Bulk cargo stacking method based on inclined gradient warehouse stacking method

Abstract

The invention relates to the technical field of cargo stacking, in particular to a bulk cargo stacking method based on an inclined gradient cabin stacking method, according to the stacking method, through a specific inclined gradient cabin stacking mode, a trapezoidal structure with the wide bottom and the narrow top is formed, the center of gravity of cargoes is effectively lowered, bag collapse caused by center-of-gravity shift is prevented, and potential safety hazards such as cargo falling are reduced. And life health of workers is guaranteed. In the actual bulk cargo loading and unloading operation, a traditional stacking mode is prone to causing accidents due to unstable cargo stacking, however, the stacking structure is optimized, the operation safety is greatly improved, the cargo stacking process is simplified, the shipping time is shortened, and the whole operation process is smoother and more efficient.

Description

Technical Field

[0001] The invention relates to the technical field of cargo stacking, and in particular to a bulk cargo stacking method based on an inclined gradient stacking method. Background Art

[0002] Traditional stacking methods lack scientific control over the center of gravity of cargo, and are mostly simple stacking without considering the impact of the stacking structure on the center of gravity. This makes it very easy for cargo to collapse due to the shift of the center of gravity during loading, unloading and transportation. The falling of cargo will not only cause the loss of the cargo itself, but also pose a serious threat to the life safety of on-site workers. At the bulk cargo loading and unloading sites of some ports, traditional stacking is not fully adjusted in combination with the shape of the cargo hold and the characteristics of bulk cargo, and there is a lack of effective support and restraint between the cargoes. During the ship transportation process, due to the shaking and bumping of the ship and the collision between the cargoes, the integrity of the cargo is difficult to guarantee.

[0003] With the continuous growth of bulk cargo transportation volume, the requirements for operational safety, cargo integrity and operational efficiency are increasing, and the limitations of traditional stacking methods are becoming more and more prominent. In this context, the bulk cargo stacking method based on the inclined gradient stacking method came into being, aiming to solve many problems existing in the traditional stacking method and promote the development of the bulk cargo transportation industry in a safer and more efficient direction. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention solves the technical problems by adopting a technical solution: a bulk cargo stacking method based on an inclined gradient stacking method, comprising the following steps: Step S1: Before bulk cargo is stacked, use 3D visual recognition equipment to perform a full-scale scan of the cargo hold's interior space to obtain a three-dimensional model of the cargo hold, including information such as the shape and size of the bulkhead, and the location and shape of obstacles in the hold. At the same time, scan the bulk cargo to be stacked to identify the shape, outline, and surface features of each batch of bulk cargo; Step S2: The weight of each batch of bulk cargo is measured in real time by a weight sensor connected to the loading and unloading equipment, and the volume of the bulk cargo is calculated by using a volume calculation algorithm in combination with the bulk cargo shape data obtained by the 3D visual recognition system; the cargo hold and bulk cargo data and the weight and volume data obtained by the 3D visual recognition are integrated and input into the software. Based on these data, the available space in the cargo hold, the carrying capacity of different areas and the characteristics of the bulk cargo are analyzed, and a preliminary stacking strategy is formulated; Step S3: Determine the starting position of the bottom layer of bulk cargo based on the 3D visual data and the calculated weight and volume of the bulk cargo, so that the bottom layer is not directly close to the bulkhead and space is reserved; if the bulk cargo is heavy and large in size, the area with a strong bearing capacity at the bottom of the cargo hold is preferred for stacking; if the bulk cargo is light and small in size, it can be appropriately stacked at a position close to the reserved space on the bulkhead, but the stability of the bottom layer must be ensured.

[0005] Step S4: Starting from the second layer, the cargo is gradually placed closer to the bulkhead, and the inclination gradient is dynamically adjusted according to the shape of the cargo hold and the characteristics of the bulk cargo; the stacking height, inclination angle and distance from the bulkhead of each layer of bulk cargo are monitored in real time through the 3D vision system, and the degree of proximity of each layer to the bulkhead is accurately controlled according to the weight and volume of the bulk cargo. If it is a heavier and larger bulk cargo, the proximity is relatively small to prevent the center of gravity from shifting too much; for lighter and smaller bulk cargo, the proximity can be appropriately increased, and for larger bulk cargo, staggered stacking is performed during the stacking process so that the cargoes bite each other and enhance the overall stability; for small bulk cargo, the neat stacking between layers is also guaranteed based on the 3D vision data to ensure a stable connection.

