Steel ladle lining refractory material erosion detection system

Abstract

The invention provides a steel ladle lining refractory material corrosion detection system, and belongs to the field of steel ladle lining detection. The steel ladle lining refractory material corrosion detection system comprises an identification unit used for obtaining multi-dimensional environment parameter data of the surface of a steel ladle lining when an industrial robot uses a contact type detection probe to scan the lining; and the judgment unit is used for judging whether the identification area is a false bulge or a dangerous bulge caused by surface oxidation based on the acquired multi-dimensional environment parameter data in the movement detection process of the industrial robot. According to the invention, the judgment unit is used for judging the types of the bulges, distinguishing dangerous bulges from oxidized pseudo bulges and planning a path bypassing the dangerous bulges, so that mechanical collision or scraping between the contact type detection probe and sharp fragments and hard residues is avoided, the probe part is effectively prevented from falling off, and the detection efficiency is improved. The safety of the contact detection probe is guaranteed, the detection efficiency is improved, the service life of equipment is prolonged, and the efficiency is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of ladle lining inspection technology, and more specifically to a ladle lining refractory material corrosion detection system. Background Technology

[0002] In the steel smelting process, the ladle is a key piece of equipment that receives and refines molten steel at high temperatures. The integrity of its refractory lining directly affects production safety and efficiency. Traditional inspection methods mainly rely on manual visual inspection or spot temperature gun sampling, which has significant limitations. In recent years, intelligent inspection robots integrating laser sensing, infrared temperature measurement, and machine vision have emerged. With a robotic arm equipped with multimodal sensors, it can automatically identify the ladle number, locate the eroded area of ​​the lining, and establish a three-dimensional model of refractory corrosion. Combined with intelligent algorithms, it can predict the lining life, significantly improving inspection efficiency and safety.

[0003] In existing technologies, steel ladle linings gradually develop irregular erosion patterns due to long-term use, resulting in a rough inner wall surface. Furthermore, residue accumulation and irregular protrusions are prone to occur inside the ladle. When using contact-type detection probes to scan the lining, the probe components are easily mechanically impacted and detached when they come into contact with the protruding parts of the lining surface. If there are sharp refractory brick fragments on the lining surface, the wear-resistant coating at the bottom of the probe will be exposed to the metal substrate due to rapid scraping, leading to an increase in the contact friction coefficient between the probe and the lining and significantly shortening the effective service life of the probe. Summary of the Invention

[0004] The purpose of this invention is to provide a ladle lining refractory material corrosion detection system to solve the problem in the prior art where irregular protrusions formed on the surface of the ladle lining can mechanically impact the contact detection probe during detection, causing damage to the contact detection probe.

[0005] To achieve the above objectives, embodiments of the present invention provide a steel ladle lining refractory material corrosion detection system. The system includes: an identification unit for acquiring multi-dimensional environmental parameter data of the steel ladle lining surface when an industrial robot scans the lining using a contact-type detection probe; a judgment unit for determining, based on the acquired multi-dimensional environmental parameter data, whether the identified area is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation during the industrial robot's movement and detection process; and an execution control unit for controlling the operation of the industrial robot's robotic arm according to the judgment result, thereby moving the contact-type detection probe and achieving safety protection for the probe.

[0006] Optionally, the multidimensional environmental parameter data includes the temperature of the ladle liner surface, an image, and the contact force generated when the contact detection probe contacts the liner. After acquiring the multidimensional environmental parameter data of the ladle liner surface, the identification unit is further configured to: filter, calibrate, align, and preprocess the multidimensional environmental parameter data.

[0007] Optionally, the step of determining whether the identified area is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation based on the acquired multidimensional environmental parameter data includes: calculating temperature feature values ​​using a preset temperature influence function; calculating image feature values ​​using a preset image processing method; calculating contact force feature values ​​using a preset contact force influence function; calculating protrusion feature values ​​based on the calculated temperature feature values, image feature values, and contact force feature values; and determining whether the identified area is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation based on the calculated protrusion feature values.

[0008] Optionally, determining whether the identification area is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation based on the calculated protrusion feature value includes: determining the identification area as a dangerous protrusion when the protrusion feature value is greater than a first threshold; and determining the identification area as a pseudo-protrusion when the protrusion feature value is less than a second threshold, wherein the first threshold is greater than the second threshold.

