Intelligent silo bulk cargo loading method and system

Abstract

The invention relates to the technical field of port automation, in particular to an intelligent silo bulk cargo loading method and system.The method comprises the following steps that silo bulk cargo loading information is obtained, and a silo bulk cargo-loading vehicle matching result is obtained based on the silo bulk cargo loading information; establishing a multi-dimensional loading information sensing array, and obtaining a multi-dimensional real-time loading sensing data set in combination with the silo bulk cargo-loading vehicle matching result; carrying out distribution analysis on loaded bulk materials according to the multi-dimensional real-time loading sensing data set to obtain the dynamic stacking volume of the bulk materials; building a compartment bulk cargo stacking state model based on the bulk cargo dynamic stacking volume, and obtaining a silo bulk cargo loading control instruction according to the compartment bulk cargo stacking state model; and an online adjusting and optimizing mechanism is constructed, dynamic optimization is conducted on the silo bulk material loading control instruction, loading optimization control parameters are obtained, and intelligent loading of silo bulk materials is achieved. The system adapts to port bulk cargo loading requirements, and intelligence of silo bulk cargo loading is achieved.

Description

Technical Field

[0001] This invention relates to the field of port automation technology, and in particular to an intelligent loading method and system for bulk materials from silos. Background Technology

[0002] Currently, bulk material loading from silos primarily relies on a combination of basic automated equipment and manual intervention. Specifically, existing technologies typically use weighbridges to obtain vehicle tare and gross weights, combined with quantitative feeders to control the total load. Laser scanners or single cameras perform a rough volume scan of the truck bed, and programmable logic controllers (PLCs) control the start / stop of loading equipment and basic process control. Some advanced systems incorporate vehicle recognition technology to automatically associate transport tasks, using simple proportional-integral-derivative (PID) controllers to adjust feeder speeds and guiding vehicle movement with on-site audio-visual indicators. These solutions, through the initial integration of sensors, control units, and actuators, have shifted loading operations from fully manual to semi-automated, improving efficiency to some extent.

[0003] However, existing technical solutions have several drawbacks: First, the perception system is limited in scope, and the measurement accuracy of laser or vision sensors drops sharply under harsh conditions such as dust, rain, and fog, and they cannot accurately reconstruct the bulk material accumulation pattern, resulting in significant deviations in volume calculation. Second, the control strategy lacks foresight and adaptability; using fixed PID parameters makes it difficult to cope with changes in different material characteristics, and overshoot or undershoot is common, making it impossible to achieve "one-time accurate loading." Third, the information of each subsystem is isolated, and the business scheduling, vehicle identification, and loading control links are disconnected, requiring multiple manual verifications, which can easily lead to loading errors. Finally, there is a lack of self-learning optimization capabilities; parameters rely on experience-based settings and cannot continuously optimize the control strategy from historical operation data, resulting in limited overall intelligence and affecting the improvement of loading accuracy and operational efficiency. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent loading method and system for bulk materials from silos.

[0005] To achieve the above objectives, in a first aspect, the present invention provides an intelligent loading method for bulk materials from silos. The method includes the following steps: acquiring silo bulk material loading information; obtaining a silo bulk material-loading vehicle matching result based on the silo bulk material loading information; establishing a multi-dimensional loading information sensing array; obtaining a multi-dimensional real-time loading sensing dataset by combining the silo bulk material-loading vehicle matching result; performing distribution analysis on the loaded bulk material according to the multi-dimensional real-time loading sensing dataset to obtain the dynamic stacking volume of the bulk material; constructing a cargo compartment bulk material stacking state model based on the dynamic stacking volume of the bulk material; obtaining silo bulk material loading control commands according to the cargo compartment bulk material stacking state model; and constructing an online optimization mechanism to dynamically optimize the silo bulk material loading control commands to obtain loading optimization control parameters, thereby realizing intelligent loading of bulk materials from silos. The present invention first completes the matching of silos and vehicles based on loading information, and then, through multi-dimensional sensing, state modeling, and dynamic optimization, effectively avoids problems such as incorrect loading and spillage, improves loading accuracy and efficiency, reduces manual intervention, adapts to the large-scale bulk material loading needs of ports, and realizes intelligent loading of bulk materials from silos throughout the entire process.

[0006] Optionally, obtaining silo bulk material loading information and obtaining silo bulk material-loading vehicle matching results based on the silo bulk material loading information includes: obtaining loading vehicle information, the planned loading quantity of bulk material, and bulk material type information as the silo bulk material loading information; establishing business logic verification rules, comparing and verifying the silo bulk material loading information according to the business logic verification rules to obtain loading verification results; and obtaining the silo bulk material-loading vehicle matching results based on the loading verification results. The silo bulk material-loading vehicle matching results include the target loading silo number, the allowed loading time window, and the initial loading control parameters. This invention comprehensively verifies the silo bulk material loading information through business logic verification rules to ensure the information is true and valid. Based on the verification results, it outputs the target silo, time window, and initial parameters, providing accurate basis for subsequent loading stages, avoiding loading delays or errors caused by information errors, and improving the reliability and efficiency of pre-loading preparation.