[0006] Step S5: When stacking to the top layer, the bulk cargo is completely attached to the bulkhead, forming a trapezoidal structure with a wide bottom, a narrow top, and an overall tilt toward the bulkhead. After the top layer is stacked, the 3D visual recognition system is used again to conduct a comprehensive scan of the entire cargo pile, and combined with the weight data, the stability of the cargo pile, whether the tilt angle meets the requirements, and whether the connection between the cargo is tight. If any problems are found, timely adjustments and reinforcements are made.

[0007] The present invention is further configured such that the 3D visual recognition equipment in step S1 includes a laser scanner and a camera, which are used to obtain high-precision three-dimensional point cloud data inside the cargo hold. The 3D visual recognition system constructs a deep learning algorithm model through multi-view point cloud scanning, detects the material stacking posture in real time, predicts the risk of bag collapse and feeds back to the control terminal.

[0008] The present invention is further configured that the deep learning algorithm steps of the 3D visual recognition system in step S1 are: Step A1: Use a multi-view 3D camera or laser scanner to collect point cloud data of materials in the cabin to obtain the shape, position, layer height and gap information of the materials; and pre-process the data, denoise the point cloud data, remove abnormal points or background interference, convert the point cloud data into a voxel grid for input into the deep learning model, and normalize the data to ensure the scale consistency of the input data.

[0009] Step A2: Use the point cloud processing network as the basic model to extract the geometric features and spatial distribution characteristics of materials; collect a large amount of point cloud data of different material stacking scenarios, including normal stacking, inclined stacking and collapsed packages, and annotate the data. The annotation content includes material boundaries, layer heights, gaps and collapse risk levels. Input the preprocessed data into the deep learning model and train the model through supervised learning.

[0010] Step A3: Input the real-time collected point cloud data into the trained deep learning model to extract the geometric features and spatial distribution features of the material; judge whether the stacking posture of the current material meets the requirements of the trapezoidal structure through the classification network, identify whether there are risks of inclination, offset or collapse, and predict the stability score of the material stacking through the regression network based on the extracted features to evaluate the collapse risk level.

[0011] Step A4: According to the output results of the deep learning model, dynamically adjust the stacking gradient, displacement parameters and inclination angle to ensure the geometric accuracy and stability of the trapezoidal structure; feedback the optimized parameters to the control terminal to guide the stacking equipment to make real-time adjustments, continuously collect data during the operation process, update the model parameters, and achieve adaptive optimization.

[0012] Step A5: In actual operation, continuously collect new point cloud data, update the model through incremental learning, and adapt to different scenarios and material characteristics.

[0013] The present invention is further configured such that after the bulk cargo stacking is completed in step S5, a 3D vision recognition device is used to review the stacking result to ensure that the stacking structure meets the preset inclination gradient requirements.

[0014] The present invention is further configured such that the width of the reserved gap in step S3 is 40% - 60% of the width of a single package of goods, and the inclination angle of the trapezoidal structure in step S4 is 5 - 15°. The specific value of the inclination angle is automatically adapted according to the ship navigation environment parameters.

[0015] The present invention is further configured such that in step S4, the specific operation steps of the dynamic adjustment algorithm are as follows: Step C1: Real-time collect the point cloud data on the surface of the bulk cargo through a 3D vision recognition device, perform denoising and filtering processing on the point cloud data, and extract the geometric features of the surface of the bulk cargo; The denoising processing formula is:

[0016] Where, is the intensity value of point after denoising, is the normalized weight, is the neighborhood centered on is the spatial Gaussian kernel, controlling the weight of the spatial distance of neighborhood points, is the spatial standard deviation; is the range Gaussian kernel, controlling the weight of the intensity value difference of neighborhood points, is the range standard deviation, is the intensity value of point and are points and point The original intensity value; Using the principal component analysis algorithm. Let the point cloud data matrix be , and its covariance matrix , is the number of points in the point cloud. Perform eigenvalue decomposition on , . The eigenvector corresponding to the largest eigenvalue in the eigenvector matrix corresponds to the main direction of the point cloud, and the geometric features of the bulk cargo surface are extracted accordingly; Step C2: Real-time obtain the weight data of each batch of bulk cargo through the weight sensor on the loading and unloading equipment, and extract the geometric information of the bulkhead from the three-dimensional model of the cargo hold; Step C3: Use the least squares algorithm to perform plane fitting on the point cloud data of the bulk cargo surface, obtain the geometric plane equation of the bulk cargo surface, and calculate the current inclination angle between the bulk cargo surface and the horizontal plane; Let the point cloud data points on the bulk cargo surface be , and the plane equation be , . The objective is to minimize the error function ; Construct the matrix , the vector , the vector . Then, obtain the plane equation coefficients by solving the normal equation ; Calculation of the inclination angle: Let the normal vector of the plane be , and the normal vector of the horizontal plane be . According to the vector dot product formula , the inclination angle between the bulk cargo surface and the horizontal plane can be calculated as .

[0017] Step C4: Obtain the target inclination angle of the current layer from the preset inclination gradient stowage requirements, and calculate the deviation between the current inclination angle and the target inclination angle; Let the current inclination angle be , and the target inclination angle of the current layer obtained from the preset inclination gradient stowage requirements be . Then, the inclination angle deviation .

[0018] Step C5: There is a deviation between the current inclination angle and the target inclination angle. The control system adjusts the operation parameters of the loading and unloading equipment according to the magnitude and direction of the deviation; The control system adjusts the operation parameters of the loading and unloading equipment according to the deviation . Assume that the operation parameters of the loading and unloading equipment are speed , angle , etc. The adjustment rule can be set as , where , is a proportionality coefficient set according to the actual situation and is used to control the amplitude of adjustment; Step C6: After each adjustment, use the 3D vision recognition device to rescan the bulk cargo stacking state, verify whether the tilt angle reaches the target value, and record the adjustment data of each tilt angle into the control system to form historical data.

[0019] After each adjustment, the 3D vision recognition device rescans to obtain new point cloud data of the bulk cargo surface, and repeats Step C3 to calculate the new tilt angle and verify whether it meets , is a preset precision threshold. Record the tilt angle deviation of each adjustment, the operating parameters after adjustment and other data into the control system. Let the recorded data format be to form historical data for subsequent analysis and optimization.

[0020] The present invention is further configured such that in Step C4, if the result data shows that the angle exceeds the set angle threshold, the tilt angle is too small; otherwise, the tilt angle is too large until the angle difference is within ±0.5° of the set temperature threshold; in Step C5, when increasing the tilt angle, move the bulk cargo towards the bulkhead direction, and when decreasing the tilt angle, move the bulk cargo towards the center direction of the cargo hold.

[0021] The beneficial effects of the present invention are as follows: 1. The stacking method of the present invention forms a trapezoidal structure with a wide bottom and a narrow top through a specific inclined gradient stacking method in the hold, effectively reducing the center of gravity of the goods, preventing package collapse caused by the center of gravity shift, reducing potential safety hazards such as goods falling, and protecting the lives and health of workers. In actual bulk cargo loading and unloading operations, traditional stacking methods are prone to accidents due to unstable stacking of goods. However, the present invention optimizes from the stacking structure, greatly improving the operation safety, simplifying the cargo stacking process, shortening the ship loading time, and making the entire operation process more smooth and efficient.

[0022] 2. The present invention sets aside space from the bottom layer, gradually approaches the bulkhead layer by layer, and finally fits the top layer to the bulkhead. The entire stacking process is dynamically adjusted in close combination with the shape of the cargo hold and the characteristics of the bulk cargo, enabling the goods to support and restrain each other, enhancing the overall stability, protecting the integrity of the goods, and reducing damage to the goods caused by shaking and collision during transportation.

