An accurate positioning control system and method for an electro-permanent magnet lifter

Abstract

The application provides an accurate positioning control system and method of an electric permanent magnet lifter, controls the electric permanent magnet lifter to perform initialization self-checking; acquires multi-modal sensing data, determines a magnetic circuit adsorption characteristic vector according to a magnetic field distribution signal, performs state prediction based on the magnetic circuit adsorption characteristic vector, obtains a magnet displacement change trend and an adsorption force change trend; performs positioning pin hole identification through a target container top image, performs visual deviation calibration according to the identified positioning pin hole point set and environmental disturbance data, and obtains a positioning drift vector; performs servo active compensation using the positioning drift vector, the magnet displacement change trend and the adsorption force change trend, generates a positioning motion trajectory, and drives the electric permanent magnet lifter to perform accurate pin hole positioning based on the positioning motion trajectory. The technical scheme provided by the application can fuse multi-modal sensing perception, predict a magnetic circuit state, and thus perform servo active compensation on a positioning trajectory, so as to realize accurate positioning control of the electric permanent magnet lifter.

Description

Technical Field

[0001] This application relates to the field of crane positioning control technology, and more specifically, to a precision positioning control system and method for an electro-permanent magnet crane. Background Technology

[0002] Port container handling is a core link in the modern logistics chain, and its efficiency and safety directly affect the port's operational benefits. Currently, electro-permanent magnet lifters, which rely on energized permanent magnets to generate attraction for lifting heavy objects, offer advantages such as low energy consumption, reliable structure, and no loss of magnetism even when power is off. They are widely used in material handling scenarios such as steel and port container handling. Especially with the rapid growth in demand for intelligent and remote container lifting in ports, electro-permanent magnet lifters are becoming an important development direction, replacing traditional mechanical locks and electromagnetic lifting devices.

[0003] However, in terms of magnetic adsorption process control, existing electro-permanent magnet cranes lack the ability to perceive and predict the magnetic circuit state in real time. Traditional systems cannot predict the magnetic circuit closure state and the trend of adsorption force establishment during descent, leading to unstable adsorption or vacuuming due to premature or delayed magnetic activation. Furthermore, in terms of dynamic environmental adaptability, traditional systems cannot effectively compensate for real-time alignment drift caused by crane structural swaying, wind disturbances, and slight changes in container posture, resulting in insufficient alignment accuracy between the guide rod and pin hole, frequently causing collisions or requiring repeated adjustments, severely affecting operational smoothness. Therefore, how to integrate multimodal sensing and predict the magnetic circuit state to perform servo active compensation for the positioning trajectory, thereby achieving precise positioning control of electro-permanent magnet cranes, is a challenge facing the industry. Summary of the Invention

[0004] This application provides a precise positioning control system and method for an electro-permanent magnet lifter, which can integrate multi-modal sensing and predict the magnetic circuit state to perform servo active compensation for the positioning trajectory, thereby achieving precise positioning control of the electro-permanent magnet lifter.

[0005] In a first aspect, this application provides a precise positioning control method for an electro-permanent magnet lifter, comprising the following steps: After the port container crane is started, the control electro-permanent magnet lifter performs an initial self-test and moves the electro-permanent magnet lifter to the airspace above the target container. Multimodal sensing data of an electro-permanent magnet lifter is acquired. The magnetic circuit adsorption feature vector is determined based on the magnetic field distribution signal in the multimodal sensing data. State prediction is performed based on the magnetic circuit adsorption feature vector, thereby obtaining the magnet displacement change trend and the attraction force change trend. The positioning pin holes are identified by using the top image of the target container in the multimodal sensing data. Then, visual deviation calibration is performed based on the identified positioning pin hole point set and the environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. The descent trajectory of the electro-permanent magnet lifter is actively compensated by the alignment drift vector, the magnet displacement change trend, and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter. Based on the positioning motion trajectory, the electro-permanent magnet lifter is driven to perform precise pin hole positioning.

[0006] In some embodiments, controlling the electro-permanent magnet lifter to perform an initial self-test specifically includes: Perform zero-point regression on the gear position sensor, record the encoder zero position, perform zero-point drift measurement and save the bias for all magnetic field sensors, force sensors, vision sensors and anemometers, and read and verify the health status of each sensor.

[0007] In some embodiments, the multimodal sensing data includes magnetic field distribution signals, images of the top of the target container, and environmental disturbance data.

[0008] In some embodiments, determining the magnetic circuit adsorption feature vector based on the magnetic field distribution signal in the multimodal sensing data specifically includes: An initial magnetic field distribution matrix is ​​constructed using the magnetic field distribution signals from the multimodal sensing data; The initial magnetic field distribution matrix is ​​filtered, denoised, and temperature compensated to obtain a standardized magnetic field distribution map. The magnetic field uniformity index is obtained by extracting the magnetic field uniformity from the standardized magnetic field distribution map. The load magnetic flux is obtained by estimating the magnetic flux through the standardized magnetic field distribution map. Edge gradient extraction is performed on the standardized magnetic field distribution map to obtain the magnetic field leakage coefficient; A magnetic circuit adsorption feature vector is constructed based on the magnetic field uniformity index, the load magnetic flux, and the magnetic field line leakage coefficient.