[0007] Optionally, establishing business logic verification rules includes: constructing vehicle whitelist verification rules, electronic delivery note validity status verification rules, and material consistency verification rules as the business logic verification rules. This invention filters unqualified vehicles through vehicle whitelist rules, ensures the legality of delivery through electronic delivery note rules, and avoids incorrect loading through material consistency rules. The combination of these three rules forms a comprehensive verification system, controlling the preconditions for loading, reducing the risk of illegal loading, and ensuring the compliant and orderly operation of bulk material loading from silos.

[0008] Optionally, establishing a multi-dimensional loading information sensing array and combining it with the silo bulk material-loading vehicle matching results to obtain a multi-dimensional real-time loading sensing dataset includes: establishing the multi-dimensional loading information sensing array based on a non-repeating scanning LiDAR, a millimeter-wave radar level gauge, and an industrial camera; during the loading process based on the silo bulk material-loading vehicle matching results, acquiring three-dimensional point cloud data using the non-repeating scanning LiDAR, obtaining the height data below the hopper using the millimeter-wave radar level gauge, and obtaining auxiliary two-dimensional image data based on the industrial camera; and establishing the multi-dimensional real-time loading sensing dataset based on the three-dimensional point cloud data, the height data below the hopper, and the auxiliary two-dimensional image data. This invention utilizes a multi-dimensional loading information sensing array to collect data in real time during the loading process based on the silo bulk material-loading vehicle matching results, forming a multi-dimensional real-time loading sensing dataset that comprehensively reflects the loading status. Compared to single-sensor monitoring, the data is richer and more accurate, providing high-quality data support for subsequent bulk material volume analysis and status modeling.

[0009] Optionally, obtaining the dynamic stacking volume of bulk materials by analyzing the distribution of loaded bulk materials based on the multi-dimensional real-time loading perception dataset includes: constructing a three-dimensional mesh model of the bulk material stacking surface based on the multi-dimensional real-time loading perception dataset; projecting the three-dimensional mesh model of the bulk material stacking surface onto the coordinate system of the bottom surface of the loading vehicle to construct a bulk material surface height function; establishing a gradient compensation term; and calculating the dynamic stacking volume of bulk materials by combining the bulk material surface height function and the gradient compensation term. This invention, through multi-step data processing and model construction, fully leverages the value of perception data. The addition of the gradient compensation term can correct the impact of surface height changes on volume calculation, significantly improving the accuracy of dynamic stacking volume calculation and providing a precise data foundation for subsequent state modeling.

[0010] Optionally, the step of establishing a gradient compensation term and calculating the dynamic stacked volume of the bulk materials by combining the surface height function of the bulk materials and the gradient compensation term includes: performing an integral operation based on the surface height function of the bulk materials to obtain a basic volume estimate of the bulk material stack; obtaining the height change rate of each point in the surface height function of the bulk materials, and obtaining surface gradient information based on the height change rate; performing a second integral operation based on the surface gradient information to obtain a gradient compensation volume value, and using the gradient compensation volume value as the gradient compensation term; and superimposing the basic volume estimate and the gradient compensation volume value to obtain the dynamic stacked volume of the bulk materials. This invention effectively optimizes the volume results through a step-by-step calculation method. The gradient compensation term effectively corrects the volume error caused by the unevenness of the bulk material surface. Compared with direct volume calculation, the results are more consistent with the actual loading situation, further improving the accuracy and reliability of bulk material volume calculation.

[0011] Optionally, the step of constructing a bulk material stacking state model for the wagon based on the dynamic stacking volume of the bulk materials, and obtaining silo bulk material loading control instructions based on the bulk material stacking state model, includes: using the dynamic stacking volume of the bulk materials as the core model data source, and combining it with the multi-dimensional real-time loading perception dataset to construct the bulk material stacking state model for the wagon; performing forward prediction of the bulk material stacking state for the wagon based on the bulk material stacking state model to obtain the predicted bulk material loading state; using loading accuracy, loading efficiency, bulk material stacking uniformity, and no spillage as multiple optimization objectives, and constructing a multi-objective optimization function based on the multiple optimization objectives; solving the multi-objective optimization function in combination with the predicted bulk material loading state to obtain the silo bulk material loading control instructions, which include feeding control instructions and vehicle movement instructions. This invention constructs a state model using dynamic volume as the core data source, accurately reflecting the bulk material stacking situation in the wagon; forward prediction can predict the loading state in advance, facilitating timely adjustments; and the silo bulk material loading control instructions are obtained by solving the multi-objective optimization function, achieving uniform bulk material loading and improving overall loading efficiency.

[0012] Optionally, the construction of the online optimization mechanism includes: during the execution of the silo bulk material loading control command, calculating the bulk material loading performance evaluation index in real time, and simultaneously monitoring and quantifying real-time loading disturbance characteristics; establishing a parameter optimization knowledge base and a control rule base based on historical loading data, and establishing the online optimization mechanism in conjunction with the real-time loading disturbance characteristics. This invention calculates performance indicators and monitors and quantifies disturbance characteristics in real time during command execution, and simultaneously builds a knowledge base and rule base based on historical data. The constructed online optimization mechanism can dynamically respond to changes during the loading process; it can promptly detect loading performance problems and provide a basis for optimization based on historical data, ensuring a stable and efficient loading process.

