An automatic stacking method, device and system

Abstract

The present invention relates to the field of container stacking and is applied to rail-mounted container cranes, and provides an automatic stacking method, device and system, including a trolley, a trolley frame, a spreader and a controller. Target detection information is obtained through a 3D laser scanner, and the controller processes the target detection information to obtain the target detection displacement of the trolley frame and the target detection displacement of the spreader. Compensation detection information is obtained through an inclination sensor, and the compensation detection information includes the compensated trolley frame inclination angle and the compensated spreader inclination angle. The controller processes the target detection information and the compensation detection information to obtain the compensated displacement of the trolley frame and the compensated displacement of the spreader. The actual displacement of the trolley frame is obtained by summing the target detection displacement of the trolley frame and the compensated displacement of the trolley frame, and the actual displacement of the spreader is obtained by summing the target detection displacement of the spreader and the compensated displacement of the spreader. The actual displacement of the trolley frame and the actual displacement of the spreader are automatically executed.

Description

Technical Field

[0001] The present invention relates to the field of container stacking, and more specifically, it is applied to a rail-mounted container crane, and relates to an automatic stacking method, device and system. Background Art

[0002] With the increase in the operation time of the crane and the use time of the yard, the operating environment of the equipment changes rapidly, and the accuracy of the automatic container stacking system in the yard will gradually decrease. There is a huge difference between the actual operation effect and the expected value, and the situation of uneven automatic container stacking often occurs. The main reasons are the change of the inclination angle of the trolley track and the change of the height difference between the two tracks on both sides of the gantry crane, which cause the trolley frame of the crane to tilt, and the laser on the trolley frame generates errors during scanning; after the yard subsides, the inclination angle of the containers in the yard changes, which causes errors during the automatic stacking of the yard containers; the change of the inclination angle of the spreader causes the inclination angle of the container held by the spreader to change, which causes errors during the automatic stacking of the yard containers. Summary of the Invention

[0003] The present invention aims to overcome at least one defect of the above-mentioned prior art, and provides an automatic stacking method, device and system for solving the problem of uneven automatic stacking of containers.

[0004] The present invention adopts the following technical solutions:

[0005] In the first aspect, an automatic stacking method is provided, which is applied to a container crane, including a gantry crane, a trolley frame, a spreader, and a controller. The method is characterized in that the method includes:

[0006] Obtain target detection information through a 3D laser scanner, where the target detection information includes the height of the target container, the height of the container on the spreader, and the inclination angle of the target detection spreader; the controller processes the target detection information to obtain the target detection displacement of the trolley frame and the target detection displacement of the spreader;

[0007] Obtain compensation detection information through an inclination sensor, where the compensation detection information includes the inclination angle of the compensation trolley frame and the inclination angle of the compensation spreader, and the controller processes the target detection information and the compensation detection information to obtain the compensation displacement of the trolley frame and the compensation displacement of the spreader;

[0008] Sum the target detection displacement of the trolley frame and the compensation displacement of the trolley frame to obtain the actual displacement of the trolley frame, and sum the target detection displacement of the spreader and the compensation displacement of the spreader to obtain the actual displacement of the spreader;

[0009] Automatically execute the actual displacement of the trolley frame and the actual displacement of the spreader.

[0010] Further, the compensated displacement of the trolley frame is obtained by the controller's operation based on the target container height, the container height on the spreader, and the compensated trolley frame inclination angles, and the compensated trolley frame inclination angles include the front inclination angle and the rear inclination angle;

[0011] The compensated displacement of the spreader is obtained by verifying the compensated spreader inclination angles and the target detected spreader inclination angles, and the difference value obtained from the verification is the compensated displacement of the spreader. The compensated spreader inclination angles and the target detected spreader inclination angles include the front inclination angle, the rear inclination angle, the left inclination angle, and the right inclination angle.

[0012] Further, the method further includes determining the stable acquisition time. The stable acquisition time is obtained by collecting at least twice at different times and calculating based on the fluctuation ratio of the information collected continuously twice. The objects of the stable acquisition time include at least one of the target container height, the container height on the spreader, the target detected spreader inclination angle, the compensated trolley frame inclination angle, and the compensated spreader inclination angle. The actual displacement of the trolley frame and the actual displacement of the spreader are both based on the same stable acquisition time. During the process of stacking containers, after the container crane grabs a container and moves it to the accurate position, the inertia generated will cause the container to shake to a certain extent, resulting in errors in automatic stacking measurement. The static degree of the container is judged by the fluctuation ratio of the objects of the stable acquisition time during continuous measurement.

[0013] Further, the fluctuation ratio is from -5% to 5%.