[0009] In some embodiments, the state prediction based on the magnetic circuit adsorption feature vector, and the resulting magnet displacement change trend and attraction force change trend, specifically include: Obtain a pre-trained prediction model based on a long short-term memory network structure; The magnetic circuit adsorption feature vector is input into the prediction model based on the long short-term memory network structure for state prediction, thereby obtaining the magnet displacement change trend and the attraction force change trend.

[0010] In some embodiments, identifying positioning pin holes using the target container top image from the multimodal sensing data specifically includes: The top image of the target container in the multimodal sensing data is preprocessed; Key regions are located in the preprocessed top image of the target container to obtain the bounding boxes of all container corner pieces in the top image of the target container; Inside the bounding box of each located container corner piece, a detection algorithm based on circular Hough transform is used to accurately locate the precise pixel center position of the positioning pin hole. The precise pixel center position is converted into three-dimensional spatial coordinates in the spreader coordinate system, thereby identifying the set of positioning pin holes on the top of the target container.

[0011] In some embodiments, the servo-active compensation of the descent trajectory of the electro-permanent magnet lifter using the alignment drift vector, the magnet displacement change trend, and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter, specifically includes: The alignment drift vector, the magnet displacement change trend, and the attraction force change trend are integrated into a comprehensive state evaluation matrix; The descent trajectory of the electro-permanent magnet lifter is dynamically corrected based on the comprehensive state evaluation matrix to obtain the corrected trajectory. The feasibility of the corrected trajectory is verified and smoothed based on preset safety constraints, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter.

[0012] Secondly, this application provides a precision positioning control system for an electro-permanent magnet lifter, used to execute a precision positioning control method for an electro-permanent magnet lifter, including: The initial self-test module is used to control the electro-permanent magnet lifter to perform an initial self-test after the port container crane is started, and to move the electro-permanent magnet lifter to the target container. The state prediction module is used to acquire multimodal sensing data of the electro-permanent magnet lifter, determine the magnetic circuit adsorption feature vector based on the magnetic field distribution signal in the multimodal sensing data, perform state prediction based on the magnetic circuit adsorption feature vector, and then obtain the magnet displacement change trend and the attraction force change trend. The drift calibration module is used to identify the positioning pin holes through the top image of the target container in the multimodal sensing data, and then perform visual deviation calibration based on the identified positioning pin hole point set and the environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. The positioning control module is used to perform servo active compensation on the descent trajectory of the electro-permanent magnet lifter using the alignment drift vector, the magnet displacement change trend, and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter, and driving the electro-permanent magnet lifter to perform precise pin hole positioning based on the positioning motion trajectory.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described precise positioning control method for an electro-permanent magnet lifter.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described precise positioning control method for an electro-permanent magnet lifter.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: This application provides a precise positioning control system and method for an electro-permanent magnet crane. After the port container crane is started, the electro-permanent magnet crane is controlled to perform an initial self-test and moved to the airspace above the target container. Multimodal sensing data of the electro-permanent magnet crane is acquired. A magnetic circuit adsorption feature vector is determined based on the magnetic field distribution signal in the multimodal sensing data. State prediction is performed based on the magnetic circuit adsorption feature vector to obtain the magnet displacement change trend and the attraction force change trend. Positioning pin holes are identified using the top image of the target container in the multimodal sensing data. Visual deviation calibration is performed based on the identified positioning pin hole point set and the environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. The alignment drift vector, the magnet displacement change trend, and the attraction force change trend are used to perform servo active compensation on the descent trajectory of the electro-permanent magnet crane to generate the positioning motion trajectory of the electro-permanent magnet crane. Based on the positioning motion trajectory, the electro-permanent magnet crane is driven to perform precise pin hole positioning.

[0016] Therefore, this application firstly acquires multimodal sensing data of the electro-permanent magnet lifter and constructs a magnetic circuit adsorption feature vector based on the magnetic field distribution signal. Then, it combines this feature vector to predict the magnet displacement change trend and the attraction force change trend, which can achieve in-depth characterization and forward-looking judgment of the permanent magnet adsorption process. The magnetic coupling strength, adsorption uniformity, and magnetic leakage can be quantitatively evaluated through the magnetic circuit adsorption feature vector. Secondly, by identifying the positioning pin holes in the top image of the target container, the precise geometric features of the container can be captured, and the detection of alignment error can reach the millimeter level accuracy. Furthermore, by combining the identified pin hole point set with environmental disturbance data for visual deviation calibration, the actual deviation between the electro-permanent magnet lifter and the container in spatial position and attitude can be accurately quantified. Finally, by using the alignment drift vector, the magnet displacement change trend, and the attraction force change trend to perform servo active compensation on the descent trajectory of the electro-permanent magnet lifter and generating a positioning motion trajectory to drive the magnet to accurately position the pin holes, real-time comprehensive correction of static error, dynamic drift, and magnetic force change can be achieved to realize precise positioning control of the electro-permanent magnet lifter.