[0009] Optionally, when the protrusion feature value is between the first threshold and the second threshold, the judgment unit is further configured to: use a pre-trained random forest classification model to determine whether the identification region is a pseudo protrusion or a dangerous protrusion caused by surface oxidation.

[0010] Optionally, training the random forest classification model includes: acquiring historical multidimensional environmental parameter data and constructing a training set; dividing the training set using bootstrap sampling to obtain multiple training sets, with the samples not selected forming out-of-bag data; training a corresponding decision tree using each of the multiple training sets, and validating each decision tree using all data not containing out-of-bag data; and combining each trained decision tree into a trained random forest classification model.

[0011] Optionally, the execution control unit generates control commands to control the movement of the robotic arm of the industrial robot to move the contact detection probe, including: controlling the robotic arm of the industrial robot to move according to the judgment result to move the contact detection probe, including: when the identification area is a dangerous protrusion, generating a probe lifting command to make the probe move horizontally around it.

[0012] Optionally, controlling the operation of the industrial robot's robotic arm based on the judgment result includes: When the identified area is a pseudo-protrusion caused by surface oxidation, the robotic arm is controlled to decelerate; when the identified area is a dangerous protrusion, the robotic arm is controlled to perform obstacle avoidance.

[0013] Optionally, the multidimensional environmental parameter data includes contact force collected using a six-dimensional force sensor and point cloud data collected using lidar. When the identified area is a dangerous protrusion, controlling the robotic arm to perform obstacle avoidance motion includes: predicting part or all of the shape of the dangerous protrusion based on the collected point cloud data and using a density clustering algorithm; and determining the motion direction vector and contact force magnitude of the contact detection probe based on the collected contact force and the predicted part or all of the dangerous protrusion, using an artificial potential field method.

[0014] Optionally, the method of using an artificial potential field to determine the motion direction vector and contact force of the contact detection probe includes: constructing a repulsive field based on the shape of the dangerous protrusion predicted by a density clustering algorithm, generating a virtual repulsive force that pushes the probe away from the protrusion, the intensity of which increases nonlinearly with decreasing distance; constructing a directional gravitational field with the detection point of the contact detection probe as the gravitational center, the intensity of which is proportional to the Euclidean distance from the probe to the target point; combining the gravitational and repulsive force vectors to obtain a total resultant force, the direction of which is used as the probe's motion direction; and dynamically adjusting the potential field intensity based on the actual contact force, in conjunction with six-dimensional force sensor data, so that the probe can maintain a stable contact force while avoiding obstacles to complete the detection.

[0015] This invention, through an identification unit, collects real-time data on the temperature, image, and contact force of the ladle lining and extracts key features. After preprocessing, these features provide a decision-making basis for the judgment unit. The judgment unit then identifies the type of protrusion, distinguishing between dangerous protrusions and pseudo-oxidized protrusions, and outputs obstacle avoidance commands. The execution control unit initiates dynamic path planning based on the judgment results, dynamically adjusting the probe height using an artificial potential field method to ensure safe obstacle avoidance. This ensures the safety of the contact detection probe while improving detection efficiency and equipment lifespan, significantly reducing the need for manual intervention. Furthermore, dynamic path planning reduces unnecessary downtime, significantly improving efficiency.

[0016] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the structure of the refractory material corrosion detection system for steel ladle lining provided in an embodiment of the present invention. Detailed Implementation

[0018] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0019] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0020] As mentioned above, due to the accumulation of residue and irregular protrusions in the ladle lining after long-term use, the use of contact detection probes for lining scanning can easily cause the probe components to fall off due to mechanical impact.

[0021] Figure 1 This diagram illustrates a flow chart of a refractory lining corrosion detection system for a steel ladle according to an embodiment of the present invention. Figure 1 As shown, this embodiment of the invention provides a steel ladle lining refractory material corrosion detection system, including: an identification unit, used to acquire multi-dimensional environmental parameter data of the steel ladle lining surface when an industrial robot scans the lining using a contact detection probe; a judgment unit, used to determine, based on the acquired multi-dimensional environmental parameter data, whether the identification area is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation during the industrial robot's movement detection process; and an execution control unit, used to generate control commands to control the movement of the industrial robot's robotic arm according to the judgment result, so as to drive the contact detection probe to move and achieve safety protection for the contact detection probe.