[0013] Optionally, the step of dynamically optimizing the silo bulk material loading control command to obtain loading optimization control parameters and realize intelligent loading of silo bulk materials includes: obtaining an optimization strategy matching result based on the real-time loading disturbance characteristics and the parameter optimization knowledge base, and obtaining the control parameter adjustment amount by combining the control rule base; using the optimization of the bulk material loading performance evaluation index as the adjustment target, and combining the control parameter adjustment amount and the silo bulk material loading control command to obtain the loading optimization control parameters, thereby realizing intelligent loading of silo bulk materials. This invention effectively addresses real-time disturbances during loading by dynamically correcting the loading control command, ensuring high loading accuracy and efficiency even under complex operating conditions, and achieving continuous optimization and stable operation of silo bulk material loading.

[0014] Secondly, this invention provides an intelligent bulk material loading system for silos. The system executes the intelligent bulk material loading method for silos provided by this invention. The system includes an input device, an output device, a processor, and a memory, all interconnected. The memory stores a computer program, which includes program instructions, and the processor is configured to invoke these instructions. This invention, through the collaborative operation of high-performance hardware, stably realizes an intelligent bulk material loading process, adapts to the port industrial environment, efficiently supports intelligent bulk material loading operations from silos, and improves the automation and intelligence level of loading operations. Attached Figure Description

[0015] Figure 1 This is a flowchart of an intelligent loading method for bulk materials from silos according to an embodiment of the present invention; Figure 2 This is a framework diagram of an intelligent loading system for bulk materials in silos according to an embodiment of the present invention. Detailed Implementation

[0016] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0017] Throughout this specification, references to "an embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "in an embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0018] Please see Figure 1 One embodiment of the present invention provides an intelligent loading method for bulk materials from silos, the method comprising the following steps: S1. Obtain silo bulk material loading information, and obtain silo bulk material-loading vehicle matching results based on the silo bulk material loading information.

[0019] In this embodiment, the loading vehicle information, the predetermined loading volume of bulk materials, and the bulk material type information are obtained as the silo bulk material loading information; business logic verification rules are established, and the silo bulk material loading information is compared and verified according to the business logic verification rules to obtain the loading verification result; based on the loading verification result, the silo bulk material-loading vehicle matching result is obtained, which includes the target loading silo number, the allowed loading time window, and the initial loading control parameters.

[0020] Specifically, vehicle identification codes are automatically obtained by reading vehicle electronic tags using RFID readers deployed at the port entrance. Simultaneously, high-definition license plate recognition cameras capture vehicle license plate images and convert them into digital license plate information using Optical Character Recognition (OCR) technology. The vehicle identification code and license plate information are used as the vehicle loading information. The planned loading quantity and type of bulk materials are obtained by connecting to the Enterprise Resource Planning (ERP) system, retrieving detailed order information from the corresponding electronic delivery order in real time. This multi-source information is automatically associated and bound using a unified timestamp and task identifier to form structured silo bulk material loading information. The entire information collection process is fully automated, ensuring the real-time nature and accuracy of data acquisition.

[0021] In this embodiment, vehicle whitelist verification rules, electronic delivery note validity status verification rules, and material consistency verification rules are constructed as business logic verification rules.

[0022] Specifically, firstly, a vehicle whitelist verification rule is established, which involves creating a database of authorized reserved vehicles and determining whether the current vehicle is on the whitelist during verification. Secondly, an electronic delivery note validity verification rule is established, which verifies the delivery note's digital signature, validity period, and usage status (whether it has been cancelled) to determine its legality. Finally, a material consistency verification rule is established, which verifies whether the planned loading materials match the actual materials stored in the designated silos by creating a silo-material correspondence database. These three layers of verification rules use a combination of logic and relations; only when all verifications pass is the verification considered successful, thus forming a complete business logic verification rule.

[0023] Furthermore, upon receiving structured silo bulk material loading information, a multi-threaded verification process is automatically initiated. First, a vehicle whitelist verification is performed, matching the obtained vehicle license plate information with the authorized vehicle list in the reservation database. Second, an electronic delivery note status verification is performed, querying the current status and authenticity of the delivery note through a blockchain verification interface to obtain a second verification result. Finally, a material consistency verification is performed, comparing the material code in the delivery note with the material identifier of the target silo to obtain a third verification result. The first, second, and third verification results are used as the loading verification result. A preset scoring threshold is established based on actual loading requirements. All loading verification results are recorded in real-time and weighted. A verification pass label is generated only when all verification item scores reach the scoring threshold; otherwise, a verification failure report containing the reason for the failure is generated, triggering an alarm mechanism to notify engineering personnel for emergency handling.

[0024] In this embodiment, after obtaining the loading verification result, the target loading silo number is determined by querying the silo-material distribution map based on the material type code in the delivery note; the optimal operation time window is calculated based on the current operating load and queuing situation, and an allowable loading time window is generated in conjunction with transportation time restrictions; based on historical loading data and the characteristic parameters of the material (such as angle of repose and density), initial loading control parameters, including the initial speed of the feeder and the size of the hopper opening, are intelligently generated. The final silo bulk material-loading vehicle matching result is sent to the loading control terminal and the vehicle navigation system to start the subsequent automatic loading process.

[0025] It should be noted that the loading process is as follows: First, the initial material passage is controlled by a gate valve. Then, the feeding amount is adjusted by a variable speed feeder. Subsequently, the material is transported by a scraper elevator and then enters a dust suppression hopper to reduce dust. Finally, the material is loaded into a container truck. All links work together to complete the loading operation.