[0017] In summary, the technical solution adopted in this application can integrate multimodal sensing and predict the magnetic circuit state, thereby performing servo active compensation for the positioning trajectory to achieve precise positioning control of the electro-permanent magnet lifter. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this embodiment of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is an exemplary flowchart of a precise positioning control method for an electro-permanent magnet lifter according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the determination of magnetic circuit adsorption feature vectors according to some embodiments of this application; Figure 3 This is a schematic diagram of the structure of a precision positioning control system for an electro-permanent magnet lifter according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a computer device for implementing a precise positioning control method for an electro-permanent magnet lifter, according to some embodiments of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0021] This application provides a precise positioning control system and method for an electro-permanent magnet crane. The core of this system involves controlling the electro-permanent magnet crane to perform an initial self-check after the port container crane starts, and then moving the electro-permanent magnet crane above the target container. Multimodal sensing data of the electro-permanent magnet crane is acquired, and a magnetic circuit adsorption feature vector is determined based on the magnetic field distribution signal in the multimodal sensing data. State prediction is performed based on the magnetic circuit adsorption feature vector to obtain the magnet displacement change trend and the attraction force change trend. Positioning pin holes are identified using the top image of the target container in the multimodal sensing data. Visual deviation calibration is then performed based on the identified positioning pin hole point set and the environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. The alignment drift vector, the magnet displacement change trend, and the attraction force change trend are used to perform servo active compensation on the descent trajectory of the electro-permanent magnet crane, thereby generating the positioning motion trajectory of the electro-permanent magnet crane. Based on this positioning motion trajectory, the electro-permanent magnet crane is driven to perform precise pin hole positioning. The above scheme can integrate multimodal sensing and predict the magnetic circuit state, thereby performing servo active compensation for the positioning trajectory to achieve precise positioning control of the electro-permanent magnet lifter.

[0022] To better understand the above technical solutions, a detailed description of the technical solutions will be provided below in conjunction with the accompanying drawings and specific embodiments. (Refer to...) Figure 1 The figure is an exemplary flowchart of a precise positioning control method for an electro-permanent magnet lifter according to some embodiments of this application. The figure mainly includes the following steps: In step S101, after the port container crane is started, the electro-permanent magnet jack is controlled to perform an initial self-test and then moved to the airspace above the target container.

[0023] In some embodiments, the control of the electro-permanent magnet lifter to perform initial self-test can be carried out in the following manner: Perform zero-point regression on the gear position sensor, record the encoder zero position, perform zero-point drift measurement and save the bias for all magnetic field sensors, force sensors, vision sensors and anemometers, and read and verify the health status of each sensor.

[0024] In practical implementation, after the port container crane starts, the system first performs an initialization self-check operation on the electro-permanent magnet lifter, which involves zero-point return operation on the gear position sensor, driving the gear to slowly move to the mechanical limit position, triggering the limit switch to confirm the zero point, and recording the encoder zero-point value; zero-point drift measurement is performed on all magnetic field sensors, acquiring the magnetic field readings of each measuring point of the Hall array with the magnet fully retracted, and performing temperature compensation based on the ambient temperature, saving the measured bias to the control system to ensure the accuracy of subsequent magnetic field measurements; zero-point drift measurement is performed on the force sensor, acquiring the force sensor output value under no-load conditions, recording the zero-point drift, and saving the calibration parameters. The system provides a reference for force feedback when the spreader contacts the container; it calibrates the vision sensors, activates the camera and image processing module, acquires standard reference images, and checks the camera's exposure, focal length, and image distortion correction functions to ensure subsequent target recognition accuracy; it performs zero-point drift measurement and health status verification on the anemometer and other environmental sensors, acquires static wind speed or environmental baseline data, records the offset, and checks the sensor communication and data update status to confirm normal sensor operation; it integrates the health status of each sensor, reads the self-test flag, communication status, and response delay of each sensor, and marks the initialization self-test as complete if all are normal; otherwise, it outputs a fault warning, requiring manual intervention or a re-self-test. After completing the self-test, the control system drives the crane trolley, crane carriage, and spreader drive mechanism to move the electro-permanent magnet lifter above the target container according to remote operator instructions or preset trajectories.

[0025] In step S102, multi-modal sensing data of the electro-permanent magnet lifter is acquired, magnetic circuit adsorption feature vector is determined based on the magnetic field distribution signal in the multi-modal sensing data, state prediction is performed based on the magnetic circuit adsorption feature vector, and then the magnet displacement change trend and the attraction force change trend are obtained.

[0026] It should be noted that the multimodal sensing data includes magnetic field distribution signals, images of the top of the target container, and environmental disturbance data. Specifically, firstly, the Hall array sensor of the electro-permanent magnet crane is activated. When the gear is in its current position, it collects real-time data on the magnetic induction intensity around the magnet, and simultaneously reads information from the gear displacement encoder to record the magnet's extension and contraction state, thus obtaining the magnetic field distribution signal. Then, a binocular or multi-view vision system at the bottom of the spreader can simultaneously acquire images of the top of the target container. During image acquisition, the camera undergoes automatic exposure, focus adjustment, and distortion correction to ensure high-resolution, low-noise images. Finally, disturbance parameters such as wind speed, wind direction, boom tilt angle, and vibration can be collected in real-time using boom attitude sensors, wind speed sensors, and other environmental monitoring equipment, thus obtaining environmental disturbance data.

[0027] Preferably, in some embodiments, reference is made to Figure 2As shown, this figure is an exemplary flowchart of determining the magnetic circuit adsorption feature vector according to some embodiments of this application. In this embodiment, determining the magnetic circuit adsorption feature vector based on the magnetic field distribution signal in the multimodal sensing data can be achieved by the following steps: In step S1021, an initial magnetic field distribution matrix is ​​constructed using the magnetic field distribution signal in the multimodal sensing data; In step S1022, the initial magnetic field distribution matrix is ​​subjected to filtering, noise reduction, and temperature compensation processing to obtain a standardized magnetic field distribution map; In step S1023, magnetic field uniformity is extracted from the standardized magnetic field distribution map to obtain a magnetic field uniformity index. In step S1024, magnetic flux estimation is performed on the standardized magnetic field distribution map to obtain the load magnetic flux; In step S1025, edge gradient extraction is performed on the standardized magnetic field distribution map to obtain the magnetic field leakage coefficient; In step S1026, a magnetic circuit adsorption feature vector is constructed based on the magnetic field uniformity index, the load magnetic flux, and the magnetic field line leakage coefficient.