[0014] Further, the method further includes a monitoring and warning method, and the method includes:

[0015] Set the normal inclination angle range of the trolley frame and set the normal inclination angle range of the spreader;

[0016] Obtain the compensated trolley frame inclination angle and the compensated spreader inclination angle at the stable acquisition time;

[0017] Judge whether the compensated trolley frame inclination angle at the stable acquisition time meets the normal inclination angle range of the trolley frame, and judge whether the compensated spreader inclination angle at the stable acquisition time meets the normal inclination angle range of the spreader;

[0018] Prompt and record the compensated trolley frame inclination angle and the compensated spreader inclination angle that do not meet the normal inclination angle range, and cut off the automatic operation command of the machine.

[0019] Further, the normal inclination angle range of the trolley frame is -0.8 to 0.8 degrees;

[0020] The normal inclination angle range of the spreader is -0.8 to 0.8 degrees.

[0021] Further, in the method, the compensated displacement of the trolley frame is obtained by a compensated displacement machine learning model, including:

[0022] Input the characteristics of the trolley frame, where the characteristics of the trolley frame include the height of the target container, the height of the container on the spreader, and the inclination angle of the compensating trolley frame;

[0023] The compensating displacement machine learning model processes the characteristics of the trolley frame. The compensating displacement machine learning model is established by extracting historical automatic stacking data, and the historical data includes stacking effect labels, the inclination angle of the trolley frame, the position of the gantry crane, the position of the trolley frame, the lifting height, the encoder value of the spreader push rod motor, and the height of the target container, the height of the container on the spreader, and the inclination angle of the target detection spreader in the target detection information;

[0024] Output the predicted value of the compensating displacement of the trolley frame.

[0025] Furthermore, the decision function of the compensating displacement machine learning model is Y, and Y = sin(x3) * (x2 - x1);

[0026] where x1 is the height of the target container, x2 is the height of the container on the spreader, x3 is the inclination angle of the compensating trolley frame, and the range of Y is from -25 cm to 25 cm.

[0027] In a second aspect, the present invention provides an automatic stacking device, which includes a gantry crane, a trolley frame, a spreader, a 3D laser scanner, an inclination sensor, an actuator, and a controller;

[0028] The 3D laser scanner is installed below the platform where the trolley frame is located. The 3D laser scanner is connected to the controller, and the controller processes the 3D laser scanner;

[0029] The inclination sensors are respectively installed on the trolley frame and the spreader. The inclination sensors are connected to the controller by high-quality shielded cables, and the controller processes the inclination sensors;

[0030] The actuator includes a trolley frame mechanism unit and a fine movement unit of the spreader upper frame, and is used to execute the actual displacement of the trolley frame and the actual displacement of the spreader. The actuator is connected to the controller and receives the controller's instructions;

[0031] The controller is a programmable logic controller, which is used to analyze and process the data collected by the 3D laser scanner and the inclination sensors.

[0032] In a third aspect, the present invention provides an automatic stacking system, which includes a data perception layer, a data processing layer, a core control layer, and a mechanism execution layer;

[0033] The data perception layer includes a 3D laser scanner and an inclination sensor. The 3D laser scanner collects the target detection information of the trolley frame and the spreader in real time, and the inclination sensor collects the inclination information of the trolley frame and the spreader in real time;

[0034] The data processing layer includes a target detection unit and an automatic stacking machine learning model unit. The target detection unit is used to process the information collected by the 3D laser scanner; the automatic stacking machine learning model unit includes a database, feature extraction, and a machine learning model, and is used to establish an automatic stacking machine learning model;

[0035] The core control layer includes a programmable logic controller, which is used to process the information of the data processing layer;

[0036] The mechanism execution layer includes a trolley frame mechanism unit and a spreader upper frame fine movement unit. The execution includes that the trolley frame mechanism unit moves the trolley frame forward and backward, and the spreader upper frame fine movement unit rotates the spreader left and right.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] 1. Aiming at the problem of frequent misalignment of stacked containers in the yard caused by the changes in the inclination angle of the trolley track of the rail-mounted container gantry crane, the height difference between the two sides of the gantry track, and the inclination angle of the spreader, a complete set of error compensation devices and methods is constructed. The inclination sensors are used to collect the inclination information of the trolley frame and the spreader, and the compensation parameters are calculated in the algorithm program of the programmable logic controller, and then the electric control system controls the rotation of the spreader and the movement of the trolley to complete the control compensation, realizing the stable operation and safe production of the port rail-mounted container gantry full-automatic crane.