[0023] 3. The present invention uses a 3D vision recognition device and a weight sensor to quickly obtain key data such as the cargo hold space, the weight and volume of the bulk cargo, accurately plan the stacking position and method, avoid time waste caused by blind stacking, and at the same time, the dynamic adjustment algorithm can optimize the stacking process in real time, reduce stacking errors and subsequent adjustments, effectively shorten the ship loading time, and improve the operation efficiency. Detailed implementation manners

[0024] The present invention will be further described in detail below in conjunction with specific implementation manners. The embodiments of the present invention are given for the purposes of illustration and description, and are not exhaustive or limit the present invention to the disclosed form. Many modifications and variations will be obvious to those of ordinary skill in the art. The embodiments are selected and described to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thus design various embodiments with various modifications suitable for specific purposes. Embodiment

[0025] The present invention provides a technical solution: a bulk cargo stacking method based on an inclined gradient code bin method, comprising the following steps: Step S1: Before stacking the bulk cargo, use a 3D vision recognition device to perform an omni-directional scan of the internal space of the cargo hold to obtain a three-dimensional model of the cargo hold; including information such as the shape and size of the hold wall, and the position and shape of obstacles inside the hold. At the same time, scan the bulk cargo to be stacked and identify the shape, contour, and surface characteristics of each batch of bulk cargo.

[0026] Step S2: Through a weight sensor connected to the loading and unloading equipment, measure the weight of each batch of bulk cargo in real time, and combine it with the bulk cargo shape data obtained by the 3D vision recognition system. Using a volume calculation algorithm, calculate the volume of the bulk cargo; integrate the cargo hold and bulk cargo data obtained by 3D vision recognition, as well as the weight and volume data, and input them into software. Based on these data, analyze the available space of the cargo hold, the bearing capacity of different regions, and the characteristics of the bulk cargo, and formulate a preliminary stacking strategy.

[0027] Step S3: According to the 3D vision data and the calculated weight and volume of the bulk cargo, determine the starting position of the bottom layer of the bulk cargo stacking, so that the bottom layer does not directly adhere to the hold wall, leaving a space; if the bulk cargo is heavy and large in volume, preferentially select the area with stronger bearing capacity at the bottom of the cargo hold for stacking; if the bulk cargo is light and small in volume, it can be appropriately stacked starting from a position near the space reserved by the hold wall, but the stability of the bottom layer should be ensured.

[0028] Step S4: Starting from the second layer, the goods are gradually placed closer to the bulkhead while loading bulk cargo. The inclination gradient is dynamically adjusted according to the shape of the cargo hold and the characteristics of the bulk cargo. The 3D vision system is used to monitor in real time the stacking height, inclination angle, and distance from the bulkhead of each layer of bulk cargo. According to the weight and volume of the bulk cargo, the amplitude of approaching the bulkhead for each layer is precisely controlled. If it is a large-volume and heavy bulk cargo, the approaching amplitude is relatively small to prevent excessive center-of-gravity shift. For light and small-volume bulk cargo, the approaching amplitude can be appropriately increased. For large-volume bulk cargo, interlaced stacking is carried out during the stacking process to make the goods bite each other and enhance the overall stability. For small-volume bulk cargo, the neat stacking between layers is also ensured according to the 3D vision data to ensure firm connection.

[0029] Step S5: When stacking reaches the top layer, the bulk cargo is made to fully fit against the bulkhead, forming a trapezoidal structure that is wider at the bottom, narrower at the top, and overall inclined towards the bulkhead. After the top-layer stacking is completed, the 3D vision recognition system is used again to comprehensively scan the entire cargo stack. Combining with the weight data, the stability of the cargo stack, whether the inclination angle meets the requirements, and the tightness of the connection between the goods are checked. If any problems are found, adjustments and reinforcements are made in a timely manner.

[0030] The 3D vision recognition device in Step S1 includes a laser scanner and a camera, which are used to obtain high-precision three-dimensional point cloud data inside the cargo hold. The 3D vision recognition system constructs a deep learning algorithm model through multi-viewpoint cloud scanning, real-time detects the stacking posture of the materials, predicts the risk of package collapse, and feeds it back to the control terminal.

[0031] The deep learning algorithm steps of the 3D vision recognition system in Step S1 are as follows: Step A1: Through a multi-view 3D camera or a laser scanner, the point cloud data of the materials in the hold is collected to obtain the shape, position, layer height, and gap information of the materials. And the data is preprocessed. The point cloud data is denoised, abnormal points or background interference are removed, and the point cloud data is converted into a voxel grid for input into the deep learning model. The data is normalized to ensure the scale consistency of the input data.