[0022] For example, when using a contact detection probe to inspect the surface of a steel ladle liner, the identification area is the local detection range defined by the contact detection probe on the steel ladle liner surface. The temperature distribution of the steel ladle liner surface can be collected by an equipped infrared temperature sensor, and image information of the steel ladle liner surface can be captured by an equipped industrial camera, recording features such as texture, edge shape, and color distribution of the raised areas. By integrating a force sensor onto the contact detection probe, contact force data is obtained. By fusing multi-dimensional data such as temperature, image, and contact force, a judgment is made as to whether the raised areas on the steel ladle liner surface are false raised areas caused by surface oxidation or dangerous raised areas. Compared with identification methods based on a single data dimension, this method can more comprehensively capture the differences in the physical characteristics of the raised areas, reduce the false judgment rate, effectively prevent dangerous raised areas from being missed and thus avoid probe collision risks, and ensure the accuracy and reliability of the judgment results.

[0023] The preferred multidimensional environmental parameter data in this embodiment of the invention includes the temperature of the ladle liner surface, an image, and the contact force generated when the contact detection probe contacts the liner. After acquiring the multidimensional environmental parameter data of the ladle liner surface, the identification unit is further used to: filter, calibrate, align, and preprocess the multidimensional environmental parameter data.

[0024] As illustrated by the example, filtering, calibration, alignment, and preprocessing of multidimensional environmental parameter data can effectively eliminate sensor noise interference; correct zero-point drift of temperature and lens distortion of images; and extract multidimensional features such as texture, edge, color, and shape of the protrusions on the inner surface of the ladle lining in the image to provide objective visual evidence for distinguishing between false protrusions and dangerous protrusions. Furthermore, by aligning the time of occurrence of temperature changes and contact force changes through timestamps, and linking coordinate transformations to the spatial position of the protrusion area in the image with the probe, the reliability and consistency of the data can be ensured.

[0025] In a preferred embodiment of the present invention, determining whether an identification area is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation based on the acquired multidimensional environmental parameter data may include: calculating temperature feature values ​​using a preset temperature influence function; calculating image feature values ​​using a preset image processing method; calculating contact force feature values ​​using a preset contact force influence function; calculating protrusion feature values ​​based on the calculated temperature feature values, image feature values, and contact force feature values; and determining whether an identification area is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation based on the calculated protrusion feature values.

[0026] For example, regarding temperature data for the identified area Ts Temperature data can be used Ts Input a preset temperature influence function to generate the corresponding temperature feature values. f ( Ts The temperature effect function is, for example, as follows: sigmoid The function reflects the influence of the temperature difference between the identification area and the reference temperature on the determination of the protrusion type. Invalid sampling points are removed through filtering, and the filtered valid parameters are input into the following temperature influence function:

[0027] in, This is the difference between the target area and the reference temperature. The weighting coefficient represents the impact of temperature differences on the final result.

[0028] For the image data Ig of the recognition area, a preset image processing method can be used to process the image data Ig to generate corresponding image feature values. The image feature value is a comprehensive value obtained by weighting edge, shape and area features, which reflects the degree of influence of the visual characteristics of the target area on the determination of the protrusion type.

[0029] Feature validity verification involves labeling extracted feature parameters from image data (Ig) as valid or invalid features. For example, feature validity verification can be performed through threshold judgment and morphological analysis. For instance, a minimum threshold for the edge gradient magnitude E can be set to remove blurry edges below the threshold; morphological operations can be used to filter regions with areas within a reasonable range, labeling feature parameters that meet the criteria as valid features and those that do not as invalid features.

[0030] By removing parameters labeled as invalid features, the remaining valid feature parameters are substituted into the image processing method function:

[0031] Where C represents the shape feature; The weights are respectively for edge, shape, and area, and , This represents the maximum area.

[0032] Contact force data for the identification area Contact force data can be used Input a preset contact force influence function to generate the corresponding contact force characteristic values. The contact force characteristic value is a comprehensive value obtained by weighting the force change rate, peak difference, and peak size, which reflects the degree of influence of the contact force characteristics of the target area on the protrusion type judgment.