[0039] 2. The inclination sensors are used to detect the inclination angles of the trolley frame and the spreader of the crane in real time to obtain the attitude information of the trolley frame and the spreader. Through logical judgment in the algorithm program of the programmable logic controller, when the spreader or the trolley frame is too inclined, an abnormal working condition alarm is given, and the equipment operation is stopped, improving the operation safety of the equipment.

[0040] 3. The database is used to organize, store, and manage the crane operation status data, unify the management of more and larger data, and has a more efficient data query function. During the automatic stacking process of the rail-mounted container gantry crane, a large amount of data will be generated, and more accurate data extraction, better data organization, more convenient data maintenance, more rigorous data control, and more effective data utilization are required. Using the database management system, a large number of crane operation data can be stored and managed, and the operation status data during the automatic stacking of the crane can be better stored and managed. Description of the Drawings

[0041] Figure 1 It is a schematic diagram of an automatic stacking method in the present invention;

[0042] Figure 2 It is a schematic diagram of a monitoring and warning method in the present invention;

[0043] Figure 3Schematic diagram of sample data in the present invention;

[0044] Figure 4 Structural diagram of an automatic stacking device in the present invention;

[0045] Figure 5 Schematic diagram of an automatic stacking system in the present invention;

[0046] Reference numerals in the drawings of the present invention are: 1: trolley, 2: trolley frame, 3: spreader, 4: 3D laser scanner, 5: inclination sensor, 6: trolley frame mechanism unit, 7: upper spreader fine movement unit, 8: controller, 9: container. Detailed implementation manners

[0047] The drawings of the present invention are only for illustrative purposes and should not be construed as limitations on the present invention. To better illustrate the following embodiments, some components in the drawings may be omitted, enlarged or reduced, which do not represent the dimensions of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0048] In some embodiments, such as Figure 1 An automatic stacking method is provided, which is applied to a rail-mounted container crane and includes a trolley 1, a trolley frame 2, a spreader 3, and a controller 8. The method includes the trolley 1 entering the working container area and the spreader 3 grasping the container 9;

[0049] Target detection information is obtained through the 3D laser scanner 4. The target detection information includes the height of the target container, the height of the container on the spreader, and the inclination angle of the target detection spreader. The controller 8 processes the target detection information to obtain the target detection displacement of the trolley frame 2 and the target detection displacement of the spreader 3, where the controller 8 is a programmable logic controller (abbreviation: PLC).

[0050] When the Target Detection System (TDS) performs the automatic container placement task in the yard, it uses 3D laser technology and cooperates with a rotating motor to accurately scan the container 9 grasped by the spreader 3 and the yard environment below both sides of the spreader 3 to achieve the purpose of target positioning, thereby coordinating the control of the crane mechanism actions and realizing the automatic container alignment of the containers 9 in the yard. The 3D laser scanner 4 is installed below the platform of the crane trolley frame 2. The installation environment is not constant. As the operating time of the crane and the usage time of the yard increase, the operating environment of the equipment will change drastically. The change of the trolley track inclination angle, the settlement of the yard, and the change of the height difference between the two tracks on both sides of the trolley 1 will cause the coordinates established by the target detection system to be distorted, inaccurate target detection and positioning, resulting in a decrease in the accuracy of the automatic stacking of the containers 9 in the yard and frequent occurrences of uneven automatic container stacking.

[0051] Therefore, it is necessary to eliminate and reduce the impact of the sharp change in the equipment operating environment on the automatic stacking system of yard containers 9, and perform automatic stacking error compensation caused by the operating environment on the basis of the target detection system, assist the target detection system to automatically align the containers, improve the automatic stacking accuracy of the rail-mounted container gantry crane 9, and provide an automatic stacking method in the present invention. On the basis of the target detection system, the compensation displacement is increased for mutual verification and complementation. The inclination sensors 5 installed on the trolley frame 2 and the inclination sensors 5 on the spreader 3 collect the inclination of the trolley frame 2 and the inclination of the spreader 3 in real time, and transmit them to the controller 8 through the fieldbus communication technology for data processing using the algorithm program.

[0052] The controller 8 calculates the compensation displacement required for the trolley and the compensation displacement required for the spreader 3 according to the angle data of the inclination sensor 5, while the target detection system calculates the displacement required for the trolley and the displacement required for the spreader 3. After combination, the actual displacement required for the trolley and the actual displacement required for the spreader 3 are obtained, and then through the controller 8, the moving distance of the trolley and the moving distance of the spreader 3 are controlled to achieve a more accurate automatic container alignment function.

