A method for unloading a cold-rolled steel coil by an unmanned vehicle

Abstract

The application relates to the technical field of steel finished product intelligent logistics, in particular to a cold-rolled steel coil unloading method of an unmanned crane, which comprises the following steps: obtaining original design coordinates and self-learning compensation offsets of a target storage location, and calculating effective target coordinates; the unmanned crane travels to a coil unloading position in an empty state, clamps the steel coil and records the quality; an excitation pulse is applied, the torsional angular velocity and the swing angle are collected, the eccentric mass identification model is input, and the eccentric distance and the eccentric direction angle are output; the trajectory planning model is used to solve the acceleration instruction and control the crane to move to above the target storage location; the lifting tool is lowered until the steel coil contacts the ground, and the contact position coordinates are recorded; the self-learning compensation offsets are updated according to the deviation between the contact coordinates and the effective target coordinates; when the steel coil contacts the ground and the lifting tool is not completely loosened, the steel coil is rotated to be centered according to the target weld angle and the required angle of a packaging unit, and the unloading is completed. The application realizes self-adaptive compensation of eccentric disturbance and storage location deviation, and improves the unloading precision and consistency.

Description

Technical Field

[0001] This application relates to the field of intelligent logistics technology for finished steel products, specifically to an unmanned overhead crane method for uncoiling cold-rolled steel coils. Background Technology

[0002] Cold-rolled steel coils are a crucial finished product in the steel industry. After being slit by the rewinding unit, they need to be transported by unmanned overhead cranes to the ground storage area in the finished product warehouse for temporary storage, and then retrieved when needed by the packaging unit. With the advancement of intelligent manufacturing transformation in steel enterprises, the application of unmanned overhead crane systems in cold-rolled finished product warehouses is becoming increasingly widespread. Through the coordination of the scheduling system and the crane controller, automatic unloading, transfer, and coil placement of steel coils can be achieved, improving logistics efficiency and reducing labor costs. However, the unloading process of cold-rolled steel coils by unmanned overhead cranes involves multiple technical aspects such as heavy-load lifting, precise positioning, and posture control. The inherent center of gravity distribution characteristics of the steel coils, the long-term changes in the warehouse environment, and the dynamic coupling during the lifting process all place high demands on the final positional accuracy of the coil placement and the posture control of the steel coils.

[0003] In existing technologies, the automatic unloading scheme for unmanned overhead cranes transporting steel coils mainly involves: pre-storing the original design coordinates of each storage location; the crane controller obtaining the real-time position of the crane based on the absolute encoder on the crane and absolute position readers such as laser rangefinders or Gray line busbars on the track; and guiding the crane to the target storage location through closed-loop position control. For anti-sway control, open-loop input shaping technology is commonly used, which involves superimposing a specific pulse sequence into the crane speed command to reduce residual sway after stopping by suppressing the inherent frequency components of the lifting device. For steel coil angle control during unloading, a common practice is to install an independent rotating centering device at the entrance of the packaging unit, rotating the steel coil to the required angle after visually detecting the weld position.

[0004] However, the existing automated unloading schemes for steel coils using overhead cranes still have the following shortcomings in practical applications: During the rewinding process, uneven tension within the inner coils or uneven force during cutting may cause the center of gravity to deviate from its geometric center, resulting in eccentric mass; the ground in long-term used warehouses may undergo plastic deformation due to heavy load settlement, causing a deviation of several centimeters between the actual center position of the warehouse location and the original design coordinates; and the weld angle of the steel coil after unloading often does not match the feeding requirements of the packaging unit. Under the combined effect of these factors, the existing unloading method using overhead cranes cannot guarantee the positional accuracy and posture consistency of the steel coil when it is finally placed in the warehouse location. After unloading, the end face of the steel coil is prone to collision with the warehouse stop blocks and deviate from the effective range of the warehouse location, affecting production efficiency and equipment investment. Summary of the Invention

[0005] To address the technical challenges of existing automated unloading schemes for steel coils using unmanned overhead cranes, which struggle to handle the combined effects of torsional sway caused by coil eccentricity, storage location coordinate deviation due to ground subsidence, and inconsistencies between the unloading angle and loading requirements, thus failing to guarantee the positional accuracy and posture consistency of the unloading, this application provides an unmanned overhead crane unloading method for cold-rolled steel coils. By reading self-learning compensation offset to correct target coordinates, identifying eccentricity and inputting it into the model predictive control, updating compensation offset based on contact deviation, and performing angle alignment while the steel coil is still in contact with the ground, this method achieves adaptive compensation for eccentric disturbances and storage location deviations, as well as online calibration of the unloading posture. This significantly improves the positional accuracy, torsional resistance, and angle consistency of the unloading.

[0006] In a first aspect, this application provides an unmanned overhead crane method for uncoiling cold-rolled steel coils, comprising the following steps: S1. Obtain the original design coordinates of the target storage location and the target weld angle of the target steel coil, and read the self-learning compensation offset corresponding to the target storage location from the storage location management database. Calculate the effective target coordinates based on the original design coordinates and the self-learning compensation offset. The target storage location is the designated unwinding location for the target steel coil; S2. Control the unmanned crane to travel unloaded to the unloading position of the target steel coil, lower the crane's lifting device, clamp the target steel coil and lift it to a safe height, and record the weight of the target steel coil; S3. Apply an excitation pulse to the spreader, collect the torsional angular velocity and swing angle of the spreader in real time, input the torsional angular velocity, swing angle and the weight of the target steel coil into the eccentric mass identification model, and output the eccentricity and eccentricity direction angle of the center of gravity of the target steel coil relative to the center of the spreader through iterative calculation; Applying excitation pulses to the lifting device specifically involves controlling the unmanned vehicle to perform acceleration-uniform speed-deceleration operations sequentially along two different horizontal directions. S4. Input the effective target coordinates, eccentricity, and eccentricity direction angle into the trajectory planning model constructed based on model predictive control, solve for the acceleration command, and control the unmanned vehicle to move directly above the target storage location based on the acceleration command; S5. Lower the lifting device until the target steel coil contacts the ground at the target storage location, and record the coordinates of the contact position; S6. Based on the deviation between the contact position coordinates and the effective target coordinates, update the self-learning compensation offset of the target storage location and write the updated self-learning compensation offset into the storage location management database; S7. With the target steel coil in contact with the ground and the lifting device not fully released, rotate the lifting device to center the target steel coil according to the target angle of the weld and the preset angle required by the packaging unit, and then fully release the lifting device to complete the unwinding.

[0007] It should be further noted that the storage location management database stores the original design coordinates of each storage location and its corresponding self-learning compensation offset.

[0008] It should be further noted that in step S1, the target angle of the weld is the circumferential angle of the weld of the target steel coil relative to the outer reference mark of the steel coil.

[0009] It should be further noted that in step S1, the self-learning compensation offset is expressed as... ; The original design coordinates are represented as follows ; The formula for calculating the effective target coordinates is:

[0010]

[0011] in, The effective target coordinates are in the first horizontal direction; The component of the original design coordinates in the first horizontal direction; The component of the self-learning compensation offset in the first horizontal direction; The effective target coordinates are in the second horizontal direction; The component of the original design coordinates in the second horizontal direction; The component of the self-learning compensation offset in the second horizontal direction; The first horizontal direction is perpendicular to the second horizontal direction.