[0032] Step A2: A point cloud processing network is used as the basic model to extract the geometric features and spatial distribution features of the materials. A large amount of point cloud data of different material stacking scenarios is collected, including normal stacking, inclined stacking, and package-collapse scenarios. The data is labeled, and the labeling content includes material boundaries, layer height, gaps, and package-collapse risk levels. The preprocessed data is input into the deep learning model, and the model is trained through supervised learning.

[0033] Step A3: Input the real-time collected point cloud data into the trained deep learning model to extract the geometric features and spatial distribution features of the materials; use the classification network to determine whether the stacking posture of the current material meets the requirements of the trapezoidal structure, identify whether there are risks of inclination, offset or collapse, and based on the extracted features, predict the stability score of the material stacking through the regression network to evaluate the collapse risk level.

[0034] Step A4: According to the output results of the deep learning model, dynamically adjust the stacking gradient, displacement parameters and inclination angle to ensure the geometric accuracy and stability of the trapezoidal structure; feedback the optimized parameters to the control terminal to guide the stacking equipment for real-time adjustment, continuously collect data during the operation process, update the model parameters, and achieve adaptive optimization.

[0035] Step A5: In actual operation, continuously collect new point cloud data and update the model through incremental learning to adapt to different scenarios and material characteristics.

[0036] In step S5, after the bulk cargo stacking is completed, use the 3D vision recognition device to review the stacking result to ensure that the stacking structure meets the preset inclination gradient requirements. The width of the reserved gap in step S3 is 40% - 60% of the width of a single package of goods, and the inclination angle of the trapezoidal structure in step S4 is 5 - 15°.

[0037] In step S4, the specific operation steps of the dynamic adjustment algorithm are as follows: Step C1: Use the 3D vision recognition device to collect the point cloud data on the surface of the bulk cargo in real time, denoise and filter the point cloud data, and extract the geometric features of the surface of the bulk cargo. Step C2: Use the weight sensor on the loading and unloading equipment to obtain the weight data of each batch of bulk cargo in real time, and extract the geometric information of the cargo hold wall from the three-dimensional model of the cargo hold. Step C3: Use the least squares algorithm to perform plane fitting on the point cloud data on the surface of the bulk cargo to obtain the geometric plane equation of the surface of the bulk cargo, and calculate the current inclination angle between the surface of the bulk cargo and the horizontal plane. Step C4: Obtain the target inclination angle of the current layer from the preset inclination gradient stowage requirements, and calculate the deviation between the current inclination angle and the target inclination angle. Step C5: If there is a deviation between the current inclination angle and the target inclination angle, the control system adjusts the operation parameters of the loading and unloading equipment according to the magnitude and direction of the deviation. Step C6: After each adjustment, use the 3D vision recognition device to re-scan the stacking state of the bulk cargo to verify whether the inclination angle reaches the target value, and record the adjustment data of each inclination angle into the control system to form historical data.

[0038] In step C4, if the angle shown in the result data exceeds the set angle threshold, the inclination angle is too small; otherwise, the inclination angle is too large, until the angle difference is within ±0.5° of the set temperature threshold. In step C5, when increasing the inclination angle, move the bulk cargo towards the bulkhead direction, and when decreasing the inclination angle, move the bulk cargo towards the center direction of the cargo hold.

[0039] Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art and related fields without creative efforts shall fall within the scope of protection of the present invention. The structures, devices, and operation methods not specifically described and explained in the present invention, unless otherwise specified and limited, shall be implemented according to the conventional means in the art.

Claims

1. A bulk cargo stacking method based on the inclined gradient code bin method, characterized in that It includes the following steps: Step S1: Before bulk goods are stacked, use a 3D vision recognition device to conduct an all-round scan of the internal space of the cargo hold to obtain a three-dimensional model of the cargo hold; Step S2: Through a weight sensor connected to the loading and unloading equipment, measure the weight of each batch of bulk goods in real time, and combine the shape data of the bulk goods obtained by the 3D vision recognition system. Using a volume calculation algorithm, calculate the volume of the bulk goods; Step S3: Based on the 3D vision data and the calculated weight and volume of the bulk goods, determine the starting position for stacking the bottom layer of bulk goods, so that the bottom layer does not directly adhere to the cabin wall, leaving a space; Step S4: Starting from the second layer, gradually place the goods closer to the cabin wall. The inclination gradient is dynamically adjusted according to the shape of the cargo hold and the characteristics of the bulk goods; Step S5: When stacking reaches the top layer, make the bulk goods fully fit against the cabin wall, forming a trapezoidal structure that is wide at the bottom, narrow at the top, and inclined as a whole towards the cabin wall.