[0053] The actual displacement of the trolley frame 2 is obtained by summing the target detection displacement of the trolley frame 2 and the compensation displacement of the trolley frame, and the actual displacement of the spreader 3 is obtained by summing the target detection displacement of the spreader 3 and the compensation displacement of the spreader. During automatic stacking, the target detection system outputs the displacement A required for the trolley and the displacement B required for the spreader 3 to the controller 8. The controller 8 calculates the compensation displacement C required for the trolley frame mechanism unit 6 and the compensation displacement D required for the upper spreader micro-motion unit 7 according to the measured angle data of the inclination sensor 5. Finally, the actual displacement required for the trolley (A + C) and the actual displacement required for the spreader 3 (B + D) are obtained, and then through the electric control system, the rotation of the spreader 3 and the movement of the trolley are controlled.

[0054] The compensation detection information is obtained through the inclination sensor 5, and the compensation detection information includes the compensated trolley frame inclination and the compensated spreader inclination. The controller 8 processes the target detection information and the compensation detection information to obtain the trolley frame compensation displacement and the spreader compensation displacement. Automatically execute the actual displacement of the trolley frame 2 and the actual displacement of the spreader 3.

[0055] The trolley frame mechanism unit 6 and the upper spreader micro-motion unit 7 reach the target position according to the displacement compensation data result through process control and coordinated drive of the trolley frame mechanism unit 6 and the upper spreader micro-motion unit 7.

[0056] Using two inclination sensors 5, the inclination angles of the trolley frame 2 and the spreader 3 are detected in real time, and the data is then transmitted to the controller 8 through fieldbus communication technology. Compensation parameters are calculated in the program of the controller 8 to control the left and right rotation of the spreader 3 and the forward and backward movement of the trolley, so as to achieve automatic stacking error compensation and improve the automatic stacking accuracy of the rail-mounted container gantry crane, effectively improving the safety, efficiency and reliability of fully automatic container handling.

[0057] The compensated displacement of the trolley frame is obtained through calculation by the controller 8 based on the target container height, the container height on the spreader, and the compensated inclination angle of the trolley frame. The compensated inclination angle of the trolley frame includes the front inclination angle and the rear inclination angle; the compensated displacement of the spreader is obtained by verifying the compensated inclination angle of the spreader and the target detected inclination angle of the spreader, and the difference value obtained by the verification is the compensated displacement of the spreader. The compensated inclination angle of the spreader and the target detected inclination angle of the spreader include the front inclination angle, the rear inclination angle, the left inclination angle, and the right inclination angle.

[0058] The process of the controller 8 calculating the compensated displacement required for the trolley based on the angle data of the inclination sensor 5 of the trolley frame 2: The compensated value of the target position of the trolley is calculated through the target container height scanned by the target detection system, the container height on the spreader, and the front and rear inclination data of the inclination sensor 5 of the trolley frame 2.

[0059] The process of the controller 8 calculating the compensated value based on the angle data of the inclination sensor 5 of the spreader 3: The compensated value of the inclination angle of the spreader 3 is calculated through mathematical operations based on the front and rear inclination data of the inclination sensor 5 of the spreader 3, and then transmitted to the target detection system through Ethernet to participate in the data verification and positioning calculation of the target detection. In the target detection system, the front and rear inclination and left and right inclination angles of the spreader 3 detected by the inclination sensor 5 are double-verified with the front and rear inclination and left and right inclination angles of the spreader 3 scanned by the target detection, improving the accuracy of the inclination angle data of the spreader 3.

[0060] The method further includes determining the stable acquisition time. The stable acquisition time is obtained through at least two acquisitions at different times, and the stable acquisition time is obtained according to the fluctuation ratio of the information collected continuously twice. The objects of the stable acquisition time include at least one of the target container height, the container height on the spreader, the target detected inclination angle of the spreader, the compensated inclination angle of the trolley frame, and the compensated inclination angle of the spreader. The actual displacements of the trolley frame 2 and the spreader 3 are both based on the same stable acquisition time.

[0061] During the stacking of containers 9, after the container crane grabs the container 9 and moves the container 9 to the accurate position, the inertia generated will cause the container 9 to shake to a certain extent. If sampling is carried out when the container 9 is not static, it will cause errors in the automatic stacking measurement, thus affecting the overall stacking effect. The static degree of the container 9 is judged by the fluctuation ratio of the object with a stable acquisition time during continuous measurement. The fluctuation ratio is set to -5% to 5%, indicating that the fluctuation ratio of the object with a stable acquisition time for two consecutive times is small and can be regarded as static. Among them, the fluctuation ratio is the ratio of the difference between the latter measurement and the former measurement to the former measurement. The object with a stable acquisition time can be one kind or a combination of multiple kinds.