[0012] It should be further noted that the initial value of the self-learning compensation offset is... .

[0013] It should be further noted that in step S2, the safe height is ≥2.5m above the ground.

[0014] It should be further noted that the formula for calculating the weight m of the target steel coil is:

[0015] in, This is the total load reading after the lifting device has gripped and lifted the target steel coil. This is the total load reading when the spreader is unloaded; the total load reading at this time is the spreader's own weight.

[0016] It should be further explained that, in step S3, applying the excitation pulse to the spreader specifically involves controlling the unmanned vehicle to perform the same actions in the first and second horizontal directions sequentially: Control the unmanned vehicle from a stationary state at a speed of 0.2~0.3 m / s 2 The acceleration increases to 0.5~0.7 m / s, and the constant velocity is maintained for 0.4~0.6 s, then decreases to (-0.2)~(-0.3) m / s. 2 The acceleration decelerates to 0; The first horizontal direction is perpendicular to the second horizontal direction.

[0017] It should be further noted that in step S3, the torsional angular velocity is the rotational angular velocity of the spreader about the vertical axis, which is determined by real-time measurement of the torsional angle of the spreader about the vertical axis. The rate of change over time was obtained; The swing angle includes a first swing angle of the spreader in the first horizontal direction and a second swing angle in the second horizontal direction. The first swing angle is obtained by measuring the angle of the spreader in a vertical plane perpendicular to the first horizontal direction in real time; the second swing angle is obtained by measuring the angle of the spreader in a vertical plane perpendicular to the second horizontal direction in real time.

[0018] It should be further noted that in step S3, the eccentric mass identification model uses a torsional dynamics equation to describe the torsional motion of the lifting device and the target steel coil. The torsional dynamics equation is as follows:

[0019] in, The total moment of inertia of the lifting device and the target steel coiled around the vertical axis is obtained through no-load calibration. The damping coefficient; It is the acceleration due to gravity; It is the torsional angular velocity; This is the first swing angle; This is the second swing angle; It is the eccentricity; The eccentricity direction angle is defined as the angle between the eccentricity vector and the first horizontal direction.

[0020] It should be further noted that the damping coefficient c ranges from 0.05 to 0.15 N·m·s / rad.

[0021] It should be further explained that the solution process for the eccentricity quality identification model includes the following steps: S301. At each sampling time Calculate the measured torsional torque :

[0022] in, The sampling period; S302. Construct the observation matrix :

[0023] S303. Set the vector of parameters to be estimated :

[0024] S304. Iteratively update the parameter estimates according to the recursive formula, which is:

[0025]

[0026]

[0027] in, Sampling time The gain matrix; It is the vector of parameters to be estimated. At sampling time The estimated value Sampling time The covariance matrix; The preset forgetting factor; S305. When the number of sampling points calculated in the iterative calculation reaches the preset target number of sampling points or the estimated value of the parameter vector to be estimated. After convergence, based on the current vector of parameters to be estimated... Calculate the eccentricity and eccentricity direction angle:

[0028]

[0029] in, The vector of parameters to be estimated The first component, corresponding to ; The vector of parameters to be estimated The second component corresponds to ; The number of sampling points corresponds to the number of data points collected within the duration of the excitation pulse.

[0030] It should be further noted that the sampling period The value range is 0.002~0.01s; Forgetting factor The value range is 0.95 to 0.99; The target number of sampling points is 300-500.

[0031] It should be further noted that in step S4, the trajectory planning model adopts model predictive control, and the state variables of this model predictive control include the position of the unmanned vehicle in the first horizontal direction. Position in the second horizontal direction The first swing angle of the lifting device in the first horizontal direction The second swing angle in the second horizontal direction The angle of torsion of the lifting device about the vertical axis ; The control variable is the acceleration of the autonomous vehicle in the first horizontal direction. Acceleration of autonomous vehicles in the second horizontal direction ; The motion equations used in model predictive control include the position change equation of the unmanned vehicle, the swing equation of the spreader, and the torsional equation of the spreader. The torsional equation of the spreader is as follows:

[0032] The moment of inertia of the lifting device and the target steel coiled around the vertical axis; This is the torsional damping coefficient; The weight of the target steel coil; The objective function for model predictive control is:

[0033] in, To predict the number of sampling points in the time domain; The effective target coordinates are the components in the first horizontal direction; The effective target coordinates are the components in the second horizontal direction.

[0034] It should be further explained that the equation for the position change of the autonomous vehicle is:

[0035]

[0036] The equation for the swing of the lifting device is:

[0037]

[0038] in The pre-calibrated oscillation damping ratio; The natural frequency of the spreading device is calculated using the following formula:

[0039] The effective length of the rope is the vertical distance between the point where the rope exits from the rope guide wheel of the unmanned vehicle lifting mechanism and the point where the lifting device connects to the rope. This is the acceleration due to gravity.

[0040] It should be further noted that the sampling period for model predictive control is 50ms, the prediction time domain is 2s, and there are 40 sampling points.

[0041] It should be further noted that the constraints of model predictive control include:

[0042]

[0043] The autonomous vehicle's movement area does not enter the restricted zone extending 0.5 meters outward from the pre-marked obstacle coordinates; When the distance between the autonomous vehicle and the center of the target storage location is less than 1 meter, the weight coefficient of the position term in the objective function is automatically increased to 5 times the original value.

[0044] It should be further noted that model predictive control uses the interior-point method to solve the optimization problem in each sampling period to obtain the optimal acceleration command. and After being integrated, the data is sent to the autonomous vehicle's drive system and analyzed in real time. , , The state variables are reinitialized and solved in each sampling period to form a closed-loop control.

[0045] It should be further noted that in step S5, when lowering the spreader, the spreader is first lowered at a speed of 0.3 m / s until the target steel coil is ≥200 mm above the ground, and then the spreader is lowered at a speed of ≤0.05 m / s until the target steel coil contacts the ground.

[0046] It should be further noted that, in step S5, the condition for determining that the target steel coil has come into contact with the ground of the target storage location is: The following conditions must be met for three consecutive sampling periods:

[0047] The absolute value of the vertical acceleration of the spreader exceeds ; in, Sampling time The load reading of the lifting device; This is the acceleration due to gravity.

[0048] It should be further noted that in step S6, the deviation between the contact position coordinates and the effective target coordinates includes a first horizontal deviation δx and a second horizontal deviation δy, calculated using the following formulas:

[0049]

[0050] in, The component of the contact position coordinates in the first horizontal direction; The component of the contact position coordinates in the second horizontal direction; The rule for updating the self-learning compensation offset of the target storage location is as follows: when or When the offset exceeds 5mm, update the self-learning compensation offset according to the following formula:

[0051]

[0052] in, The component of the updated self-learning compensation offset in the first horizontal direction; The updated self-learning compensation offset component in the second horizontal direction α is the preset learning rate.

[0053] It should be further noted that the learning rate α ranges from 0.3 to 0.7.

[0054] It should be further noted that in step S7, the rotating spreader drives the target steel coil to rotate at an angle for centering. The calculation formula is:

[0055] in, The angle is set according to the preset requirements of the packaging unit; The target angle for the weld.