[0026] In an optional embodiment, in the application scenario of the quantitative variable speed feeder, to solve the problem of material accumulation, a dust collection device is set up to collect the return material and transfer it directly to the scraper conveyor. The device and the scraper conveyor adopt an integral sealed structure to ensure a closed dust suppression effect. To address belt misalignment, an automatic misalignment device is equipped. When misalignment occurs, the misalignment idler roller is automatically adjusted, and the centering of the receiving device is adjusted to prevent misalignment from the source. An adjusting rod is set at the tail roller to adjust the roller angle and reduce misalignment. To ensure weighing accuracy, a frame structure is adopted, equipped with two sets of weighing units, paired with high-precision sensors, and an overload protection device is set up to prevent damage to the sensors due to overload during operation or transportation. In terms of maintenance, the sealing cover is designed as a detachable modular structure with an observation window. The belt tensioning device is external for easy adjustment, and an external hanging weight verification device is configured for easy weighing calibration.

[0027] S2. Establish a multi-dimensional loading information sensing array, and combine the silo bulk material-loading vehicle matching results to obtain a multi-dimensional real-time loading sensing dataset.

[0028] In this embodiment, a multi-dimensional loading information sensing array is established based on a non-repeating scanning lidar, a millimeter-wave radar level gauge, and an industrial camera. During the loading process of matching bulk materials in the silo with the loading vehicle, three-dimensional point cloud data is acquired using the non-repeating scanning lidar, the height data below the hopper is obtained using the millimeter-wave radar level gauge, and auxiliary two-dimensional image data is obtained using the industrial camera. A multi-dimensional real-time loading sensing dataset is established based on the three-dimensional point cloud data, the height data below the hopper, and the auxiliary two-dimensional image data.

[0029] Two non-repetitive scanning LiDARs are symmetrically deployed on both sides of the loading chute. Their installation positions are precisely calculated to ensure a complete scanning coverage without blind spots, fully covering the entire process from empty to full load. A millimeter-wave radar level gauge is installed at the outlet of the dust suppression hopper. Its installation angle is adjusted so that its beam is perpendicular to the falling trajectory of the bulk material, monitoring the distance from the hopper to the material surface in real time. A high-definition industrial camera is mounted above the loading area, using a wide-angle lens to ensure complete coverage of the entire truck compartment area. It is equipped with an automatic dust removal device and a waterproof housing to adapt to the complex operating environment. All sensors are connected to the central processing unit via industrial Ethernet and are synchronized using a unified timing controller to ensure the spatiotemporal consistency of the collected data, ultimately forming a multi-dimensional loading information sensing array capable of simultaneously acquiring point clouds, distance, and images.

[0030] Upon receiving the matching result between the bulk material from the silo and the loading vehicle, confirming the vehicle's positioning, and authorizing the start of the loading process, the multi-dimensional sensing array works collaboratively according to preset instructions. A non-repetitive scanning LiDAR emits a laser beam at a frequency of 20Hz, performing high-speed scanning of the interior of the truck bed to generate three-dimensional point cloud data, reflecting the surface morphology and stacking height of the material in real time. A millimeter-wave radar level gauge continuously emits frequency-modulated continuous waves at a sampling frequency of 100Hz, using the echo phase difference to accurately calculate the real-time distance between the hopper outlet and the material surface, obtaining the height data below the hopper. Its millimeter-wave characteristics effectively penetrate dust interference generated during loading. An industrial camera acquires two-dimensional image data of the loading process at a frame rate of 30fps, extracting the truck bed edge features and material distribution visual information in real time through a built-in image processing unit. All sensor data is timestamped with millisecond-level precision and transmitted in real time to the central data processing unit.

[0031] After receiving the raw, timestamped sensor data, the central processing unit first performs spatiotemporal registration, unifying all data to the world coordinate system through coordinate transformation. The 3D point cloud data is preprocessed using voxel mesh downsampling and statistical outlier removal algorithms to obtain laser point cloud data. The height data below the hopper measured by the millimeter-wave radar level gauge is cross-validated with the point cloud height data, and then fused using a Kalman filter algorithm to generate reliable material level height information. Industrial image data undergoes distortion correction and feature extraction processing to assist in identifying the vehicle boundary and the approximate material distribution area. Finally, the laser point cloud data, the fused material level height data, and the preprocessed image feature data are aligned according to time series, and data quality identifiers and environmental parameter metadata are added to form a structured, multi-dimensional real-time loading perception dataset, providing a complete and reliable data foundation for subsequent distribution analysis and volume calculation.

[0032] S3. Based on the multi-dimensional real-time loading perception dataset, perform distribution analysis on the loaded bulk materials to obtain the dynamic stacking volume of the bulk materials.

[0033] In this embodiment, a three-dimensional mesh model of the bulk material stacking surface is constructed based on a multi-dimensional real-time loading perception dataset; the three-dimensional mesh model of the bulk material stacking surface is projected onto the coordinate system of the bottom surface of the loading vehicle to construct a bulk material surface height function; a gradient compensation term is established, and the dynamic stacking volume of the bulk material is calculated by combining the bulk material surface height function and the gradient compensation term.