[0062] In some embodiments, such as Figure 2 provides a monitoring and warning method, and the method includes:

[0063] Set the normal inclination angle range of the trolley frame 2 and set the normal inclination angle range of the spreader 3;

[0064] Obtain the compensated inclination angle of the trolley frame and the compensated inclination angle of the spreader at the stable acquisition time;

[0065] Judge whether the compensated inclination angle of the trolley frame at the stable acquisition time meets the normal inclination angle range of the trolley frame 2, and judge whether the compensated inclination angle of the spreader at the stable acquisition time meets the normal inclination angle range of the spreader 3;

[0066] Prompt and record the compensated inclination angle of the trolley frame and the compensated inclination angle of the spreader that do not meet the normal inclination angle range, and cut off the automatic operation command of the machine.

[0067] First, record the inclination angles of the trolley frame 2 and the spreader 3 in the normal state, and then, during the operation process, real-time monitor the attitude data of the trolley frame 2 and the spreader 3 and compare them with the normal data. When the data deviates too much from the normal value, an abnormal state safety alarm is given, and the equipment action is stopped to ensure the safety of the automatic equipment operation.

[0068] The controller 8 calculates the inclination angle of the spreader 3 relative to the trolley frame 2 according to the angle data of the inclination angle sensor 5 of the trolley frame 2 and the inclination angle sensor 5 of the spreader 3, and realizes the process of alarming when the inclination angle of the spreader 3 is too large.

[0069] The present invention will give an alarm under abnormal working conditions when the spreader 3 or the trolley frame 2 is too inclined, improving the safety of equipment operation. Real-time detect the inclination angles of the crane trolley frame 2 and the spreader 3, obtain the attitude information of the trolley frame 2 and the spreader 3, and give an alarm in the abnormal state during the automatic control process. The safety of the full-automatic loading and unloading of the rail-mounted container gantry crane is improved.

[0070] In some embodiments, the normal inclination angle range of the trolley frame 2 is -0.8 to 0.8 degrees, and the normal inclination angle range of the spreader 3 is -0.8 to 0.8 degrees.

[0071] In some embodiments, a method for obtaining the compensated displacement of the trolley frame through a compensated displacement machine learning model is provided, including:

[0072] Input the characteristics of the trolley frame 2, where the characteristics of the trolley frame 2 include the target container height, the container height on the spreader, and the compensated trolley frame inclination angle;

[0073] The compensated displacement machine learning model processes the characteristics of the trolley frame 2. The compensated displacement machine learning model is established by extracting historical automatic stacking data, and the historical data includes stacking effect labels, the inclination angle of the trolley frame 2, the position of the gantry 1, the position of the trolley frame 2, the lifting height, the encoder value of the spreader push rod motor, and the target container height, the container height on the spreader, and the target detection spreader inclination angle in the target detection information;

[0074] Output the predicted value of the compensated displacement of the trolley frame.

[0075] It is found through a large amount of on-site test data that there is a correlation between the neatness of the automatic container stacking of the crane in the yard container 9 through the target detection system and the inclination angles of the trolley frame 2 and the spreader 3. By continuously collecting the state information of the equipment in the automatic control process, the current actual working state of the equipment is determined, the automatic control process is optimized, an adaptive control law is generated, so as to adjust the automatic control parameters in real time, keep the automatic stacking system of the yard container 9 in the optimal operating state all the time, and correct the rotation amount of the spreader 3 and the movement amount of the trolley in the automatic control process to adapt to the changes in the installation environment.

[0076] After each successful automatic stacking and container boxing, the automatic stacking historical database obtains the state information of the equipment such as the angle of the inclination sensor 5, the position of the gantry 1, the trolley position, the lifting height, the encoder value of the spreader push rod motor, and the data of the target detection system from the controller 8, and stores it. At the same time, the data is sent to the data platform for feature extraction, and a successful automatic stacking machine learning model is established. According to the state information of the equipment during a large number of successful automatic container boxings in the database, the correlation between the neatness of the crane's automatic container boxing and the inclination angles of the trolley frame 2 and the spreader 3 is determined, and then the algorithm program of the controller 8 is carried out to calculate the compensation data through the angle value of the inclination sensor 5, which is convenient for subsequent data analysis and algorithm optimization.

[0077] After each successful automatic stacking of the container 9, the automatic stacking historical database obtains the device status information such as the angle of the inclination sensor 5, the position of the trolley 1, the position of the carriage, the lifting height, the encoder value of the spreader push rod motor, and the data of the target detection system from the controller 8, and stores it. At the same time, the data is sent to the data platform for feature extraction, and a machine learning model for successful automatic stacking is established. Theoretically, when the same container 9 is automatically stacked repeatedly at the same position, the rotation amount of the spreader 3 and the movement amount of the carriage are similar during the automatic control process. Therefore, the data platform uses the model with similar device status data such as the lifting position, the position of the carriage, the data of the inclination sensor 5, and the weight of the spreader 3 when the automatic container stacking is successful as the same feature model. When a new automatic container stacking task is executed, the database collects the device status information from the controller 8 again, sends it to the data platform, and compares it with the successful automatic stacking model with the same features for data analysis to analyze the parameter feature differences of the automatic stacking.