[0033] Contact force data is filtered through force signal processing. Interference signals are labeled as interference values ​​or effective force values. For example, interference signals can be filtered using a low-pass filter and a force range threshold. For instance, a 5Hz low-pass filter can be used to remove mechanical vibration noise, and an effective range of contact force can be set. Signals within the range after processing can be labeled as effective force values, while those outside the range can be labeled as interference values.

[0034] By removing the parameters corresponding to the sampling points marked as interference values, the effective force value parameters are substituted into the contact force influence function:

[0035] in, For the rate of change of force, For maximum contact force, For average contact force, These are the weighting coefficients, and .

[0036] For the calculated temperature feature value f(Ts), image feature value Contact force characteristics Based on the fusion algorithm, corresponding convex feature values ​​can be generated. The convexity feature value is obtained by weighted summation of temperature, image, and contact force features, reflecting the degree of influence of the overall characteristics of the target area on the convexity type judgment.

[0037] By verifying the reasonableness of the weights, the weight coefficients corresponding to the aforementioned feature values ​​are labeled as reasonable or unreasonable weights. This verification can be performed through training with historical data and adjustments based on expert experience. For example, the information gain of each feature can be calculated based on the past 1000 labeled samples to determine the initial weights. After verification by domain experts, those that meet the verification standards are marked as reasonable weights, and those that do not meet the standards are marked as unreasonable weights.

[0038] Mapping the aforementioned feature values ​​to the fusion model upon which the convex feature value calculation relies, the influence of feature values ​​labeled as having unreasonable weights is removed; that is, the weights of feature values ​​related to unreasonable weights are set to 0. The remaining feature values ​​are then weighted and summed with the reasonable weights, and substituted into the fusion formula:

[0039] in,

[0040] .

[0041] A preferred embodiment of the present invention determines whether an identification region is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation based on the calculated protrusion feature value, including: when the protrusion feature value is greater than a first threshold, determining that the identification region is a dangerous protrusion; when the protrusion feature value is less than a second threshold, determining that the identification region is a pseudo-protrusion, wherein the first threshold is greater than the second threshold.

[0042] To illustrate, for the calculated convex feature value Based on this raised feature value and using preset threshold judgment rules, the identified area can be marked as a false raised label caused by surface oxidation or a dangerous raised label. A high-risk threshold can be used. (First threshold) and low-risk threshold (Second threshold) is used for label determination, for example, when When, it is marked as a hazard raised label; when When, it is marked as a pseudo-raised label caused by surface oxidation; when Then, a pre-trained random forest classification model is used to further determine and label the data.

[0043] Each identification area is labeled as a false bump or a dangerous bump to avoid triggering unnecessary obstacle avoidance actions due to misjudgment, reduce detection interruptions, and ensure scanning efficiency.

[0044] In a preferred embodiment of the present invention, when the protrusion feature value is between the first threshold and the second threshold, the judgment unit is further configured to: use a pre-trained random forest classification model to determine whether the identification region is a pseudo protrusion or a dangerous protrusion caused by surface oxidation.

[0045] To illustrate, the decision unit calls a pre-trained random forest classification model to perform more refined classification based on multi-dimensional feature values, outputting the final convexity type label. Each decision tree in the model independently classifies the feature vector and outputs the prediction results for dangerous convexities and pseudo-convexities. The results of all decision trees are integrated through a majority voting mechanism, outputting the probability that the target region is a dangerous convexity and the probability that it is a pseudo-convexity. The final label is determined based on the probability value: if the probability of dangerous convexity is higher, it is judged as dangerous convexity; otherwise, it is judged as pseudo-convexity. This provides a basis for judgment for the trajectory update of the control unit.

[0046] The preferred method for training the random forest classification model in this embodiment of the invention includes: acquiring historical multidimensional environmental parameter data and constructing a training set; dividing the training set using bootstrap sampling to obtain multiple training sets, with the samples not selected forming out-of-bag data; training a corresponding decision tree using each of the multiple training sets, and validating each decision tree using all data not containing out-of-bag data; and combining each trained decision tree into a trained random forest classification model.