[0034] Specifically, preprocessed laser point cloud data is extracted from a multi-dimensional real-time loading and sensing dataset. A deep learning-based point cloud segmentation algorithm is used to divide the point cloud data into three independent clusters: ground point cloud, carriage wall point cloud, and bulk material surface point cloud. Data simplification is performed on the bulk material surface point cloud to reduce computational complexity while preserving the surface characteristics. To improve model accuracy, millimeter-wave radar ranging data is integrated to fill in the gaps in the laser point cloud in dust-affected areas. Texture information collected by an industrial camera is used for texture mapping of the model, enhancing its visualization and feature recognition. This generates a discrete geometric structure composed of vertices, edges, and faces as a 3D mesh model of the bulk material accumulation surface. This model contains complete geometric information such as vertex coordinates, normal vectors, and triangular facet topological relationships, accurately representing the microscopic undulations and macroscopic accumulation morphology of the bulk material surface.

[0035] Furthermore, a coordinate system for the bottom surface of the cargo compartment is established, with the plane of the cargo compartment floor as the XY plane and the vertical upward direction as the Z-axis. The origin and orientation of the coordinate system are automatically calibrated using positioning targets installed at the four corners of the cargo compartment. A nearest-point iteration algorithm is used to transform the global coordinate system of the 3D mesh model of the bulk material accumulation surface to the coordinate system of the cargo compartment floor, ensuring accurate correspondence between the model and the actual spatial position of the cargo compartment. The plane of the cargo compartment floor is uniformly meshed, generating a regular array of sampling points in the XY direction. For each sampling point (x, y), its intersection with the 3D mesh model in the Z-axis direction is calculated. If an intersection exists, its height value z(x, y) is recorded; otherwise, the height value is calculated through interpolation of neighboring points. An adaptive sampling strategy is adopted for special areas (such as the edges of the cargo compartment and steep slopes of the material pile), increasing the sampling density in areas with drastic height changes. Finally, a height function of the bulk material surface is generated and stored in the form of a two-dimensional array, where each element stores the height value at the corresponding mesh point, and the confidence evaluation result of each point is recorded, providing a high-precision digital elevation data foundation for subsequent volume calculations.

[0036] In this embodiment, an integral operation is performed based on the surface height function of the bulk material to obtain the basic volume estimate of the bulk material stack; the height change rate of each point in the surface height function of the bulk material is obtained, and the surface gradient information is obtained based on the height change rate; a second integral operation is performed based on the surface gradient information to obtain the gradient compensation volume value, and the gradient compensation volume value is used as the gradient compensation term; the basic volume estimate and the gradient compensation volume value are superimposed to obtain the dynamic bulk material stack volume.

[0037] Specifically, the basic volume estimate is first calculated by performing a double integral operation based on the bulk material surface height function. The integration region is the projected area of ​​the truck bed bottom. The basic volume value is obtained by integrating the bulk material surface height function over the projected area. Then, the central difference method is used to calculate the height change rate at each point in the bulk material surface height function, obtaining the partial derivatives in the X and Y directions, and thus calculating the surface gradient at each point. A second integral operation is performed based on the surface gradient information to calculate the gradient compensation volume value. A dynamic adjustment coefficient related to the material's angle of repose is introduced to adjust the compensation intensity to adapt to the flow characteristics of different materials. Finally, the accurate dynamic bulk material stack volume is obtained by linearly superimposing the basic volume estimate and the gradient compensation volume value. The gradient compensation term effectively corrects the volume calculation error caused by the natural angle of repose of the bulk material, making it particularly suitable for calculating the volume of bulk material stacks with complex surface morphologies, significantly improving the accuracy and reliability of bulk material volume measurement. The entire calculation process is accelerated using a parallel computing architecture to ensure that the timeliness requirements of real-time loading control are met.

[0038] The above-mentioned dynamic stacking volume of bulk materials satisfies the following relationship: in, For the dynamic stacking volume of bulk materials, The horizontal projected area of ​​the carriage floor. As a function of the surface height of the bulk material, The height of the carriage floor. This is a dynamic adjustment coefficient related to the angle of repose of the material. For the height gradient of the bulk material surface, These are the tiny area units divided during the integration process.

[0039] In an optional embodiment, the ratio of the integrated dynamic stacked volume of bulk material to the cumulative feed weight data at the same time is calculated to estimate the instantaneous dynamic density of bulk material. The calculation confidence level of the dynamic stacked volume of bulk material is estimated by monitoring the variance or coefficient of variation of the instantaneous dynamic density.

[0040] S4. Construct a bulk material stacking state model for the wagon based on the dynamic stacking volume of the bulk materials, and obtain the silo bulk material loading control command according to the bulk material stacking state model for the wagon.

[0041] In this embodiment, the dynamic stacking volume of bulk materials is used as the core model data source, and a bulk material stacking state model of the car body is constructed by combining a multi-dimensional real-time loading perception dataset. Based on the bulk material stacking state model of the car body, the bulk material stacking state of the car body is forward-predicted to obtain the bulk material loading prediction state. Loading accuracy, loading efficiency, bulk material stacking uniformity, and no spillage are used as multiple optimization objectives, and a multi-objective optimization function is constructed based on the multiple optimization objectives. The multi-objective optimization function is solved in combination with the bulk material loading prediction state to obtain the silo bulk material loading control command, which includes feeding control command and vehicle movement command.