[0078] According to the device status information of a large number of successful automatic container stackings in the database, the correlation between the neatness of the crane's automatic container stacking and the inclination of the carriage 2 and the inclination of the spreader 3 is determined. Then, through the algorithm program design of the controller 8, the displacement compensation amount is calculated from the value of the inclination sensor 5.

[0079] The machine learning model is used to realize the mapping from the sample x to the sample Y, that is, f(x) → Y. In this automatic stacking machine learning model, the independent variables include the height x1 of the target container scanned by the target detection, the height x2 of the container on the spreader, and the front and rear inclination data x3 of the inclination sensor 5 of the carriage 2. The dependent variable is the compensation value Y of the target position of the carriage, that is, Y = f(x1, x2, x3).

[0080] When the automatic stacking machine learning model knows x1, x2, x3 in multiple samples of successful automatic stackings and their corresponding compensation values Y of the target position of the carriage, it uses the assumed known function form Y = f(x1, x2, x3) to fit the objectively existing mapping function as much as possible and ensure that it has as similar fitting ability on unknown samples. The establishment process of the automatic stacking machine learning model is actually to learn the mapping from input to output through known successful automatic stacking samples through learning strategies and optimization algorithms. Calculate the corresponding relationship between the known data (the height of the target container scanned by the target detection, the height of the container on the spreader, the front and rear inclination data of the inclination sensor 5 of the carriage 2) and the unknown data (the compensation value of the target position of the carriage).

[0081] By learning the experience of the "successful automatic stacking" feature, after knowing x1, x2, x3 of the unknown template, the compensation value Y of the corresponding target position of the carriage is calculated.

[0082] Successfully automatically stacking data is the basic raw material for an automatic stacking machine learning system. The data sets of each successful automatic stacking sample, such as Figure 3 The schematic diagram of the sample data in Figure 3 is a partial data set of successful automatic stacking samples recorded in the database, including features and results. Among them, the features are the height x1 of the target container scanned by TDS, the height x2 of the container on the spreader, and the front and rear tilt data x3 of the two-angle sensors on the trolley frame 2. The result obtained is the compensation value Y of the trolley target position.

[0083] Learning a "good" model is the direct goal of machine learning. The automatic stacking machine learning model is a function that learns the internal laws of data features. The machine learning model first selects a certain model method, and then learns from the data samples (x1, x2, x3), optimizes the model parameters w to adjust the effective expression of each feature, and finally obtains the corresponding decision function f(x1, x2, x3; w). This function maps the input variables x1, x2, x3 under the action of the parameter w to the output prediction Y, that is, Y =

[0084] f(x1, x2, x3; w). The device currently obtains the corresponding decision function as Y = sin(x3)*(x2 - x1), Y ∈ [-25, 25].

[0085] The learning goal of the automatic stacking machine learning model is that the error between the predicted value and the actual value is as low as possible, that is, the predicted compensation value of the trolley target position can make the automatic stacking effect better and make the containers 9 stacked more neatly after automatic stacking.

[0086] In some embodiments, such as Figure 4 There is provided an automatic stacking device, which includes a gantry 1, a trolley frame 2, a spreader 3, a 3D laser scanner 4, an inclination sensor 5, a trolley frame mechanism unit 6, a fine movement unit 7 on the spreader upper frame, and a controller 8; the 3D laser scanner 4 is installed below the platform where the trolley frame 2 is located, the 3D laser scanner 4 is connected to the controller 8, and the controller 8 processes the 3D laser scanner 4;

[0087] The inclination sensors 5 are respectively installed on the trolley frame 2 and the spreader 3. The inclination sensors 5 are connected to the controller 8 by high-quality shielded cables, and the controller 8 processes the inclination sensors 5;

[0088] This device has a total of two inclination sensors 5. The surface of the steel plate where they are installed is flat, with little vibration, firm and reliable, and high-quality shielded cables are used to connect the sensors to the controller 8 for real-time acquisition, analysis, processing, and calculation of the inclination sensors 5.

[0089] The inclination sensor 5 with a robust design can reliably monitor the inclination in a harsh application environment and accurately complete the inclination measurement. A high-quality Profibus-DP cable is used, with double shielding of aluminum foil and bare wire braiding, which is especially suitable for installation in an industrial environment vulnerable to electromagnetic interference. The continuity of grounding can be achieved through the outer layer of the grounding contact of the bus terminal, and it has a more reliable anti-interference function.