[0056] Secondly, this application provides an unmanned crane uncoiling system for cold-rolled steel coils, used to implement the aforementioned unmanned crane uncoiling method, including: The effective target coordinate calculation module is used to obtain the original design coordinates of the target storage location and the weld target angle of the target steel coil, and read the self-learning compensation offset corresponding to the target storage location from the storage location management database, and calculate the effective target coordinates based on the original design coordinates and the self-learning compensation offset. The target steel coil clamping module is used to control the unmanned vehicle to travel unloaded to the unloading position of the target steel coil, lower the lifting device of the unmanned vehicle, clamp the target steel coil and lift the target steel coil to a safe height, and record the weight of the target steel coil. The eccentricity parameter generation module is used to apply excitation pulses to the spreader, collect the torsional angular velocity and swing angle of the spreader in real time, input the torsional angular velocity, swing angle and the weight of the target steel coil into the eccentricity mass identification model, and output the eccentricity distance and eccentricity direction angle of the center of gravity of the target steel coil relative to the center of the spreader through iterative calculation. The trajectory planning and execution module is used to input the effective target coordinates, eccentricity and eccentricity direction angle into the trajectory planning model constructed based on model predictive control, and output acceleration commands after solving. Based on the acceleration commands, the unmanned vehicle is controlled to move directly above the target storage location. The contact position coordinate recording module is used to lower the lifting device until the target steel coil makes contact with the ground of the target storage location and record the contact position coordinates. The self-learning compensation offset update module is used to update the self-learning compensation offset of the target storage location based on the deviation between the contact position coordinates and the effective target coordinates, and write the updated self-learning compensation offset into the storage location management database. The angle alignment module is used to rotate the lifting device to center the target steel coil at an angle when the target steel coil has already contacted the ground and the lifting device is not fully released, based on the target angle of the weld and the preset angle required by the packaging unit. Then, the lifting device is fully released to complete the unwinding of the coil.

[0057] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described unmanned vehicle unloading method.

[0058] Fourthly, this application provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described unmanned vehicle unloading method.

[0059] As can be seen from the above technical solutions, this application has the following advantages: 1. This application calculates the effective target coordinates by reading the self-learning compensation offset and combining it with the original design coordinates. Before hoisting, it identifies the eccentricity and eccentricity direction angle of the steel coil and inputs them together with the effective target coordinates into a trajectory planning model based on model predictive control to solve for acceleration commands. After the steel coil contacts the ground, it updates the self-learning compensation offset of the storage position according to the contact deviation. When the steel coil has contacted the ground and the lifting device has not been fully released, it rotates and centers according to the weld target angle. This forms a closed-loop control of the entire process from task reception, hoisting, contact detection to coordinate compensation and angle calibration, which significantly improves the position accuracy of the coil drop, the attitude control effect and the system's adaptability to complex working conditions.

[0060] 2. This application applies excitation pulses along two different horizontal directions to the spreader after it grips and lifts the steel coil to a safe height. The torsional angular velocity and swing angle of the spreader are collected in real time. The torsional angular velocity, swing angle, and steel coil mass are input into an eccentric mass identification model. Through iterative calculation, the eccentricity and eccentricity direction angle of the steel coil's center of gravity relative to the spreader center are output. Subsequently, the effective target coordinates, eccentricity, and eccentricity direction angle are input into a trajectory planning model constructed based on model predictive control. The acceleration command is then obtained and used to control the movement of the unmanned crane. This allows the optimization process of the acceleration command to explicitly consider the influence of eccentric mass on the spreader's torsional motion, achieving active anti-torsional control of the eccentric steel coil. This effectively reduces the residual torsional angle of the spreader after an emergency stop of the crane and reduces the risk of collision between the steel coil end face and the storage block during coil unloading.

[0061] 3. This application obtains the original design coordinates of the target storage location, then reads the corresponding self-learning compensation offset from the storage location management database, calculates the effective target coordinates based on the original design coordinates and the self-learning compensation offset; after lowering the lifting device until the steel coil contacts the ground, the contact position coordinates are recorded, and the self-learning compensation offset of the target storage location is updated based on the deviation between the contact position coordinates and the effective target coordinates. The updated self-learning compensation offset is then written into the storage location management database. This allows the actual contact position of each coil drop to provide feedback and correct the coordinate offset of the corresponding storage location. Subsequent coil drops automatically compensate for deviations caused by factors such as ground subsidence. After multiple coil drops, the accuracy of the storage location coordinates is significantly improved, ensuring that the steel coil can be accurately placed into the storage location base without scratching adjacent steel coils.

[0062] 4. This application achieves angle alignment of the steel coil by rotating the lifting device while the steel coil is in contact with the ground and the lifting device is not fully released, based on the target angle of the weld and the preset angle required by the packaging unit. This utilizes the friction generated by the contact between the steel coil and the ground to suppress swaying disturbances during rotation, avoiding the additional risks caused by suspended rotation. At the same time, the angle calibration process is integrated into the unwinding process, so that the steel coil can directly meet the feeding requirements of the packaging unit after being unwound, without the need to add an additional rotation alignment device at the entrance of the packaging unit, reducing equipment investment and shortening the operating cycle of the packaging unit. Attached Figure Description

[0063] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 This is a flowchart of an unmanned crane unwinding method for cold-rolled steel coils in one embodiment of this application.

[0065] Figure 2 This is a schematic block diagram of an unmanned crane unwinding system for cold-rolled steel coils in one embodiment of this application.

[0066] Figure 3 This is a schematic diagram of the hardware structure of an electronic device in one embodiment of this application. Detailed Implementation

[0067] To make the purpose, features, and advantages of this application more apparent and understandable, specific embodiments and accompanying drawings will be used to clearly and completely describe the technical solution protected by this application. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0068] The unmanned vehicle unloading method of this application will be described in detail below. Specific details such as particular system structures and technologies are presented for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details.

[0069] In the unmanned vehicle unwinding method of this application, the term "comprising" indicates the presence of the described feature, integral, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0070] To facilitate a clear description of the technical solutions of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" do not necessarily imply that they are different.

[0071] The terms "one embodiment" or "some embodiments" used in this application mean that one or more embodiments of this application include the specific features, structures, or characteristics described in that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this application do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized.

[0072] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0073] The unmanned crane unwinding method provided in this application embodiment is executed by computer equipment, and correspondingly, the unmanned crane unwinding system for cold-rolled steel coils runs in the computer equipment.

[0074] Figure 1 This is a flowchart of an unmanned crane uncoiling method for cold-rolled steel coils according to an embodiment of this application. Wherein, Figure 1 The implementing entity can be an unmanned vehicle unloading system. Depending on different requirements, the order of the steps in this flowchart can be changed, and some can be omitted.

[0075] like Figure 1 As shown, the unmanned crane uncoiling method for this cold-rolled steel coil includes: Step S1: Obtain the original design coordinates of the target storage location and the target weld angle of the target steel coil, and read the self-learning compensation offset corresponding to the target storage location from the storage location management database. Calculate the effective target coordinates based on the original design coordinates and the self-learning compensation offset. The target storage location is the designated storage location for unwinding the target steel coil.