[0042] A 3D visualization modeling platform based on digital twin technology was established, using real-time acquired dynamic bulk material stacking volume data as the core, and integrating 3D point cloud data, hopper height data, and auxiliary 2D image data from a multi-dimensional perception dataset. A multi-layer data fusion architecture was adopted: the bottom layer processes geometric information reconstruction, generating an accurate geometric model of the bulk material stack through point cloud registration and surface reconstruction algorithms; the middle layer incorporates physical properties, using a material density database and mechanical property parameters to assign physical attributes such as mass distribution, internal friction angle, and cohesion to the geometric model; the top layer integrates state information, updating dynamic parameters such as stacking height, volume change rate, and surface flatness in real time. This constructs a comprehensive digital representation system of the bulk material stacking in the truck bed, including geometric features, physical properties, and state parameters, which can fully reflect the current state and characteristics of the bulk material stack, providing an accurate holographic data foundation for subsequent prediction and optimization. The truck bed bulk material stacking state model adopts an incremental update mechanism, refreshing in real time at a frequency of 20Hz to ensure synchronization accuracy with the physical world at the millisecond level.

[0043] A dual forward prediction mechanism combining physics-based simulation prediction and data-driven prediction is employed. The physics simulation prediction uses the discrete element method to establish a model of bulk material particle motion, simulating the trajectory and stacking morphology changes of the bulk material during continuous loading by solving the Newton-Euler equations. The data-driven prediction uses a long short-term memory neural network to learn the dynamic laws of bulk material stacking by training on historical loading data, predicting the stacking state at future time steps. The outputs of the two prediction methods are fused using Kalman filtering to generate the final bulk material loading prediction state. The prediction state includes information such as the expected stacking height distribution, material surface morphology evolution, and potential off-center loading area warnings every 0.1 seconds within the next 30 seconds. During the prediction process, influencing factors such as the current feed rate, material flow landing point, and vehicle attitude are considered in real time, and a rolling prediction mechanism is used, immediately re-initializing the prediction model based on the latest state after each prediction cycle to ensure the accuracy and timeliness of the prediction results.

[0044] In an optional embodiment, the forward prediction process is described using a state transition equation that satisfies the following relationship: in, For a moment The system state vector, For the current moment, This is the time period for forward estimation. This is the state transition function. For a moment The system state vector, For a moment The control input vector, For a moment The process noise vector.

[0045] Furthermore, a precise quantitative mapping relationship is first established for the four optimization objectives. The loading accuracy objective is quantified as the squared error term between the actual loaded weight and the target weight; the loading efficiency objective is defined as the ratio of loading quantity to loading time; the stacking uniformity objective is defined as the standard deviation of the material surface height distribution to minimize surface fluctuations; and the no-spill objective is defined as a penalty term based on an indicator function to eliminate the risk of spillage. Subsequently, a dynamic weighted summation structure is used to integrate the multiple objectives. The weight coefficients of each objective are automatically calculated using the analytic hierarchy process and real-time operating conditions. The initial stage emphasizes efficiency weights, while the final stage emphasizes accuracy weights. Finally, an integral multi-objective optimization function in the time domain is formed, satisfying the following relationship: in, Indicates minimization. To optimize variables, For multi-objective optimization functions, To optimize the start time of the process, To optimize the process termination time, For index variables of multiple optimization objectives, For the first An optimization objective at time... The weighting coefficients, For the first A quantitative index function for an optimization objective. For state variables, It is a tiny time unit.

[0046] In this embodiment, the optimal control strategy is solved in real time based on the predicted evolution trend of the bulk material loading state, combined with a multi-objective optimization function. The optimization solution employs a constrained multi-objective particle swarm optimization algorithm to find the optimal combination of control parameters that optimizes the objective function while satisfying equipment physical limitations and operational safety constraints. The generation of feeding control commands uses a fuzzy adaptive PID control algorithm, combined with a multi-objective optimization algorithm, to dynamically adjust the feeder speed and hopper opening based on the deviation between the predicted and target states. Control commands include parameters such as the target speed value, acceleration curve, and runtime. Vehicle movement commands are generated based on a material surface flatness optimization strategy. When an off-center load or peak load is predicted, the optimal vehicle movement distance and direction are obtained by combining the multi-objective optimization function, generating a command sequence including parameters such as movement distance, movement speed, and pause time. All control commands are sent to the execution device in real time via Ethernet and are equipped with command verification and feedback confirmation mechanisms. Simultaneously, a command execution effect evaluation system is established to monitor the deviation between the actual effect after command execution and the expected target in real time, and dynamically adjust the generation strategy of subsequent commands accordingly, forming a complete closed-loop control circuit.

[0047] S5. Construct an online optimization mechanism to dynamically optimize the silo bulk material loading control command to obtain loading optimization control parameters, thereby realizing intelligent loading of silo bulk materials.

[0048] In this embodiment, during the execution of the bulk material loading control command from the silo, the bulk material loading performance evaluation index is calculated in real time, and the real-time loading disturbance characteristics are monitored and quantified simultaneously. A parameter optimization knowledge base and control rule base are established based on historical loading data, and an online tuning mechanism is established in combination with the real-time loading disturbance characteristics.

[0049] Specifically, during the execution of loading control commands, real-time loading information is acquired, and four core performance evaluation indicators for bulk material loading are calculated based on this information: the loading rate stability indicator is calculated using the feeder speed fluctuation variance; the volume estimation accuracy indicator is calculated using the relative error between the laser scanning volume and the weighing volume; the material surface flatness indicator is calculated using the standard deviation of the material surface height distribution; and the equipment operational health indicator is calculated using vibration spectrum analysis and temperature monitoring data. Simultaneously, multi-source disturbance observers are used to monitor and quantify real-time loading disturbance characteristics, including quantifying the humidity change index through material image texture analysis, analyzing dust concentration changes through millimeter-wave radar echo characteristics, and monitoring material flowability changes through motor load current, thus forming real-time loading disturbance characteristics.