[0090] The actuator includes a carriage mechanism unit 6 on the carriage 2 and a fine movement unit 7 on the upper carriage of the spreader 3, which are used to execute the actual displacement of the carriage 2 and the actual displacement of the spreader 3. The actuator is connected to the controller 8 and receives instructions from the controller 8.

[0091] The controller 8 is used to analyze and process the data collected by the 3D laser scanner 4 and the inclination sensor 5.

[0092] In some embodiments, such as Figure 5 An automatic stacking system is provided, which includes a data perception layer, a data processing layer, a core control layer, and a mechanism execution layer.

[0093] The data perception layer includes a 3D laser scanner 4 and an inclination sensor 5. The 3D laser scanner 4 collects the target detection information of the carriage 2 and the spreader 3 in real time, and the inclination sensor 5 collects the inclination information of the carriage 2 and the spreader 3 in real time. The data processing layer includes a target detection unit and an automatic stacking machine learning model unit. The target detection unit is used to process the information collected by the 3D laser scanner 4. The automatic stacking machine learning model unit includes a database, feature extraction, and a machine learning model, which are used to establish an automatic stacking machine learning model.

[0094] Using a database management system, a large amount of crane operation status data can be stored and managed, which can better store and manage the operation status data during the automatic stacking of the crane and has a more efficient data query function. It is beneficial to the data analysis of the equipment operation status, data tracking, historical fault troubleshooting, historical fault recording, establishing an automatic stacking model, and unified management of data.

[0095] Using a database management system, the key status data of automatic stacking is stored, analyzed, and managed efficiently and organized. The data center preprocesses the real-time sampled data, extracts effective feature data, stores the feature data of successful automatic stacking in the automatic stacking historical database, establishes an automatic stacking machine learning model, and then trains the automatic stacking machine learning model with the automatic stacking historical database to deduce the accurate compensation data difference law.

[0096] The core control layer includes a controller 8, which is used to process the information of the data processing layer.

[0097] Two inclination sensors 5 are used to detect the inclination angles of the car body frame 2 and the spreader 3 in real time. Then, the data is transmitted to the controller 8 of the core control layer through fieldbus communication technology. The program algorithm of the controller 8 analyzes, processes, and calculates the angle data in real time. During automatic container alignment, the rotation amount of the spreader 3 and the displacement of the car body are corrected in the automatic control process to achieve automatic stacking error compensation, adapt to changes in the installation environment, and improve the automatic stacking accuracy of the rail-mounted container gantry crane.

[0098] The mechanism execution layer includes the car body frame mechanism unit 6 on the car body frame 2 and the fine movement unit 7 of the spreader upper frame. The execution includes moving the car body frame 2 forward and backward by executing the car body frame mechanism unit 6 and rotating the spreader 3 left and right by executing the fine movement unit 7 of the spreader upper frame on the spreader 3.

[0099] Obviously, the above-mentioned embodiments of the present invention are merely examples for clearly illustrating the technical solutions of the present invention, rather than limitations on the specific implementation manners of the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the claims of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. An automatic stacking method applied to an orbital container crane, comprising a trolley, a trolley frame, a spreader, and a controller, characterized in that the method comprises: obtaining target detection information through a 3D laser scanner, where the target detection information includes the height of the target container, the height of the container on the spreader, and the inclination angle of the target detection spreader; the controller processes the target detection information to obtain the target detection displacement of the trolley frame and the target detection displacement of the spreader; obtaining compensation detection information through an inclination sensor, where the compensation detection information includes the compensated trolley frame inclination angle and the compensated spreader inclination angle, and the controller processes the target detection information and the compensation detection information to obtain the compensated displacement of the trolley frame and the compensated displacement of the spreader; summing the target detection displacement of the trolley frame and the compensated displacement of the trolley frame to obtain the actual displacement of the trolley frame, and summing the target detection displacement of the spreader and the compensated displacement of the spreader to obtain the actual displacement of the spreader; automatically executing the actual displacement of the trolley frame and the actual displacement of the spreader.

2. The automatic stacking method according to claim 1, characterized in that the compensated displacement specifically includes: the compensated displacement of the trolley frame is obtained by the controller's operation based on the height of the target container, the height of the container on the spreader, and the compensated trolley frame inclination angle, and the compensated trolley frame inclination angle includes the front inclination angle and the rear inclination angle; the compensated displacement of the spreader is obtained by verifying the compensated spreader inclination angle and the target detection spreader inclination angle, and the difference value obtained from the verification is the compensated displacement of the spreader, and the compensated spreader inclination angle and the target detection spreader inclination angle include the front inclination angle, the rear inclination angle, the left inclination angle, and the right inclination angle.