[0076] By acquiring the original design coordinates of the target storage location and the target weld angle of the target steel coil, and reading the self-learning compensation offset corresponding to the target storage location from the storage location management database, and then calculating the effective target coordinates based on the original design coordinates and the self-learning compensation offset, the target position used for subsequent trolley movement is no longer the fixed original design coordinates, but a self-learning correction value that incorporates historical coiling experience. Thus, without the need for manual measurement of ground settlement, the plastic deformation generated by long-term use of the storage location is automatically included in the target coordinate calculation, improving the authenticity of the target coordinates and the initial accuracy of coiling positioning.

[0077] In some specific embodiments, the storage location management database stores the original design coordinates of each storage location and its corresponding self-learning compensation offset.

[0078] The storage location management database independently stores the original design coordinates and corresponding self-learning compensation offset for each storage location. This ensures that the coordinate correction values ​​between different storage locations do not interfere with each other. An offset caused by ground subsidence or equipment collision in one storage location will not be incorrectly applied to other unaffected storage locations. At the same time, the centralized storage method of the database allows the unmanned vehicle controller to quickly read the complete coordinate information of the required storage location at the beginning of each unloading task and write the updated compensation offset back to the database after the task is completed. This ensures the persistent storage of self-learning data and the continuity across tasks. In some specific embodiments, the weld target angle is the circumferential angle of the weld of the target steel coil relative to the outer reference mark of the steel coil.

[0079] By defining the target weld angle as the circumferential angle of the weld of the target steel coil relative to the outer reference mark of the steel coil, this angle parameter can directly correspond to the actual measurement results of the weld detection device at the rewinding unit exit. There is no need for additional coordinate transformation or angle conversion in the trolley control system, thereby reducing the links that introduce errors in the data transmission process. Furthermore, this definition clearly defines a unique and repeatable location marker of the weld on the circumference of the steel coil, providing a clear target reference for the subsequent angle alignment steps and ensuring that the rotation operation can accurately adjust the weld to the orientation required by the packaging unit.

[0080] In some specific embodiments, the self-learning compensation offset is expressed as ; The original design coordinates are represented as follows ; The formula for calculating the effective target coordinates is:

[0081]

[0082] in, The effective target coordinates are in the first horizontal direction; The component of the original design coordinates in the first horizontal direction; The component of the self-learning compensation offset in the first horizontal direction; The effective target coordinates are in the second horizontal direction; The component of the original design coordinates in the second horizontal direction; The component of the self-learning compensation offset in the second horizontal direction; The first horizontal direction is perpendicular to the second horizontal direction.

[0083] By decomposing the self-learning compensation offset into two mutually perpendicular horizontal components and expressing the effective target coordinates as a linear superposition of the original design coordinates and the compensation offset, the correction of the storage location coordinates can be performed independently and uncoupled in the direction of the large vehicle's movement and the direction of the small vehicle's movement. This adapts to the non-uniform deformation that may exist in different directions of ground settlement. At the same time, this linear superposition calculation method is extremely simple and efficient, without the need for complex nonlinear transformations or iterative solutions, reducing the real-time computational burden of the unmanned vehicle controller.

[0084] In some specific embodiments, the initial value of the self-learning compensation offset is .

[0085] By setting the initial value of the self-learning compensation offset to zero, the effective target coordinates are directly equal to the original design coordinates when a new storage location is first put into use or when there are no historical records in the database. That is, the first roll-out is executed normally according to the design position. This initial value does not introduce any bias from human assumptions, ensuring the robustness of the system when there is a lack of learning experience. At the same time, starting from zero also makes it easier for the deviation calculated after the first contact to fully and accurately reflect the actual offset, avoiding the problem of needing multiple iterations to converge due to improper selection of the initial compensation value.

[0086] Step S2: Control the unmanned crane to travel unloaded to the unloading position of the target steel coil, lower the crane's lifting device, clamp the target steel coil and lift it to a safe height, and record the weight of the target steel coil.

[0087] By controlling the unmanned overhead crane to travel unloaded to the unloading position of the target steel coil, lowering the spreader to clamp the target steel coil and raising it to a safe height, and recording the weight of the target steel coil at the same time, the weight parameters of the steel coil required for subsequent eccentric mass identification and trajectory planning can be obtained in real time after actual clamping. This avoids the errors caused by relying on preset weight data. Furthermore, raising the steel coil to a safe height ensures that the steel coil will not collide with the ground or surrounding equipment during subsequent excitation pulse application and loaded travel, providing reliable weight input and a safe working space for eccentric identification and anti-sway control.

[0088] In some specific embodiments, the safe height is ≥2.5m above the ground.

[0089] By setting the safety height to no less than 2.5m above the ground, a sufficient vertical distance is maintained between the bottom of the steel coil and the ground after the spreader grips and lifts it. This distance can fully accommodate any equipment protrusions, steel coil saddles, or passageways that may exist below. At the same time, during subsequent application of excitation pulses for eccentricity identification and during loaded travel, even if the spreader swings or twists to a certain extent due to acceleration or deceleration, the steel coil will not accidentally touch the ground. This provides a safe operating height guarantee for the two key steps of eccentricity mass identification and trajectory planning.

[0090] In some specific embodiments, the formula for calculating the weight m of the target steel coil is:

[0091] in, This is the total load reading after the lifting device has gripped and lifted the target steel coil. This is the total load reading when the spreader is unloaded; the total load reading at this time is the spreader's own weight.

[0092] The steel coil mass is calculated by subtracting the weight of the spreader itself when unloaded from the total load reading after the spreader has gripped and lifted the coil. This eliminates the need for additional specialized weighing sensors; instead, the existing load detection device on the gantry crane spreader is used directly. Furthermore, the difference calculation automatically eliminates the influence of the spreader's own weight on the mass reading. This simplifies the system hardware configuration and reduces equipment costs and maintenance complexity while ensuring the accuracy of mass measurement.

[0093] Step S3: Apply an excitation pulse to the spreader, collect the torsional angular velocity and swing angle of the spreader in real time, input the torsional angular velocity, swing angle and the weight of the target steel coil into the eccentric mass identification model, and output the eccentricity and eccentricity direction angle of the target steel coil's center of gravity relative to the center of the spreader through iterative calculation. Applying excitation pulses to the lifting device specifically involves controlling the unmanned vehicle to perform acceleration-uniform speed-deceleration operations sequentially along two different horizontal directions.

[0094] By applying acceleration-uniform speed-deceleration excitation pulses sequentially along two different horizontal directions to the spreader, the torsional angular velocity and swing angle of the spreader are collected in real time. The torsional angular velocity, swing angle, and weight of the target steel coil are input into the eccentricity mass identification model. After iterative calculation, the eccentricity distance and eccentricity direction angle of the target steel coil's center of gravity relative to the spreader center are output. This allows the unmanned crane to proactively know the eccentricity characteristics of the steel coil before lifting begins, without relying on external measuring equipment or factory preset data. This solves the problem that eccentricity mass cannot be directly measured or estimated, and provides accurate eccentricity parameters for anti-torsion control during subsequent loaded travel.

[0095] In some specific embodiments, in step S3, applying an excitation pulse to the lifting device specifically involves controlling the unmanned vehicle to perform the same actions in the first horizontal direction and the second horizontal direction sequentially: Control the unmanned vehicle from a stationary state at a speed of 0.2~0.3 m / s 2 The acceleration increases to 0.5~0.7 m / s, and the constant velocity is maintained for 0.4~0.6 s, then decreases to (-0.2)~(-0.3) m / s. 2 The acceleration decelerates to 0; The first horizontal direction is perpendicular to the second horizontal direction.