[0047] Following the example above, let the training set be... The sample size is ; Generated by sampling from Bootstrap a sample set Each sample set contains N samples; the unselected samples constitute the out-of-bag data. Each sample set Used to train a decision tree During training, some features are randomly selected to participate in node splitting of each tree, ensuring tree structure diversity. Samples in , using all that do not contain The decision tree is used for prediction, and the percentage of incorrect predictions is the out-of-bag error for that sample. The mean out-of-bag error of all samples is used as an evaluation index of the model's generalization ability, enabling precise judgment of the type of protrusion within the threshold range and improving the reliability of the steel ladle lining refractory material corrosion detection system.

[0048] In a preferred embodiment of the present invention, the operation of the robotic arm of the industrial robot is controlled according to the judgment result to move the contact detection probe, including: when the identification area is a dangerous protrusion, a probe lifting command is generated so that the probe can pass around it horizontally.

[0049] According to the determination result of the protrusion type by the judgment unit, the embodiment of the present invention can plan a path to avoid dangerous protrusions, avoid mechanical impact or scratching between the contact detection probe and sharp fragments or hard residues, effectively prevent probe parts from falling off and wear-resistant coating from being damaged, and significantly extend the service life of the equipment.

[0050] In a preferred embodiment of the present invention, the operation of the robotic arm of the industrial robot is controlled according to the judgment result, including: when the identified area is a pseudo-protrusion caused by surface oxidation, the robotic arm is controlled to decelerate; when the identified area is a dangerous protrusion, the robotic arm is controlled to perform obstacle avoidance.

[0051] For example, when passing through an area identified as a false protrusion, the robotic arm automatically reduces its operating speed to slowly pass through the oxidized area, avoiding probe slippage or data distortion caused by rapid movement. When passing through an area identified as a dangerous protrusion, after predicting the shape of the protrusion based on a density clustering algorithm, an bypass path is generated using an artificial potential field method. The robotic arm lifts the probe and dynamically adjusts the repulsive force intensity based on contact force feedback to ensure that the probe maintains safe contact with the lining while avoiding obstacles. After completing the local inspection, the original path is restored to continue scanning.

[0052] The preferred multidimensional environmental parameter data in this embodiment of the invention includes contact force collected by a six-dimensional force sensor and point cloud data collected by a lidar. When the identified area is a dangerous protrusion, controlling the robotic arm to perform obstacle avoidance movement includes: predicting part or all of the shape of the dangerous protrusion based on the collected point cloud data and using a density clustering algorithm; and determining the motion direction vector of the contact detection probe and the magnitude of the contact force based on the collected contact force and the predicted part or all of the dangerous protrusion, using an artificial potential field method.

[0053] Continuing with the example above, point cloud data is a set of three-dimensional coordinate points generated by scanning the surface of the ladle lining with LiDAR. Each point contains spatial location information, directly reflecting the geometric features of the ladle lining surface, including key information such as the location, height, and distribution range of protrusions. This complements the six-dimensional force sensor data. Point cloud data provides a visual three-dimensional morphology, while contact force data provides mechanical interactive feedback. Density clustering algorithms are used to "filter and group" these point clouds, eliminating irrelevant background points and clustering points belonging to the same dangerous protrusion into one category, thus separating the point cloud subset of dangerous protrusions. Based on the clustered protrusion point cloud subset, their shape characteristics are predicted, providing key parameters for the subsequent construction of a repulsive force field using the artificial potential field method. This ensures that the obstacle avoidance path of the robotic arm can conform to the actual shape of the protrusion and avoid collisions.

[0054] In a preferred embodiment of the invention, the motion direction vector and contact force of the contact detection probe are determined using an artificial potential field method. This includes: constructing a repulsive field based on the shape of the dangerous protrusion predicted by a density clustering algorithm, generating a virtual repulsive force that pushes the probe away from the protrusion, with the intensity of the virtual repulsive force increasing nonlinearly with decreasing distance; constructing a directional gravitational field with the detection point of the contact detection probe as the gravitational center, the intensity of which is proportional to the Euclidean distance from the probe to the target point; combining the gravitational and repulsive force vectors to obtain a total resultant force, the direction of which is used as the probe's motion direction; and dynamically adjusting the potential field intensity based on the actual contact force, combined with data from a six-dimensional force sensor, so that the probe maintains a stable contact force while avoiding obstacles to complete the detection.