2. The bulk cargo stacking method based on the inclined gradient code bin method according to claim 1, characterized in that: In step S1, the 3D vision recognition device includes a laser scanner and a camera, which are used to obtain high-precision three-dimensional point cloud data inside the cargo hold. The 3D vision recognition system constructs a deep learning algorithm model through multi-viewpoint cloud scanning, real-time detects the stacking posture of materials, predicts the risk of package collapse, and feeds it back to the control terminal.

3. A bulk cargo stacking method based on the inclined gradient code bin method according to claim 2, characterized in that: The deep learning algorithm steps of the 3D vision recognition system in step S1 are as follows: Step A1: Through a multi-view 3D camera or laser scanner, collect the point cloud data of the materials in the cabin to obtain information on the shape, position, layer height, and gap of the materials; Step A2: Use a point cloud processing network as the basic model to extract the geometric features and spatial distribution features of the materials; Step A3: Input the real-time collected point cloud data into the trained deep learning model to extract the geometric features and spatial distribution features of the materials; Step A4: According to the output results of the deep learning model, dynamically adjust the stacking gradient, displacement parameters, and inclination angle to ensure the geometric accuracy and stability of the trapezoidal structure; Step A5: During actual operations, continuously collect new point cloud data, update the model through incremental learning, and adapt to different scenarios and material characteristics.

4. A bulk cargo stacking method based on the inclined gradient code bin method according to claim 1, characterized in that: In step S5, after the bulk goods are stacked, use a 3D vision recognition device to review the stacking result to ensure that the stacking structure meets the preset inclination gradient requirements.

5. A bulk cargo stacking method based on the inclined gradient code bin method according to claim 1, characterized in that: The width of the reserved gap in step S3 is 40% - 60% of the width of a single package of goods. The inclination angle of the trapezoidal structure in step S4 is 5 - 15°.

6. The bulk cargo stacking method based on the inclined gradient code bin method according to claim 1, characterized in that: In step S4, the specific operation steps of the dynamic adjustment algorithm are as follows: Step C1: Through a 3D vision recognition device, collect the point cloud data on the surface of the bulk goods in real time, perform denoising and filtering processing on the point cloud data, and extract the geometric features of the surface of the bulk goods; Step C2: Through the weight sensor on the loading and unloading equipment, obtain the weight data of each batch of bulk goods in real time, and extract the geometric information of the cabin wall from the three-dimensional model of the cargo hold; Step C3: Use the least squares algorithm to perform plane fitting on the point cloud data on the surface of the bulk goods to obtain the geometric plane equation of the surface of the bulk goods, and calculate the current inclination angle between the surface of the bulk goods and the horizontal plane; Step C4: Obtain the target tilt angle of the current layer from the preset requirements for the inclined gradient code bin, and calculate the deviation between the current tilt angle and the target tilt angle; Step C5: If there is a deviation between the current tilt angle and the target tilt angle, the control system adjusts the operating parameters of the loading and unloading equipment according to the magnitude and direction of the deviation; Step C6: After each adjustment, use the 3D vision recognition device to rescan the bulk cargo stacking state, verify whether the tilt angle reaches the target value, and record the adjustment data of each tilt angle into the control system to form historical data.

7. A bulk cargo stacking method based on the inclined gradient code bin method according to claim 6, characterized in that: In step C4, if the result data shows that the angle exceeds the set angle threshold, the tilt angle is too small; otherwise, the tilt angle is too large, until the angle difference is within ±0.5° of the set temperature threshold; in step C5, when increasing the tilt angle, move the bulk cargo towards the bulkhead direction, and when decreasing the tilt angle, move the bulk cargo towards the center of the cargo hold.

Images