[0096] By sequentially applying the same excitation pulse consisting of an acceleration phase, a constant speed holding phase, and a deceleration phase in two mutually perpendicular horizontal directions, and optimizing the acceleration and velocity amplitudes of each phase, the applied pulses can generate sufficiently rich torsional dynamic response signals without excessively exciting the spreader to swing significantly. This ensures that the eccentricity mass identification model can effectively extract eccentricity and eccentricity direction angle information from the collected data. At the same time, the total duration of the pulses is short and will not significantly affect the cycle time of the entire unwinding operation.

[0097] In some specific embodiments, the torsional angular velocity is the rotational angular velocity of the spreader about the vertical axis, measured in real time by the torsional angle of the spreader about the vertical axis. The rate of change over time was obtained; The swing angle includes a first swing angle of the spreader in the first horizontal direction and a second swing angle in the second horizontal direction. The first swing angle is obtained by measuring the angle of the spreader in a vertical plane perpendicular to the first horizontal direction in real time; the second swing angle is obtained by measuring the angle of the spreader in a vertical plane perpendicular to the second horizontal direction in real time.

[0098] By defining the torsional angular velocity as the rotational angular velocity of the spreader about the vertical axis and obtaining it from the rate of change of the torsional angle over time, and decomposing the swing angle into two independent swing angles corresponding to the two horizontal directions respectively, the spatial attitude of the spreader is completely and uncoupledly decomposed into torsional motion about the vertical axis and swing motion about the two horizontal axes. This decomposition method is completely consistent with the variable definitions in the torsional dynamics equations used in the subsequent eccentric mass identification model, thus ensuring the physical correspondence between the collected data and the mathematical model and improving the accuracy of the identification results.

[0099] In some specific embodiments, the eccentric mass identification model uses a torsional dynamics equation to describe the torsional motion of the lifting device and the target steel coil. The torsional dynamics equation is as follows:

[0100] in, The total moment of inertia of the lifting device and the target steel coiled around the vertical axis is obtained through no-load calibration. The damping coefficient; It is the acceleration due to gravity; It is the torsional angular velocity; This is the first swing angle; This is the second swing angle; It is the eccentricity; The eccentricity direction angle is defined as the angle between the eccentricity vector and the first horizontal direction.

[0101] The torsional motion of the lifting device and the steel coil is described by a torsional dynamics equation consisting of a product of total moment of inertia and torsional angular acceleration, a damping term, and an excitation term composed of the coil mass, gravitational acceleration, eccentricity, eccentricity direction angle, and swing angle. The total moment of inertia is obtained through no-load calibration, so that the influence of the eccentric mass on the torsional motion is explicitly expressed as a coupling relationship between the measurable swing angle and the eccentricity parameter to be identified. This provides a mathematical model basis for the algorithm that accurately reflects the physical laws, and the eccentricity parameter can be solved in reverse from the measured torsional angular velocity and swing angle.

[0102] In some specific embodiments, the damping coefficient c ranges from 0.05 to 0.15 N·m·s / rad.

[0103] By limiting the damping coefficient in the torsional dynamics equation to an optimal range of 0.05~0.15 N·m·s / rad, which was calibrated through numerous no-load and loaded tests, the damping term in the model can truly reflect the energy dissipation characteristics of the lifting device and steel coil system under actual working conditions. Selecting the damping coefficient within this optimal range can balance the convergence speed and steady-state accuracy of the identification algorithm, ensuring the reliability of eccentric mass identification.

[0104] In some specific embodiments, the solution process for the eccentricity mass identification model includes the following steps: S301. At each sampling time Calculate the measured torsional torque :

[0105] in, The sampling period; S302. Construct the observation matrix :

[0106] S303. Set the vector of parameters to be estimated :

[0107] S304. Iteratively update the parameter estimates according to the recursive formula, which is:

[0108]

[0109]

[0110] in, Sampling time The gain matrix; It is the vector of parameters to be estimated. At sampling time The estimated value Sampling time The covariance matrix; The preset forgetting factor; S305. When the number of sampling points calculated in the iterative calculation reaches the preset target number of sampling points or the estimated value of the parameter vector to be estimated. After convergence, based on the current vector of parameters to be estimated... Calculate the eccentricity and eccentricity direction angle:

[0111]

[0112] in, The vector of parameters to be estimated The first component, corresponding to ; The vector of parameters to be estimated The second component corresponds to ; The number of sampling points corresponds to the number of data points collected within the duration of the excitation pulse.

[0113] By calculating the measured torsional torque based on the torsional angular velocity at each sampling moment and the previous moment, an observation matrix consisting of swing angle and mass is constructed. The parameter vector to be estimated is set, and iterative calculation is performed according to the update formulas of gain matrix, parameter estimation and covariance matrix. Finally, the eccentricity and eccentricity direction angle are calculated from the converged parameter vector, so that the identification process can be carried out online in real time without storing all historical data. Furthermore, the introduction of the forgetting factor allows the algorithm to gradually reduce the influence of old data, thereby quickly converging to a stable parameter estimate within the finite duration of the excitation pulse.

[0114] In some specific embodiments, the sampling period The value range is 0.002~0.01s; Forgetting factor The value range is 0.95 to 0.99; The target number of sampling points is 300-500.

[0115] By limiting the range of values ​​for the sampling period, forget factor, and target number of sampling points, the sampling frequency is sufficient to capture the dynamic changes in torsional angular velocity and swing angle. Within this range, the forget factor can achieve a balance between quickly tracking parameter changes and suppressing measurement noise, while the target number of sampling points ensures that enough iteration steps are obtained to converge, without prolonging the identification time and affecting the efficiency of the operation due to too many sampling points.

[0116] Step S4: Input the effective target coordinates, eccentricity, and eccentricity direction angle into the trajectory planning model constructed based on model predictive control. After solving, the acceleration command is output. Based on the acceleration command, the unmanned vehicle is controlled to move directly above the target storage location.

[0117] By inputting the effective target coordinates, eccentricity, and eccentricity direction angle into a trajectory planning model constructed based on model predictive control, the model outputs acceleration commands after solving the model. Based on these acceleration commands, the unmanned tractor moves to the top of the target storage location. This ensures that the acceleration decision at every moment during the tractor's loaded journey explicitly considers the influence of the eccentric mass of the steel coil on the torsional motion of the spreader. Furthermore, model predictive control can predict the system state over a future time domain and adjust the control inputs in advance. This proactively suppresses torsional sway caused by eccentricity during the tractor's acceleration, constant speed, and deceleration phases, improving the positioning accuracy and attitude stability of the tractor when it reaches the top of the target storage location.

[0118] In some specific embodiments, the trajectory planning model employs model predictive control, and the state variables of this model predictive control include the position of the unmanned vehicle in the first horizontal direction. Position in the second horizontal direction The first swing angle of the lifting device in the first horizontal direction The second swing angle in the second horizontal direction The angle of torsion of the lifting device about the vertical axis ; The control variable is the acceleration of the autonomous vehicle in the first horizontal direction. Acceleration of autonomous vehicles in the second horizontal direction ; The motion equations used in model predictive control include the position change equation of the unmanned vehicle, the swing equation of the spreader, and the torsional equation of the spreader. The torsional equation of the spreader is as follows:

[0119] The moment of inertia of the lifting device and the target steel coiled around the vertical axis; This is the torsional damping coefficient; The weight of the target steel coil; The objective function for model predictive control is:

[0120] in, To predict the number of sampling points in the time domain; The effective target coordinates are the components in the first horizontal direction; The effective target coordinates are the components in the second horizontal direction.