[0055] Continuing with the example above, the target detection position of the contact detection probe is taken as the gravitational field target point. According to the formula Construct a gravitational field; among which, The gravitational coefficient determines the strength of the attraction between the target and the probe; the larger the coefficient, the more significant the pulling effect of the target on the probe. This indicates the current pose of the probe. Reach the target pose The greater the Euclidean distance, the higher the gravitational potential energy, and the stronger the tendency for the probe to move towards the target. The gravitational field simulates the target's pull on the probe, ensuring that the probe always maintains a tendency to move towards the target's detection position.

[0056] Set the location of the dangerous protrusion as the obstacle pose, construct a repulsive force field, and when the probe is in its current pose... Extracted by laser point cloud clustering distance When ensuring sufficient space for obstacle avoidance by the probe, the formula is as follows:

[0057] Calculate the potential energy. Among them, The repulsive coefficient is the coefficient of force. The closer the probe is to the protrusion, the faster the potential energy of the repulsive field increases, generating a strong repulsive force that pushes the probe away from the dangerous protrusion and avoids collision.

[0058] By analyzing the gravitational potential energy Repulsive field potential energy Calculate the gradients separately, according to the formula. Calculate the resultant force The direction of the total resultant force is the motion direction vector of the contact detection probe, guiding the probe towards the target detection position while avoiding dangerous protrusions. During probe movement, dynamic adaptation is achieved by combining data from a six-dimensional force sensor: if the contact force is abnormal (too large or too small), the potential field parameters (such as adjusting the attraction and repulsion coefficients) can be indirectly adjusted so that the probe neither collides with the protrusion nor fails to maintain stable contact with the inner lining surface to complete the detection, ensuring system safety and detection effectiveness.

[0059] This invention also provides an industrial robot operation support system, which includes the aforementioned steel ladle lining refractory material corrosion detection system, and further includes the following modules: The protection module is used to construct a safety barrier for robot operation, including: setting up a safety fence using high-strength materials, equipped with a safety door and a safety lock with a self-locking function; setting up protective railings between the robot's operating area and surrounding equipment, and covering key components such as motors and reducers with transparent polycarbonate sheet protective covers; marking conspicuous indicator lines on the ground, and equipping it with emergency lighting and safety exit indicator lights. The monitoring and early warning module is used to collect environmental parameters and equipment operating status data, including: monitoring ambient temperature, humidity, dust concentration, as well as motor current, reducer vibration speed, and bearing temperature, and activating alarms when thresholds are exceeded. The self-testing module is used for automatic detection, diagnosis, and emergency handling of the industrial robot's operation support system, including: automatically detecting the hardware and software status of the controller, servo drive, motor, reducer, sensors, encoders, etc., when powered on to ensure that the system is in a normal initial state; and analyzing monitoring data in real time during operation to identify the type and location of faults.

[0060] This embodiment uses the steel ladle lining inspection scenario in the iron and steel metallurgy industry as an example to illustrate the working process of the industrial robot operation support system: After powering on, the safety fence and guardrail gate are locked. Furthermore, the self-test module automatically checks whether the controller, motor, and reducer are operating normally. If any abnormality is found, a self-test alarm is immediately triggered and the fault location is displayed. The monitoring and early warning module collects data from the work area in real time. When environmental parameters exceed preset thresholds, an early warning is immediately triggered. The monitoring and early warning module also monitors the operating status of key robot components. If abnormal current, vibration, or temperature is detected in the motor, reducer, etc., an immediate warning is issued and the direction of inspection is indicated. After the detection task is completed, the system automatically records the multi-dimensional data of this operation as historical data, providing data support for subsequent optimization of threshold parameters, improvement of fault diagnosis models, and path planning algorithms. In addition, the protection module maintains a safe state after the task is completed: the safety fence and safety gate remain locked until unlocked by authorized personnel; emergency lighting and safety exit indicator lights remain in standby mode, ensuring that the work area still has basic lighting guidance functions in the event of a sudden power outage.

[0061] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0062] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0063] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0064] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0065] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0066] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0067] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0068] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0069] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A system for detecting corrosion of refractory lining materials in steel ladles, characterized in that, The ladle lining refractory material corrosion detection system includes: The identification unit is used to acquire multi-dimensional environmental parameter data of the ladle lining surface when the industrial robot scans the lining using a contact detection probe; The judgment unit is used to determine, based on the acquired multidimensional environmental parameter data, whether the identification area is a false protrusion or a dangerous protrusion caused by surface oxidation during the mobile detection process of the industrial robot. An execution control unit is used to control the operation of the robotic arm of the industrial robot according to the judgment result, so as to drive the contact detection probe to move and achieve safety protection for the contact detection probe.