[0028] In practical implementation, firstly, an initial magnetic field distribution matrix can be constructed using the magnetic field distribution signal from the multimodal sensing data. This initial matrix is ​​formed by combining the magnetic field distribution signal from the multimodal sensing data with information from the gear displacement encoder, mapping the magnetic field values ​​of each measuring point in the magnetic field distribution signal to the lifting coordinate system. Then, the initial magnetic field distribution matrix can be filtered, denoised, and temperature compensated. This involves applying medium-range filtering, wavelet denoising, or low-pass filtering to remove impulse noise and high-frequency interference, and using temperature sensor data for temperature compensation correction, thereby obtaining a standardized magnetic field distribution map. Secondly, the magnetic field uniformity index can be obtained by extracting the magnetic field uniformity from the standardized magnetic field distribution map. This index is used to characterize the uniformity of the magnetic field distribution on the adsorption working surface of the electro-permanent magnet lifter, reflecting whether the magnetic force acts uniformly on the surface of the container steel plate. When the magnet is aligned with the steel plate for adsorption, the more uniform the magnetic field in the adsorption area, the more fully the attraction force is transmitted and the higher the adsorption stability. If the magnetic field distribution is uneven, there may be risks of biased attraction, local demagnetization, or lateral slippage. In practice, the mean and standard deviation of the magnetic induction intensity of all measuring points in the standardized magnetic field distribution map can be calculated, and the ratio of the standard deviation to the mean can be subtracted from 1. The calculation result can be used as the magnetic field uniformity index.

[0029] In addition, in practical implementation, magnetic flux estimation can be performed on the standardized magnetic field distribution map to obtain the load magnetic flux. This load magnetic flux represents the effective magnetic flux of the permanent magnet passing through the adsorption surface of the steel plate under the current adsorption conditions, and is used to characterize the actual adsorption capacity output by the permanent magnet. The larger the load magnetic flux, the stronger the attraction; a decrease in the load magnetic flux indicates that the magnetic field path is blocked, the distance is increased, or the magnetic circuit is not fully closed. In practical implementation, the magnetic induction intensity can be integrated over the effective adsorption region to calculate the load magnetic flux. Then, edge gradient extraction can be performed on the standardized magnetic field distribution map to obtain the magnetic field leakage coefficient. This magnetic field leakage coefficient represents the degree of magnetic field leakage outside the effective adsorption region, and is used to reflect the magnetic field concentration and the quality of magnetic circuit closure. That is, the more leakage, the more wasted the magnetic field adsorption force and the unstable adsorption; the less leakage, the more concentrated the magnetic field force and the higher the adsorption efficiency. In practical implementation, the edges of the standardized magnetic field distribution map can be extracted using the Sobel operator and the gradient mean can be calculated, and the calculation result can be used as the magnetic field leakage coefficient. Finally, the eigenvector composed of the magnetic field uniformity index, the load magnetic flux, and the magnetic line leakage coefficient can be used as the magnetic circuit adsorption eigenvector.

[0030] Based on the magnetic circuit adsorption feature vector, state prediction is performed to obtain the magnet displacement change trend and the attraction force change trend, specifically including: Obtain a pre-trained prediction model based on a long short-term memory network structure; The magnetic circuit adsorption feature vector is input into the prediction model based on the long short-term memory network structure for state prediction, thereby obtaining the magnet displacement change trend and the attraction force change trend.

[0031] In practice, firstly, a pre-trained prediction model based on a long short-term memory network structure can be obtained. This involves reading the pre-trained prediction model based on a long short-term memory network structure from the device's local or cloud secure storage and completing model initialization and version verification to ensure that the model parameters are consistent with the operating environment. Then, the magnetic circuit adsorption feature vector can be input into the prediction model based on the long short-term memory network structure for state prediction. That is, the real-time generated magnetic circuit adsorption feature vector is organized into a time-series sample sequence according to the time window, and combined with the feature vectors of several historical moments to form the model input sequence. Then, the sequence is normalized and imputed for missing values ​​in the same way as during training to ensure that the input scale and statistical characteristics match. The preprocessed time-series input is sent into the long short-term memory network for forward inference. The network automatically captures the dynamic correlation of the feature vector in the time-series context and outputs the state prediction results for several future control cycles. Specifically, it includes the short-term change sequence of magnet displacement and the short-term trend sequence of attraction force change, that is, the magnet displacement change trend and the attraction force change trend. Among them, the magnet displacement change trend refers to the dynamic trend information of the positional offset of the magnet assembly of the electro-permanent magnet lifter changing with time during the vertical approach to the surface of the target steel structure. The attraction force change trend refers to the dynamic trend information of the magnetic adsorption force between the magnet and the container steel plate during the adsorption process of the electro-permanent magnet lifter changing with time and distance.