[0121] By setting the state variables of model predictive control to the position of the gantry in two horizontal directions, the swing angle of the spreader in two directions, and the torsional angle of the spreader about the vertical axis, and setting the control variables to the acceleration of the gantry in two directions, and by including the square term of position deviation, the square penalty term of swing angle, the square penalty term of torsional angle, and the square penalty term of acceleration in the optimization objective function, and by incorporating the eccentric torque term in the torsional dynamics equation into the prediction model, model predictive control can explicitly consider the torsional excitation caused by eccentric mass when predicting the future system state, thereby generating the optimal acceleration command that can suppress both swing and torsion.

[0122] In some specific embodiments, the equation for the position change of the autonomous vehicle is:

[0123]

[0124] The equation for the swing of the lifting device is:

[0125]

[0126] in The pre-calibrated oscillation damping ratio; The natural frequency of the spreading device is calculated using the following formula:

[0127] The effective length of the rope is the vertical distance between the point where the rope exits from the rope guide wheel of the unmanned vehicle lifting mechanism and the point where the lifting device connects to the rope. This is the acceleration due to gravity.

[0128] By adopting the second-order integral relationship between the vehicle position and acceleration as the position change equation, and the standard second-order swing dynamics equation including damping ratio and natural frequency as the swing equation of the spreader, where the natural frequency is determined by the effective rope length and gravitational acceleration, and the swing damping ratio is obtained through pre-calibration, the kinematic relationship in the trajectory planning model can accurately describe the second-order dynamic coupling characteristics between the vehicle acceleration and the swing angle of the spreader. This enhances the matching accuracy of the anti-sway control to the mechanical characteristics of specific equipment and improves the actual control effect.

[0129] In some specific embodiments, the sampling period for model predictive control is 50ms, the prediction time domain is 2s, and there are 40 sampling points.

[0130] By setting the sampling period of model predictive control to 50ms and the prediction time domain to 2s corresponding to 40 sampling points, the decision frequency of the controller is sufficient to respond to the inherent frequency changes of the sway and torsion of the spreader. At the same time, the two-second prediction time domain covers the main dynamic process of the vehicle from the start of deceleration to complete stop. The 40 sampling points in the prediction time domain can provide enough control steps while ensuring that the computational complexity of the optimization problem is controllable. This allows model predictive control to plan the complete deceleration strategy in advance without frequent adjustments in the middle, thus improving the stability and real-time performance of the control.

[0131] In some specific embodiments, the constraints of model predictive control include:

[0132]

[0133] The autonomous vehicle's movement area does not enter the restricted zone extending 0.5 meters outward from the pre-marked obstacle coordinates; When the distance between the autonomous vehicle and the center of the target storage location is less than 1 meter, the weight coefficient of the position term in the objective function is automatically increased to 5 times the original value.

[0134] By limiting the vehicle's acceleration in both horizontal directions to a safe range to prevent excessive acceleration and deceleration from causing violent swaying, and by setting the movement area to avoid entering a restricted zone extending outward from the pre-calibrated obstacle coordinates to ensure a safe distance from obstacles such as pillars and maintenance areas within the parking area, and by automatically increasing the weight coefficient of the position term in the objective function to 5 times the original value when the distance between the vehicle and the center of the target parking space is less than the set value, the vehicle can more accurately track the effective target coordinates in the final stage of approaching the target parking space, thereby simultaneously improving driving safety, obstacle avoidance ability, and final parking accuracy.

[0135] In some specific embodiments, model predictive control uses the interior-point method to solve the optimization problem in each sampling period to obtain the optimal acceleration command. and After being integrated, the data is sent to the autonomous vehicle's drive system and analyzed in real time. , , The state variables are reinitialized and solved in each sampling period to form a closed-loop control.

[0136] By employing the interior point method to solve the optimization problem in each sampling cycle, the optimal acceleration command is obtained and sent to the drive system after integration. Simultaneously, based on the real-time measured swing and torsional angles, the state variables are re-initialized in each sampling cycle and solved to form a closed-loop control. This enables the model predictive control to have feedback correction capability. When the actual swing or torsion of the spreader deviates from the predictive model, the next sampling cycle will re-optimize based on the latest measurement value to promptly correct the control command. This rolling time domain control strategy effectively overcomes the cumulative effects of model errors and external disturbances, ensuring the robustness and accuracy of trajectory planning throughout the entire process.

[0137] Step S5: Lower the lifting device until the target steel coil contacts the ground at the target storage location, and record the coordinates of the contact position.

[0138] By lowering the spreader until the target steel coil makes contact with the ground of the target storage location and recording the contact position coordinates, the unmanned crane can use the actual position at the moment of physical contact to verify or correct the crane position previously calculated based on the encoder and model. Since the spreader is very low and the descent speed is slow when contact occurs, the contact position coordinates can accurately reflect the relative position relationship between the bottom center of the steel coil and the ground of the storage location, thus providing highly reliable measured feedback data for the subsequent self-learning update of the storage location coordinates.

[0139] In some specific embodiments, when lowering the spreader, the spreader is first lowered at a speed of 0.3 m / s until the target steel coil is ≥200 mm above the ground, and then the spreader is lowered at a speed of ≤0.05 m / s until the target steel coil contacts the ground.

[0140] By dividing the descent process into two stages, a faster descent speed is used in the first stage before the steel coil approaches the ground to reduce the total operation time, and an extremely slow descent speed is switched in the second stage when the steel coil is only a short distance from the ground. This ensures that the steel coil has a very low falling speed at the moment of contact with the ground, thereby avoiding damage to the end face of the steel coil caused by rapid impact with the ground, violent rebound of the lifting device, or excessive impact force at the moment of contact affecting the accurate detection of the contact position coordinates. This improves the safety of the descent process and the reliability of contact detection.

[0141] In some specific embodiments, the condition for determining that the target steel coil has come into contact with the ground of the target storage location is: The following conditions must be met for three consecutive sampling periods:

[0142] The absolute value of the vertical acceleration of the spreader exceeds ; in, Sampling time The load reading of the lifting device; This is the acceleration due to gravity.

[0143] By setting conditions for determining contact, the contact detection integrates the load drop characteristics of the load cell and the vertical acceleration impact characteristics of the inertial measurement unit. The dual conditions are used for joint determination and require that the conditions be met for multiple consecutive sampling cycles. This effectively avoids false triggering caused by sensor noise or accidental disturbances. At the same time, the two conditions confirm the contact event from different physical dimensions, which improves the reliability and anti-interference ability of the detection.

[0144] Step S6: Based on the deviation between the contact position coordinates and the effective target coordinates, update the self-learning compensation offset of the target storage location, and write the updated self-learning compensation offset into the storage location management database.