[0050] Furthermore, historical loading data is collected, including multi-dimensional data such as environmental parameters, equipment status, control parameters, and operational results. Spatiotemporal data mining algorithms are employed to analyze optimal operating modes under different working conditions, constructing a parameter optimization knowledge base that stores optimal parameter configuration combinations for various typical operating conditions. Simultaneously, a control rule base is built by combining expert operational experience and data-driven analysis, including equipment adjustment rules and process optimization rules. Finally, by integrating real-time loading disturbance characteristics, the parameter optimization knowledge base, and the control rule base, an online optimization mechanism with online learning and adaptive adjustment capabilities is established.

[0051] In an optional embodiment, taking bulk density as an example, the incremental learning and updating of the parameter optimization knowledge base in the online tuning mechanism satisfies the following relationship: in, The updated material density value. The learning rate factor, The material density value before the update. This is the total weight after loading. For vehicle tare weight, This is an estimated average volume of the loaded bulk material.

[0052] In this embodiment, the optimization strategy matching result is obtained based on the real-time loading disturbance characteristics and parameter optimization knowledge base, and the control parameter adjustment amount is obtained by combining the control rule base; the optimal bulk loading performance evaluation index is used as the adjustment target, and the loading optimization control parameters are obtained by combining the control parameter adjustment amount and the silo bulk loading control command, so as to realize intelligent loading of silo bulk materials.

[0053] Specifically, the real-time loading disturbance features detected in real-time are extracted and standardized. Then, a similarity matching algorithm is used to find the most similar historical cases in the parameter optimization knowledge base as the optimization strategy matching results, and the corresponding optimization parameter benchmark values ​​are obtained to obtain the control parameter adjustment amounts. Simultaneously, the deviation between the current performance indicators and the target values, as well as their changing trends, are input into the control rule base for inference calculation. A fuzzy inference mechanism is used to generate parameter correction amounts based on the current system state. Finally, the control parameter adjustment amounts based on historical case matching and the parameter correction amounts based on rule inference are fused and optimized to generate a control parameter adjustment scheme for the silo bulk material loading control command. This obtains loading optimization control parameters, including but not limited to key parameters such as feeder speed adjustment, gate opening adjustment, and vehicle movement adjustment, enabling intelligent loading of bulk materials from the silo.

[0054] The above parameter corrections satisfy the following relationship: in, For parameter correction amount, This represents the total number of fuzzy rules. For the index variable of the fuzzy rule, For the first The activation degree of a fuzzy rule. For the first The output value vector corresponding to each fuzzy rule.

[0055] In one optional embodiment, the tare weight (empty weight) of the container truck is obtained, and the loading weight limit is determined to calculate the weight of the material to be loaded. Then, a batching scale is used to precisely control the amount of material added. During the loading process, the batching scale accurately batches the material according to the calculated loading amount, ensuring that the target weight is achieved in one loading, realizing "accurate loading on the first try." This eliminates the need to add or subtract material from the vehicle's load later, improves loading efficiency, and completes the quantitative loading of bulk materials.

[0056] Please see Figure 2 In one optional embodiment, the present invention provides an intelligent bulk material loading system for silos. The system includes an input device, an output device, a processor, and a memory, all interconnected. The memory stores a computer program comprising program instructions. The processor is configured to invoke the program instructions to execute specific steps as described in the relevant embodiments of the intelligent bulk material loading method for silos provided by the present invention. The intelligent bulk material loading system for silos provided by the present invention has a complete and stable structure, enhancing the overall applicability and practical application capability of the present invention.

[0057] In summary, the present invention provides an intelligent loading method and system for bulk materials in silos. It collects loading environment data in real time through a multi-dimensional sensing array, accurately analyzes the dynamic stacking volume of bulk materials based on a gradient compensation algorithm, and constructs a model of the bulk material stacking state in the silo to predict the loading status. It employs multi-objective optimization to generate control commands and establishes an online learning-capable tuning mechanism. Through a parameter knowledge base and rule base, it dynamically optimizes control parameters, achieving fully automated, high-precision, and self-evolving intelligent loading. This effectively solves the pain points of insufficient accuracy, low efficiency, and reliance on manual labor in traditional loading methods. The method of this invention is easy to understand, computationally simple, requires less workload, and is convenient for engineering applications, providing a theoretical foundation and technical support for the further development of port automation technology.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for intelligent loading of bulk materials from silos, characterized in that, Includes the following steps: Obtain silo bulk material loading information, and obtain silo bulk material-loading vehicle matching results based on the silo bulk material loading information; A multi-dimensional loading information sensing array is established, and a multi-dimensional real-time loading sensing dataset is obtained by combining the silo bulk material-loading vehicle matching results. Based on the multi-dimensional real-time loading perception dataset, the distribution of loaded bulk materials is analyzed to obtain the dynamic stacking volume of the bulk materials; Based on the dynamic stacking volume of the bulk materials, a bulk material stacking state model of the wagon is constructed, and the bulk material loading control command of the silo is obtained according to the bulk material stacking state model of the wagon. An online optimization mechanism is established to dynamically optimize the silo bulk material loading control commands to obtain loading optimization control parameters, thereby realizing intelligent loading of silo bulk materials.