[0032] It should be noted that by acquiring multimodal sensing data of the electro-permanent magnet lifter and constructing a magnetic circuit adsorption feature vector based on the magnetic field distribution signal, and then combining this feature vector to predict the trend of magnet displacement change and attraction force change, a deep characterization and forward-looking judgment of the permanent magnet adsorption process can be achieved. The magnetic coupling strength, adsorption uniformity and magnetic leakage can be quantitatively evaluated through the magnetic circuit adsorption feature vector, and the trend of magnet displacement change and attraction force change can be known in advance based on the prediction model, so as to predict the stability and alignment misalignment risk of the magnetic attraction establishment process in advance.

[0033] In step S103, the positioning pin holes are identified using the top image of the target container in the multimodal sensing data. Then, visual deviation calibration is performed based on the identified positioning pin hole point set and the environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane.

[0034] In some embodiments, the identification of positioning pin holes using the target container top image in the multimodal sensing data can be specifically carried out in the following manner: The top image of the target container in the multimodal sensing data is preprocessed; Key regions are located in the preprocessed top image of the target container to obtain the bounding boxes of all container corner pieces in the top image of the target container; Inside the bounding box of each located container corner piece, a detection algorithm based on circular Hough transform is used to accurately locate the precise pixel center position of the positioning pin hole. The precise pixel center position is converted into three-dimensional spatial coordinates in the spreader coordinate system, thereby identifying the set of positioning pin holes on the top of the target container.

[0035] In practice, the first step is to preprocess the target container top image from the multimodal sensing data. This involves distortion and parallax correction, followed by adaptive exposure adjustment, denoising, and contrast enhancement to suppress shadows and highlights and improve the signal-to-noise ratio of corner pieces and pinhole edges. Then, key region localization is performed on the preprocessed target container top image. This involves using a multi-scale candidate region extraction strategy, combined with morphological filtering and deep learning-based semantic segmentation or traditional corner and edge clustering methods to quickly locate the bounding boxes of the four corner pieces on the top of the container. Irrelevant background and overlapping false detections are removed, and the bounding box coordinates of all container corner pieces in the target container top image are output. Secondly, within the bounding box of each located container corner piece, a detection algorithm based on circular Hough transform is used to accurately locate the precise pixel center position of the positioning pin hole. That is, circular Hough transform is applied to detect the circular contour within the bounding box of each corner piece. First, a coarse-scale Hough transform is used to obtain the candidate circle center and radius. Then, the candidate results are refined at the sub-pixel level (e.g., least squares optimization based on gradient direction and edge fitting) to determine the precise pixel center position of the pin hole. At the same time, false detections caused by out-of-plane or occlusion are eliminated by stereo parallax or depth constraints. Finally, the precise pixel center position can be converted into three-dimensional spatial coordinates in the spreader coordinate system. That is, the depth information is recovered at the center of each precise pixel using stereo matching. The pixel coordinates and depth values ​​are converted into three-dimensional spatial coordinates in the spreader reference coordinate system by combining the camera's intrinsic and extrinsic parameters and the spreader calibration matrix. The coordinate offset is corrected by combining gear displacement and boom posture. Thus, the geometric consistency of the three-dimensional coordinate point set identified in all corner pieces is checked (e.g., based on the four-point coplanarity and dimensional constraints of the box template). Abnormal points are removed and missing points are filled. Finally, the set of positioning pin holes on the top of the target container is output after verification.

[0036] In some embodiments, visual deviation calibration is performed based on the identified set of positioning pin holes and environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. Specifically, this can be achieved in the following manner: The identified positioning pin hole point set is spatiotemporally aligned and fused with the real-time tilt angle data and the environmental disturbance data in the multimodal sensing data. Based on the Kalman filter algorithm, the fused data is optimally estimated to predict the additional drift amount that will be generated in the next moment due to the continuous effect of environmental disturbance. A dynamic deviation compensation model is established, and the spatial distribution geometric characteristics of the pin hole point set and the point set drift mode after being affected by environmental disturbances are analyzed based on the dynamic deviation compensation model, so as to obtain the static pose deviation and dynamic drift components. The additional drift amount, the static pose deviation, and the dynamic drift component are vectorized to obtain the alignment drift vector between the target container and the electro-permanent magnet crane.

[0037] In practical implementation, firstly, the identified positioning pin hole point set can be spatiotemporally aligned and fused with real-time tilt data and environmental disturbance data from multimodal sensing data. That is, precise alignment is performed according to timestamps, and all observations are uniformly transformed to the same reference coordinate system (spreader reference system). Then, the fused data can be optimally estimated based on the Kalman filter algorithm. That is, taking the alignment drift direction as the state variable, a state-space model containing system kinematic prediction terms and environmental disturbance input terms is established. The Kalman filter is used to recursively optimize the fused data to obtain the current optimal pose estimate and state covariance. Based on this filter, a short-term prediction of the additional drift that may occur under continuous wind load and swaying action at the next moment can be made. Secondly, a dynamic deviation compensation model can be established. This model performs spatial geometric analysis on the identified pin hole point set in the spreader reference coordinate system. By comparing the actual position of the point set with the positional relationship of the ideal container geometric template, the overall translational and rotational deviations are calculated, thereby obtaining the static pose deviation. Furthermore, for each pin hole point, the three-dimensional coordinate sequence at continuous sampling times is analyzed, combined with environmental disturbance data collected by the spreader tilt sensor, wind speed sensor, and acceleration sensor, to determine the dynamic change pattern of the point set over time, such as slight swaying, vibration, or offset due to wind load. By filtering, frequency domain analysis, or mode decomposition of the time series, the main dynamic drift components are extracted, obtaining the dynamic drift components of the point set under environmental disturbance. Finally, the additional drift, static pose deviation, and dynamic drift components are superimposed in a vector manner in the same coordinate system to obtain the final alignment drift vector between the target container and the electro-permanent magnet crane.