[0145] By updating the self-learning compensation offset of the target storage location based on the deviation between the contact position coordinates and the effective target coordinates, and writing the updated self-learning compensation offset into the storage location management database, the compensation offset of the storage location can be iteratively corrected based on the actual contact results after each roll-down operation. The corrected offset is persistently stored for subsequent roll-down tasks of the same storage location to read directly, thereby realizing the adaptive learning capability of the storage location coordinates. As the number of roll-down operations increases, the effective target coordinates gradually approach the true center position of the storage location.

[0146] In some specific embodiments, the deviation between the contact position coordinates and the effective target coordinates includes a first horizontal deviation δx and a second horizontal deviation δy, calculated using the following formulas:

[0147]

[0148] in, The component of the contact position coordinates in the first horizontal direction; The component of the contact position coordinates in the second horizontal direction; The rule for updating the self-learning compensation offset of the target storage location is as follows: when or When the offset exceeds 5mm, update the self-learning compensation offset according to the following formula:

[0149]

[0150] in, The component of the updated self-learning compensation offset in the first horizontal direction; The updated self-learning compensation offset component in the second horizontal direction α is the preset learning rate.

[0151] By decomposing the deviation between the contact position coordinates and the effective target coordinates into two horizontal components, and setting the update rule to update the compensation offset according to the formula including the learning rate only when the absolute value of the deviation of either component exceeds a preset threshold, the compensation update is only performed when the deviation exceeds a certain dead zone range, avoiding unnecessary updates caused by small random errors. At the same time, the introduction of the learning rate ensures that each update only corrects a part of the deviation, playing a filtering and smoothing role and preventing a single abnormal deviation from causing a large jump in the compensation value.

[0152] In some specific embodiments, the learning rate α ranges from 0.3 to 0.7.

[0153] By limiting the learning rate to the preferred range of 0.3 to 0.7, the update of the self-learning compensation offset is neither too slow to converge due to an excessively small learning rate, requiring multiple roll-ups to approach the true offset, nor too large to cause excessive amplification of the single deviation measurement error, resulting in oscillation of the compensation value. The learning rate within this preferred range enables the storage location coordinate compensation to converge to a high accuracy after a reasonable number of roll-ups, while maintaining good noise resistance and adaptive tracking capability for slow and continuous ground subsidence.

[0154] Step S7: With the target steel coil in contact with the ground and the lifting device not fully released, rotate the lifting device to center the target steel coil according to the target angle of the weld and the preset angle required by the packaging unit, and then fully release the lifting device to complete the unwinding.

[0155] By rotating the lifting device to center the target steel coil at an angle according to the target weld angle and the preset packaging unit requirements while the target steel coil is in contact with the ground and the lifting device is not fully released, the lifting device is then fully released to complete the uncoiling. This angle centering process utilizes the frictional resistance generated by the contact between the steel coil and the ground to suppress the swing of the lifting device that may be caused by rotation. At the same time, since the lifting device is not yet released, the rotational driving force can be directly transmitted to the steel coil through the clamps, avoiding the inertial swaying and eccentric torque interference when the steel coil is suspended and rotated. Thus, the angle calibration is completed in one go at the final stage of uncoiling, ensuring that the weld orientation of the steel coil after uncoiling meets the feeding requirements of the subsequent packaging unit.

[0156] In some specific embodiments, the rotating spreader drives the target steel coil to rotate at an angle for centering. The calculation formula is:

[0157] in, The angle is set according to the preset requirements of the packaging unit; The target angle for the weld.

[0158] By adopting The calculation is used to determine the rotation angle required for the rotating hoist to drive the steel coil for angle alignment. This ensures that regardless of the sign of the difference between the target weld angle and the required angle of the packaging unit, the shortest effective rotation path can be automatically obtained and mapped to the standard circumferential angle range. This avoids the problem of the steel coil rotating more than a full circle or rotating an unnecessary angle due to incorrect angle difference calculation, thus improving the efficiency and accuracy of the angle alignment step.

[0159] The following are embodiments of the unmanned crane uncoiling system for cold-rolled steel coils provided in this application. This unmanned crane uncoiling system for cold-rolled steel coils belongs to the same inventive concept as the unmanned crane uncoiling methods in the above embodiments. For details not described in detail in the embodiments of the unmanned crane uncoiling system, please refer to the embodiments of the unmanned crane uncoiling methods for cold-rolled steel coils described above.

[0160] like Figure 2 As shown, the unmanned overhead crane uncoiling system for cold-rolled steel coils includes: The effective target coordinate calculation module is used to obtain the original design coordinates of the target storage location and the weld target angle of the target steel coil, and read the self-learning compensation offset corresponding to the target storage location from the storage location management database, and calculate the effective target coordinates based on the original design coordinates and the self-learning compensation offset. The target steel coil clamping module is used to control the unmanned vehicle to travel unloaded to the unloading position of the target steel coil, lower the lifting device of the unmanned vehicle, clamp the target steel coil and lift the target steel coil to a safe height, and record the weight of the target steel coil. The eccentricity parameter generation module is used to apply excitation pulses to the spreader, collect the torsional angular velocity and swing angle of the spreader in real time, input the torsional angular velocity, swing angle and the weight of the target steel coil into the eccentricity mass identification model, and output the eccentricity distance and eccentricity direction angle of the center of gravity of the target steel coil relative to the center of the spreader through iterative calculation. The trajectory planning and execution module is used to input the effective target coordinates, eccentricity and eccentricity direction angle into the trajectory planning model constructed based on model predictive control, and output acceleration commands after solving. Based on the acceleration commands, the unmanned vehicle is controlled to move directly above the target storage location. The contact position coordinate recording module is used to lower the lifting device until the target steel coil makes contact with the ground of the target storage location and record the contact position coordinates. The self-learning compensation offset update module is used to update the self-learning compensation offset of the target storage location based on the deviation between the contact position coordinates and the effective target coordinates, and write the updated self-learning compensation offset into the storage location management database. The angle alignment module is used to rotate the lifting device to center the target steel coil at an angle when the target steel coil has already contacted the ground and the lifting device is not fully released, based on the target angle of the weld and the preset angle required by the packaging unit. Then, the lifting device is fully released to complete the unwinding of the coil.

[0161] The unmanned crane uncoiling system in this embodiment is used to realize the unmanned crane uncoiling method for cold-rolled steel coils.

[0162] This application also provides an electronic device for implementing the various embodiments of this application. Figure 3 To illustrate the hardware structure of an electronic device according to various embodiments of this application, as shown in the following diagram... Figure 3 As shown, the electronic device includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor.

[0163] Those skilled in the art will understand that the electronic device structure involved in the embodiments of this application does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0164] In embodiments of this application, electronic devices include, but are not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0165] In this application embodiment, the processor can be implemented using at least one of an Application-Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a processor, a controller, a microcontroller, a microprocessor, or an electronic unit designed to perform the functions described herein. In some cases, such implementations can be implemented within a controller. For software implementations, implementations such as processes or functions can be implemented with separate software modules that allow the performance of at least one function or operation. The software code can be implemented by a software application (or program) written in any suitable programming language, and the software code can be stored in memory and executed by the controller.

[0166] In addition, the electronic device includes some functional modules not shown, which will not be described in detail here.