3. The automatic stacking method according to claim 1, characterized in that the method further includes determining the stable acquisition time, which is obtained by collecting at least twice at different times and calculating the stable acquisition time according to the fluctuation ratio of the information collected continuously twice. The objects of the stable acquisition time include at least one of the height of the target container, the height of the container on the spreader, the inclination angle of the target detection spreader, the compensated trolley frame inclination angle, and the compensated spreader inclination angle. The actual displacement of the trolley frame and the actual displacement of the spreader are both based on the same stable acquisition time.

4. The automatic stacking method according to claim 3, characterized in that the fluctuation ratio is from -5% to 5%.

5. The automatic stacking method according to claim 3, characterized in that the method further includes a monitoring and warning method, which includes: setting the normal inclination angle range of the trolley frame and setting the normal inclination angle range of the spreader; obtaining the compensated trolley frame inclination angle and the compensated spreader inclination angle at the stable acquisition time; judging whether the compensated trolley frame inclination angle at the stable acquisition time meets the normal inclination angle range of the trolley frame, and judging whether the compensated spreader inclination angle at the stable acquisition time meets the normal inclination angle range of the spreader; prompting and recording the compensated trolley frame inclination angle and the compensated spreader inclination angle that do not meet the normal inclination angle range, and cutting off the automatic operation command of the machine.

6. The automatic stacking method according to claim 5, characterized in that the normal inclination angle range of the trolley frame is from -0.8 to 0.8 degrees; the normal inclination angle range of the spreader is from -0.8 to 0.8 degrees.

7. An automatic stacking method according to claim 2, characterized in that, in the method, the compensated displacement of the trolley frame is obtained through a compensated displacement machine learning model, including: inputting the trolley frame features, where the trolley frame features include the height of the target container, the height of the container on the spreader, and the compensated trolley frame inclination angle; the compensated displacement machine learning model processes the trolley frame features, and the compensated displacement machine learning model is established by extracting historical data of automatic stacking. The historical data includes stacking effect labels, trolley frame inclination angles, gantry positions, trolley frame positions, lifting heights, encoder values of the spreader push rod motor, and the height of the target container, the height of the container on the spreader, and the inclination angle of the target detection spreader in the target detection information; outputting the predicted value of the compensated displacement of the trolley frame.

8. An automatic stacking method according to claim 7, characterized in that, the decision function of the compensated displacement machine learning model is Y, Y = sin(x3)*(x2 - x1); where x1 is the height of the target container, x2 is the height of the container on the spreader, x3 is the compensated trolley frame inclination angle, and the range of Y is from -25 cm to 25 cm.

9. An automatic stacking device for implementing an automatic stacking method according to any one of claims 1-8, including a gantry, a trolley frame, a spreader, a 3D laser scanner, an inclination sensor, and a controller, characterized in that, the device further includes an actuator; the 3D laser scanner is installed below the platform where the trolley frame is located, the 3D laser scanner is connected to the controller, and the controller processes the 3D laser scanner; the inclination sensors are respectively installed on the trolley frame and the spreader, the inclination sensors are connected to the controller by high-quality shielded cables, and the controller processes the inclination sensors; the actuator includes a trolley frame mechanism unit and a spreader upper frame fine movement unit for executing the actual displacement of the trolley frame and the actual displacement of the spreader. The actuator is connected to the controller and receives the controller's instructions; the controller is a programmable logic controller for analyzing and processing the data collected by the 3D laser scanner and the inclination sensor.

10. An automatic stacking system for implementing an automatic stacking method according to any one of claims 1-8, characterized in that, the system includes a data perception layer, a data processing layer, a core control layer, and an agency execution layer; the data perception layer includes a 3D laser scanner and an inclination sensor. The 3D laser scanner real-time collects the target detection information of the trolley frame and the spreader, and the inclination sensor real-time collects the inclination information of the trolley frame and the spreader; the data processing layer includes a target detection unit and an automatic stacking machine learning model unit. The target detection unit is used to process the information collected by the 3D laser scanner; the automatic stacking machine learning model unit includes a database, feature extraction, and a machine learning model for establishing an automatic stacking machine learning model; the core control layer includes a programmable logic controller for processing the information of the data processing layer; The execution layer of the mechanism includes a trolley frame mechanism unit and a fine movement unit for the upper spreader. The execution includes that the trolley frame mechanism unit moves the trolley frame forward and backward, and the fine movement unit for the upper spreader rotates the spreader left and right.

Images