[0038] It should be noted that in this application, the additional drift amount refers to the additional displacement and attitude change that the electro-permanent magnet jack may generate at the next moment due to the continuous effect of environmental disturbances, based on the current moment; the static pose deviation refers to the fixed error between the target container and the electro-permanent magnet jack under ideal alignment conditions, including translational deviation and rotational deviation; the dynamic drift component refers to the instantaneous position change pattern of the pin hole point set due to environmental disturbances in a short time series; the alignment drift vector is the result of the additional drift amount, static pose deviation and dynamic drift component superimposed on vectors in a unified coordinate system, which fully describes the current and predicted comprehensive displacement deviation between the target container and the electro-permanent magnet jack.

[0039] In addition, it should be noted that by identifying the positioning pin holes in the top image of the target container, the precise geometric features of the container can be captured, enabling the detection of alignment errors to reach millimeter-level accuracy. Furthermore, by combining the identified pin hole point set with environmental disturbance data for visual deviation calibration, the actual deviations in spatial position and attitude between the electro-permanent magnet crane and the container can be accurately quantified. This includes both static installation errors and dynamic drifts caused by environmental disturbances such as wind loads and spreader sway. By fusing environmental disturbance data with visual observations and calculating the alignment drift vector through deviation calibration, real-time and forward-looking compensation can be provided for the servo control system.

[0040] In step S104, the alignment drift vector, the magnet displacement change trend, and the attraction force change trend are used to perform servo active compensation on the descent trajectory of the electro-permanent magnet lifter, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter. Based on the positioning motion trajectory, the electro-permanent magnet lifter is driven to perform precise pin hole positioning.

[0041] In some embodiments, the descent trajectory of the electro-permanent magnet lifter is actively compensated by the alignment drift vector, the magnet displacement change trend, and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter. Specifically, this can be achieved in the following manner: The alignment drift vector, the magnet displacement change trend, and the attraction force change trend are integrated into a comprehensive state evaluation matrix; The descent trajectory of the electro-permanent magnet lifter is dynamically corrected based on the comprehensive state evaluation matrix to obtain the corrected trajectory. The feasibility of the corrected trajectory is verified and smoothed based on preset safety constraints, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter.

[0042] In practical implementation, firstly, the real-time calculated alignment drift vector, magnet displacement change trend, and attraction force change trend are uniformly sampled and transformed into the same reference coordinate system. These three are then arranged into a matrix structure according to spatial dimensions and time series to obtain a comprehensive state evaluation matrix. Next, the descent trajectory of the electro-permanent magnet lifter can be dynamically corrected based on the comprehensive state evaluation matrix. Specifically, using the comprehensive state evaluation matrix as input, a closed-loop feedback mechanism is established in the servo control algorithm. The deviation information in the matrix is ​​mapped to real-time motion commands for the lifting device fine-tuning mechanism (including cross slide translation and rotary motor attitude adjustment). Through continuous iterative calculation of the real-time corrected trajectory, the descent speed, acceleration, and direction of the electro-permanent magnet lifter are dynamically adjusted, enabling precise compensation simultaneously in both the vertical and horizontal planes, thus obtaining the corrected trajectory. Finally, the feasibility of the corrected trajectory can be verified and smoothed based on preset safety constraints. That is, after obtaining the corrected trajectory, the feasibility of the corrected trajectory is verified by combining preset safety constraints, including maximum descent speed, maximum acceleration, magnetic impact limit and lifting device attitude tolerance, eliminating trajectory segments that may cause collision or overload, and optimizing the corrected trajectory through smoothing algorithms (such as cubic spline interpolation or multi-segment acceleration limit) to make the descent motion smooth and continuous, so that the final trajectory curve is used as the positioning motion trajectory of the electro-permanent magnet lifter.

[0043] In some embodiments, the electro-permanent magnet lifter is driven to perform precise pin hole positioning based on the positioning motion trajectory. Specifically, firstly, the generated positioning motion trajectory is input into the servo control system of the electro-permanent magnet lifter, controlling the trolley, carriage, and fine-tuning mechanism of the spreader to move synchronously according to the trajectory instructions, ensuring the magnet descends smoothly in both the vertical and horizontal directions while maintaining stable posture. Secondly, during the descent, multi-modal sensors on the spreader continuously collect data on magnet displacement, attraction force, visual images, and environmental disturbances. These real-time observations are compared and fed back with predicted trends in magnet displacement, attraction force, and alignment drift vectors to form a closed-loop control. Subsequently, the servo controller adjusts the fine-tuning mechanism's displacement and posture based on real-time feedback, precisely compensating for minor deviations caused by wind load, spreader sway, or magnetic circuit non-uniformity, ensuring the magnet can smoothly approach the positioning pin hole on the top of the container. Finally, when the magnet and container pin hole are aligned, the attraction force reaches a preset stable value, and the pressure sensor confirms that the magnet has reliably contacted the container surface. The system completes precise pin hole positioning, achieving high-precision automatic adsorption and safe operation of the electro-permanent magnet lifter in complex environments.

