting speed, expected lifting height and expected lifting speed of the lifted cargo, a state equation of the lifted cargo is constructed to obtain the lifting state of the lifted cargo; Step 4: Use The algorithm selects one of several lifting actuators to perform the action; Step 5: Adjust the lifting state of the suspended cargo according to the lifting driver selected in step 4, and then return to step 2 until the suspended cargo is stably lifted.

4. The control method according to claim 3, characterized in that: The state equation of the suspended cargo in step 3 is: ; in, is the hoisting height difference gain, is the hoisting speed difference gain, is an additional increment, is the actual lifting height of the cargo being hoisted, is the actual lifting speed of the lifted cargo, is the expected lifting height of the cargo to be lifted, It is the expected lifting speed of the cargo to be lifted.

5. The control method according to claim 4, characterized in that: The execution action of the lifting driver belongs to the action set , ,in represents the extension of the main hydraulic cylinder (6), Indicates the shortening of the main hydraulic cylinder (6). represents the extension of the auxiliary hydraulic cylinder (7), Indicates the shortening of the auxiliary hydraulic cylinder (7). Indicates the tightening of the hanging wire on the capstan (202). It indicates that the hanging wire on the capstan (202) is loosened.

6. The control method according to claim 5, characterized in that: When the lifting driver performs the action in step 5, the lifted cargo enters a new lifting state. , Rewarded by the algorithm , update the value function of the execution action: ; in, is the learning rate, is the reduction factor, For the new lifting state The maximum value at which this action can be taken.

7. The control method according to claim 6, characterized in that: If the status item value of the new lifting status of the lifted cargo is less than the status item value of the original lifting status, the reward The reward is positive, and the smaller the value of the state item of the new hoisting state is, the higher the reward The larger the value is; if the value of the status item of the new lifting state of the lifted cargo is greater than the value of the status item of the original lifting state, the reward Negative reward.