2. The intelligent loading method for bulk materials from silos according to claim 1, characterized in that, The step of obtaining silo bulk material loading information and obtaining silo bulk material-loading vehicle matching results based on the silo bulk material loading information includes: The loading vehicle information, the planned loading quantity of bulk materials, and the type of bulk materials are obtained as the bulk material loading information of the silo. Establish business logic verification rules, and compare and verify the loading information of the bulk materials in the silo according to the business logic verification rules to obtain the loading verification result; The silo bulk material-loading vehicle matching result is obtained based on the loading verification result. The silo bulk material-loading vehicle matching result includes the target loading silo number, the allowed loading time window, and the initial loading control parameters.

3. The intelligent loading method for bulk materials from silos according to claim 2, characterized in that, The establishment of business logic verification rules includes: The vehicle whitelist verification rules, the electronic delivery note validity status verification rules, and the material consistency verification rules are constructed as the business logic verification rules.

4. The intelligent loading method for bulk materials from silos according to claim 1, characterized in that, The establishment of a multi-dimensional loading information sensing array, combined with the silo bulk material-loading vehicle matching results, yields a multi-dimensional real-time loading sensing dataset, including: The multi-dimensional loading information sensing array is established based on non-repetitive scanning lidar, millimeter-wave radar level gauge and industrial camera; During the loading process of the silo bulk material-loading vehicle matching result, the non-repeating scanning lidar is used to acquire three-dimensional point cloud data, the millimeter-wave radar level gauge is used to obtain the height data below the hopper, and the industrial camera is used to obtain auxiliary two-dimensional image data. The multi-dimensional real-time loading perception dataset is established based on the three-dimensional point cloud data, the height data below the hopper, and the auxiliary two-dimensional image data.

5. The intelligent loading method for bulk materials from silos according to claim 1, characterized in that, The step of obtaining the dynamic stacking volume of bulk materials by analyzing the distribution of the loaded bulk materials based on the multi-dimensional real-time loading perception dataset includes: A three-dimensional mesh model of the bulk material accumulation surface is constructed based on the multi-dimensional real-time loading perception dataset. The three-dimensional mesh model of the bulk material accumulation surface is projected onto the coordinate system of the bottom surface of the loading vehicle to construct the bulk material surface height function; A gradient compensation term is established, and the dynamic stacking volume of the bulk material is calculated by combining the surface height function of the bulk material and the gradient compensation term.

6. The intelligent loading method for bulk materials from silos according to claim 5, characterized in that, The step of establishing a gradient compensation term, and calculating the dynamic stacking volume of the bulk material by combining the surface height function of the bulk material and the gradient compensation term, includes: Based on the integration calculation of the bulk material surface height function, the basic volume estimate of the bulk material stack is obtained; Obtain the height change rate of each point in the surface height function of the bulk material, and obtain surface gradient information based on the height change rate; The gradient compensation volume value is obtained by performing a second integral operation based on the surface gradient information, and the gradient compensation volume value is used as the gradient compensation term. The estimated base volume is superimposed with the gradient compensation volume to obtain the dynamic stacking volume of the bulk material.

7. The intelligent loading method for bulk materials from silos according to claim 1, characterized in that, The process of constructing a bulk material stacking state model for the wagon based on the dynamic stacking volume of the bulk materials, and obtaining silo bulk material loading control commands based on the bulk material stacking state model, includes: The dynamic stacking volume of bulk materials is used as the core model data source, and the multi-dimensional real-time loading perception dataset is combined to construct the bulk material stacking state model of the carriage. Based on the aforementioned bulk material stacking state model of the car body, the bulk material stacking state of the car body is forward-predicted to obtain the predicted bulk material loading state; With loading accuracy, loading efficiency, uniformity of bulk material accumulation, and no spillage as multiple optimization objectives, a multi-objective optimization function is constructed based on these multiple optimization objectives; The silo bulk material loading control command is obtained by solving the multi-objective optimization function based on the predicted bulk material loading status. The silo bulk material loading control command includes a feeding control command and a vehicle movement command.

8. The intelligent loading method for bulk materials from silos according to claim 1, characterized in that, The aforementioned online optimization mechanism includes: During the execution of the bulk material loading control command in the silo, the bulk material loading performance evaluation index is calculated in real time, and the real-time loading disturbance characteristics are monitored and quantified simultaneously. A parameter optimization knowledge base and a control rule base are established based on historical loading data, and the online tuning mechanism is established in combination with the real-time loading disturbance characteristics.

9. The intelligent loading method for bulk materials from silos according to claim 8, characterized in that, The process of dynamically optimizing the loading control commands for bulk material loading from the silo to obtain optimized loading control parameters, thereby achieving intelligent loading of bulk material from the silo, includes: Based on the real-time loading disturbance characteristics and the parameter optimization knowledge base, the optimization strategy matching result is obtained, and the control parameter adjustment amount is obtained by combining the control rule base. The optimal loading performance evaluation index of bulk materials is used as the adjustment target. The loading optimization control parameters are obtained by combining the adjustment amount of the control parameters and the silo bulk material loading control command, so as to realize intelligent loading of bulk materials in silos.

10. A smart loading system for bulk materials from silos, characterized in that, The system includes an input device, an output device, a processor, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the intelligent loading method for bulk materials in silos as described in any one of claims 1-9.

Images