[0044] It should be noted that by using the alignment drift vector, the magnet displacement change trend, and the attraction force change trend to perform servo active compensation on the descent trajectory of the electro-permanent magnet lifter, and generating a positioning motion trajectory to drive the magnet to accurately position the pin hole, real-time comprehensive correction of static errors, dynamic drift, and magnetic force changes can be achieved. By fusing visual observation, magnetic field state, and environmental disturbance information, the system can fully perceive the actual spatial deviation between the spreader and the container. By predicting the magnet displacement and attraction force trends, the servo control can proactively adjust the descent trajectory and actively offset the offset caused by environmental disturbances or magnetic circuit non-uniformity, thereby achieving precise positioning control of the electro-permanent magnet lifter.

[0045] Therefore, this application firstly acquires multimodal sensing data of the electro-permanent magnet lifter and constructs a magnetic circuit adsorption feature vector based on the magnetic field distribution signal. Then, it combines this feature vector to predict the magnet displacement change trend and the attraction force change trend, which can achieve in-depth characterization and forward-looking judgment of the permanent magnet adsorption process. The magnetic coupling strength, adsorption uniformity, and magnetic leakage can be quantitatively evaluated through the magnetic circuit adsorption feature vector. Secondly, by identifying the positioning pin holes in the top image of the target container, the precise geometric features of the container can be captured, and the detection of alignment error can reach the millimeter level accuracy. Furthermore, by combining the identified pin hole point set with environmental disturbance data for visual deviation calibration, the actual deviation between the electro-permanent magnet lifter and the container in spatial position and attitude can be accurately quantified. Finally, by using the alignment drift vector, the magnet displacement change trend, and the attraction force change trend to perform servo active compensation on the descent trajectory of the electro-permanent magnet lifter and generating a positioning motion trajectory to drive the magnet to accurately position the pin holes, real-time comprehensive correction of static error, dynamic drift, and magnetic force change can be achieved to realize precise positioning control of the electro-permanent magnet lifter.

[0046] In summary, the technical solution adopted in this application can integrate multimodal sensing and predict the magnetic circuit state, thereby performing servo active compensation for the positioning trajectory to achieve precise positioning control of the electro-permanent magnet lifter.

[0047] In another aspect, in some embodiments, this application provides a precise positioning control system for an electro-permanent magnet lifter, with reference to... Figure 3 The figure is a schematic diagram of the structure of a precision positioning control system for an electro-permanent magnet lifter according to some embodiments of this application. The precision positioning control system for the electro-permanent magnet lifter includes: The initial self-test module 201 is used to control the electro-permanent magnet lifter to perform an initial self-test after the port container crane is started, and to move the electro-permanent magnet lifter to the target container. The state prediction module 202 is used to acquire multimodal sensing data of the electro-permanent magnet lifter, determine the magnetic circuit adsorption feature vector based on the magnetic field distribution signal in the multimodal sensing data, perform state prediction based on the magnetic circuit adsorption feature vector, and then obtain the magnet displacement change trend and the attraction force change trend. The drift calibration module 203 is used to identify the positioning pin holes through the top image of the target container in the multimodal sensing data, and then perform visual deviation calibration based on the identified positioning pin hole point set and the environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. The positioning control module 204 is used to perform servo active compensation on the descent trajectory of the electro-permanent magnet lifter using the alignment drift vector, the magnet displacement change trend and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter, and driving the electro-permanent magnet lifter to perform precise pin hole positioning based on the positioning motion trajectory.

[0048] In addition, this application also provides a computer device, which includes a memory and a processor. The memory stores code, and the processor is configured to acquire the code and execute the above-described electro-permanent magnet lifter precise positioning control method.

[0049] In some embodiments, reference Figure 4 The figure is a schematic diagram of the structure of a computer device for implementing a precise positioning control method for an electro-permanent magnet lifter according to some embodiments of this application. The precise positioning control method for the electro-permanent magnet lifter in the above embodiments can... Figure 4 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.

[0050] The processor 301 can be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more devices used to control the execution of the precise positioning control method for the electro-permanent magnet lifter in this application.

[0051] The communication bus 302 can be used to transmit information between the aforementioned components.

[0052] The memory 303 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 303 may exist independently and be connected to the processor 301 via the communication bus 302. The memory 303 may also be integrated with the processor 301.

[0053] The memory 303 stores program code for executing the scheme of this application, and its execution is controlled by the processor 301. The processor 301 executes the program code stored in the memory 303. The program code may include one or more software modules. In the above embodiments, the determination of the precise positioning control method of the electro-permanent magnet lifter can be achieved by the processor 301 and one or more software modules in the program code in the memory 303.

[0054] Communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0055] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0056] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0057] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described precise positioning control method for an electro-permanent magnet lifter.