[0167] Those skilled in the art will understand that the various aspects of the electronic device provided in this application can be implemented as a system, method, or program product. Therefore, the various aspects of this application can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0168] This application also provides a storage medium storing a program product capable of implementing an unmanned crane unwinding method for cold-rolled steel coils. In some possible embodiments, various aspects of this application can also be implemented as a program product comprising program code that, when run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this application.

[0169] The storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0170] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for unloading cold-rolled steel coils using an unmanned crane, characterized in that, Includes the following steps: S1. Obtain the original design coordinates of the target storage location and the target weld angle of the target steel coil, and read the self-learning compensation offset corresponding to the target storage location from the storage location management database. Calculate the effective target coordinates based on the original design coordinates and the self-learning compensation offset. S2. Control the unmanned crane to travel unloaded to the unloading position of the target steel coil, lower the crane's lifting device, clamp the target steel coil and lift it to a safe height, and record the weight of the target steel coil; S3. Apply an excitation pulse to the spreader, collect the torsional angular velocity and swing angle of the spreader in real time, input the torsional angular velocity, swing angle and the weight of the target steel coil into the eccentric mass identification model, and output the eccentricity and eccentricity direction angle of the center of gravity of the target steel coil relative to the center of the spreader through iterative calculation; Applying excitation pulses to the lifting device specifically involves controlling the unmanned vehicle to perform acceleration-uniform speed-deceleration operations sequentially along two different horizontal directions. S4. Input the effective target coordinates, eccentricity, and eccentricity direction angle into the trajectory planning model constructed based on model predictive control, solve for the acceleration command, and control the unmanned vehicle to move directly above the target storage location based on the acceleration command; S5. Lower the lifting device until the target steel coil contacts the ground at the target storage location, and record the coordinates of the contact position; S6. Based on the deviation between the contact position coordinates and the effective target coordinates, update the self-learning compensation offset of the target storage location and write the updated self-learning compensation offset into the storage location management database; S7. With the target steel coil in contact with the ground and the lifting device not fully released, rotate the lifting device to center the target steel coil according to the target angle of the weld and the preset angle required by the packaging unit, and then fully release the lifting device to complete the unwinding.

2. The unmanned vehicle unwinding method as described in claim 1, characterized in that, In step S1, the self-learning compensation offset is expressed as: ; The original design coordinates are represented as follows ; The formula for calculating the effective target coordinates is: in, The effective target coordinates are in the first horizontal direction; The component of the original design coordinates in the first horizontal direction; The component of the self-learning compensation offset in the first horizontal direction; The effective target coordinates are in the second horizontal direction; The component of the original design coordinates in the second horizontal direction; The component of the self-learning compensation offset in the second horizontal direction; The first horizontal direction is perpendicular to the second horizontal direction.

3. The unmanned vehicle unloading method as described in claim 1, characterized in that, In step S3, applying an excitation pulse to the spreader specifically involves controlling the unmanned vehicle to perform the same actions in the first and second horizontal directions sequentially: Control the unmanned vehicle from a stationary state at a speed of 0.2~0.3 m / s 2 The acceleration increases to 0.5~0.7 m / s, and the constant velocity is maintained for 0.4~0.6 s, then decreases to (-0.2)~(-0.3) m / s. 2 The acceleration decelerates to 0; The first horizontal direction is perpendicular to the second horizontal direction.

4. The unmanned vehicle unloading method as described in claim 3, characterized in that, In step S3, the torsional angular velocity is the rotational angular velocity of the spreader about the vertical axis, which is determined by real-time measurement of the torsional angle of the spreader about the vertical axis. The rate of change over time was obtained; The swing angle includes a first swing angle of the spreader in the first horizontal direction and a second swing angle in the second horizontal direction. The first swing angle is obtained by measuring the angle of the spreader in a vertical plane perpendicular to the first horizontal direction in real time; the second swing angle is obtained by measuring the angle of the spreader in a vertical plane perpendicular to the second horizontal direction in real time.

5. The unmanned vehicle unloading method as described in claim 4, characterized in that, In step S3, the eccentric mass identification model uses a torsional dynamics equation to describe the torsional motion of the lifting device and the target steel coil. The torsional dynamics equation is as follows: in, The total moment of inertia of the lifting device and the target steel coiled around the vertical axis is obtained through no-load calibration. The damping coefficient; It is the acceleration due to gravity; It is the torsional angular velocity; This is the first swing angle; This is the second swing angle; It is the eccentricity; The eccentricity direction angle is defined as the angle between the eccentricity vector and the first horizontal direction.

6. The unmanned vehicle unwinding method as described in claim 5, characterized in that, The solution process for the eccentricity mass identification model includes the following steps: S301. At each sampling time Calculate the measured torsional torque : in, The sampling period; S302. Construct the observation matrix : S303. Set the vector of parameters to be estimated : S304. Iteratively update the parameter estimates according to the recursive formula, which is: in, Sampling time The gain matrix; It is the vector of parameters to be estimated. At sampling time The estimated value Sampling time The covariance matrix; The preset forgetting factor; S305. When the number of sampling points calculated in the iterative calculation reaches the preset target number of sampling points or the estimated value of the parameter vector to be estimated. After convergence, based on the current parameter vector to be estimated... Calculate the eccentricity and eccentricity direction angle: in, The vector of parameters to be estimated The first component, corresponding to ; The vector of parameters to be estimated The second component corresponds to ; The number of sampling points corresponds to the number of data points collected within the duration of the excitation pulse.

7. The unmanned vehicle unwinding method as described in claim 1, characterized in that, In step S4, the trajectory planning model employs model predictive control, and the state variables of this model predictive control include the position of the unmanned vehicle in the first horizontal direction. Position in the second horizontal direction The first swing angle of the lifting device in the first horizontal direction The second swing angle in the second horizontal direction The angle of torsion of the lifting device about the vertical axis ; The control variable is the acceleration of the autonomous vehicle in the first horizontal direction. Acceleration of autonomous vehicles in the second horizontal direction ; The objective function for model predictive control is: in, To predict the number of sampling points in the time domain; The effective target coordinates are the components in the first horizontal direction; The effective target coordinates are the components in the second horizontal direction.

8. The unmanned vehicle unwinding method as described in claim 7, characterized in that, Model predictive control uses the interior-point method to solve the optimization problem in each sampling period to obtain the optimal acceleration command. and After being integrated, the data is sent to the autonomous vehicle's drive system and analyzed in real time. , , The state variables are reinitialized and solved in each sampling period to form a closed-loop control.

9. The unmanned vehicle unloading method as described in claim 1, characterized in that, In step S6, the deviation between the contact position coordinates and the effective target coordinates includes a first horizontal deviation δx and a second horizontal deviation δy, calculated using the following formulas: in, The component of the contact position coordinates in the first horizontal direction; The component of the contact position coordinates in the second horizontal direction; The rule for updating the self-learning compensation offset of the target storage location is as follows: when or When the offset exceeds 5mm, update the self-learning compensation offset according to the following formula: in, The component of the updated self-learning compensation offset in the first horizontal direction; The updated self-learning compensation offset component in the second horizontal direction α is the preset learning rate.

10. The unmanned vehicle unwinding method as described in claim 1, characterized in that, In step S7, the rotating spreader rotates the target steel coil to achieve angular alignment. The calculation formula is: in, The angle is set according to the preset requirements of the packaging unit; The target angle for the weld.

Images