[0058] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0059] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A precise positioning control method for an electro-permanent magnet lifter, characterized in that, Includes the following steps: After the port container crane is started, the control electro-permanent magnet lifter performs an initial self-test and moves the electro-permanent magnet lifter to the airspace above the target container. Multimodal sensing data of the electro-permanent magnet lifter is acquired, and an initial magnetic field distribution matrix is ​​constructed using the magnetic field distribution signal in the multimodal sensing data. The initial magnetic field distribution matrix is ​​then filtered for noise reduction and temperature compensation to obtain a standardized magnetic field distribution map. Magnetic field uniformity extraction, magnetic flux estimation, and edge gradient extraction are performed on the standardized magnetic field distribution map to obtain a magnetic field uniformity index, load magnetic flux, and magnetic field leakage coefficient. A magnetic circuit adsorption feature vector is constructed based on the magnetic field uniformity index, the load magnetic flux, and the magnetic field leakage coefficient. State prediction is then performed based on the magnetic circuit adsorption feature vector to obtain the magnet displacement change trend and the attraction force change trend. The magnetic field uniformity index characterizes the uniformity of the magnetic field distribution on the adsorption working surface of the electro-permanent magnet lifter, the load magnetic flux characterizes the actual adsorption capacity output by the permanent magnet, and the magnetic field leakage coefficient represents the degree of magnetic field leakage outside the effective adsorption area. Positioning pin holes are identified using the top image of the target container from the multimodal sensing data. The identified positioning pin hole set is spatiotemporally aligned and fused with real-time tilt angle data and environmental disturbance data from the multimodal sensing data. Based on the Kalman filter algorithm, the fused data is optimally estimated to predict the additional drift amount that will occur in the next moment due to the continued effect of environmental disturbance. A dynamic deviation compensation model is established, and based on this model, the spatial distribution geometric characteristics of the positioning pin hole set and the drift pattern of the set after being affected by environmental disturbance are analyzed to obtain the static pose deviation and dynamic drift components. The additional drift amount, the static pose deviation, and the dynamic drift components are vector-synthesized to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. The descent trajectory of the electro-permanent magnet lifter is actively compensated by the alignment drift vector, the magnet displacement change trend, and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter. Based on the positioning motion trajectory, the electro-permanent magnet lifter is driven to perform precise pin hole positioning.

2. The precise positioning control method for an electro-permanent magnet lifter as described in claim 1, characterized in that, The initialization self-test of the control electro-permanent magnet lifter specifically includes: Perform zero-point regression on the gear position sensor, record the encoder zero position, perform zero-point drift measurement and save the bias for all magnetic field sensors, force sensors, vision sensors and anemometers, and read and verify the health status of each sensor.

3. The precise positioning control method for an electro-permanent magnet lifter as described in claim 1, characterized in that, The multimodal sensing data includes magnetic field distribution signals, images of the top of the target container, and environmental disturbance data.

4. The precise positioning control method for an electro-permanent magnet lifter as described in claim 1, characterized in that, Based on the magnetic circuit adsorption feature vector, state prediction is performed to obtain the magnet displacement change trend and the attraction force change trend, specifically including: Obtain a pre-trained prediction model based on a long short-term memory network structure; The magnetic circuit adsorption feature vector is input into the prediction model based on the long short-term memory network structure for state prediction, thereby obtaining the magnet displacement change trend and the attraction force change trend.

5. The precise positioning control method for an electro-permanent magnet lifter as described in claim 1, characterized in that, The identification of positioning pin holes using the target container top image from the multimodal sensing data specifically includes: The top image of the target container in the multimodal sensing data is preprocessed; Key regions are located in the preprocessed top image of the target container to obtain the bounding boxes of all container corner pieces in the top image of the target container; Inside the bounding box of each located container corner piece, a detection algorithm based on circular Hough transform is used to accurately locate the precise pixel center position of the positioning pin hole. The precise pixel center position is converted into three-dimensional spatial coordinates in the spreader coordinate system, thereby identifying the set of positioning pin holes on the top of the target container.

6. The precise positioning control method for an electro-permanent magnet lifter as described in claim 1, characterized in that, The servo-active compensation for the descent trajectory of the electro-permanent magnet lifter is performed using the alignment drift vector, the magnet displacement change trend, and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter. Specifically, this includes: The alignment drift vector, the magnet displacement change trend, and the attraction force change trend are integrated into a comprehensive state evaluation matrix; The descent trajectory of the electro-permanent magnet lifter is dynamically corrected based on the comprehensive state evaluation matrix to obtain the corrected trajectory. The feasibility of the corrected trajectory is verified and smoothed based on preset safety constraints, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter.

7. A precision positioning control system for an electro-permanent magnet lifter, used to execute a precision positioning control method for an electro-permanent magnet lifter as described in any one of claims 1 to 6, characterized in that, include: The initial self-test module is used to control the electro-permanent magnet lifter to perform an initial self-test after the port container crane is started, and to move the electro-permanent magnet lifter to the target container. The state prediction module is used to acquire multimodal sensing data of the electro-permanent magnet lifter, determine the magnetic circuit adsorption feature vector based on the magnetic field distribution signal in the multimodal sensing data, perform state prediction based on the magnetic circuit adsorption feature vector, and then obtain the magnet displacement change trend and the attraction force change trend. The drift calibration module is used to identify the positioning pin holes through the top image of the target container in the multimodal sensing data, and then perform visual deviation calibration based on the identified positioning pin hole point set and the environmental disturbance data in the multimodal sensing data to obtain the alignment drift vector between the target container and the electro-permanent magnet crane. The positioning control module is used to perform servo active compensation on the descent trajectory of the electro-permanent magnet lifter using the alignment drift vector, the magnet displacement change trend, and the attraction force change trend, thereby generating the positioning motion trajectory of the electro-permanent magnet lifter, and driving the electro-permanent magnet lifter to perform precise pin hole positioning based on the positioning motion trajectory.

8. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to retrieve the code and execute the precise positioning control method for an electro-permanent magnet lifter as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the precise positioning control method for the electro-permanent magnet lifter as described in any one of claims 1 to 6.

Images