2. The refractory material corrosion detection system for steel ladle linings according to claim 1, characterized in that, The multidimensional environmental parameter data includes the temperature and image of the ladle liner surface, as well as the contact force generated when the contact detection probe contacts the liner. After acquiring the multidimensional environmental parameter data of the ladle liner surface, the identification unit is further used for: The multidimensional environmental parameter data is filtered, calibrated, aligned, and preprocessed.

3. The refractory material corrosion detection system for steel ladle linings according to claim 2, characterized in that, The process of determining whether a region is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation, based on the acquired multidimensional environmental parameter data, includes: Calculate temperature characteristic values ​​using a preset temperature influence function; Calculate image feature values ​​using a preset image processing method; The characteristic value of the contact force is calculated using a preset contact force influence function; Based on the calculated temperature feature value, image feature value, and contact force feature value, the bulge feature value is calculated; Based on the calculated protrusion feature values, the identification area is determined to be a pseudo protrusion or a dangerous protrusion caused by surface oxidation.

4. The refractory material corrosion detection system for steel ladle linings according to claim 3, characterized in that, The determination of whether a region is a pseudo-protrusion or a dangerous protrusion caused by surface oxidation, based on the calculated protrusion feature value, includes: When the protrusion feature value is greater than the first threshold, the identification area is determined to be a dangerous protrusion; When the protrusion feature value is less than the second threshold, the identification region is determined to be a pseudo-protrusion. Wherein, the first threshold is greater than the second threshold.

5. The refractory material corrosion detection system for steel ladle linings according to claim 4, characterized in that, When the protrusion feature value is between the first threshold and the second threshold, the determination unit is further configured to: Using a pre-trained random forest classification model, the region is identified as a pseudo-protrusion or a dangerous protrusion caused by surface oxidation.

6. The refractory material corrosion detection system for steel ladle linings according to claim 4, characterized in that, Training the random forest classification model includes: Acquire historical multidimensional environmental parameter data and construct a training set; The training set is divided using bootstrap sampling to obtain multiple training sets, with the unsampled samples forming out-of-bag data. Using each of the multiple training sets, train the corresponding decision tree, and validate each decision tree using all data that does not include out-of-bag data; Each trained decision tree is combined to form a trained random forest classification model.

7. The refractory material corrosion detection system for steel ladle linings according to claim 1, characterized in that, The step of controlling the robotic arm of the industrial robot to move, based on the judgment result, includes: When the identified area is a dangerous protrusion, a probe lifting command is generated, causing the probe to bypass it horizontally.

8. The refractory material corrosion detection system for steel ladle linings according to claim 1, characterized in that, The step of controlling the operation of the industrial robot's robotic arm based on the judgment result includes: When the identified area is a pseudo-protrusion caused by surface oxidation, the robotic arm is controlled to decelerate. When the identified area is a dangerous protrusion, the robotic arm is controlled to perform obstacle avoidance.

9. The refractory material corrosion detection system for steel ladle linings according to claim 8, characterized in that, The multidimensional environmental parameter data includes contact force collected using a six-dimensional force sensor and point cloud data collected using lidar. When the identified area is a dangerous protrusion, controlling the robotic arm to perform obstacle avoidance maneuvers includes: Based on the collected point cloud data and using a density clustering algorithm, the partial or complete shape of the dangerous protrusion is predicted. Based on the collected contact force and the predicted partial or complete shape of the dangerous protrusion, and using the artificial potential field method, the motion direction vector of the contact detection probe and the magnitude of the contact force are determined.

10. The refractory material corrosion detection system for steel ladle linings according to claim 8, characterized in that, The method of determining the motion direction vector and contact force of the contact detection probe using an artificial potential field includes: A repulsive field is constructed based on the shape of the dangerous protrusion predicted by the density clustering algorithm, and a virtual repulsive force is generated to push the probe away from the protrusion. The intensity of the virtual repulsive force increases nonlinearly as the distance decreases. A directional gravitational field is constructed with the detection point of the contact detection probe as the gravitational center, and the gravitational strength is proportional to the Euclidean distance from the probe to the target point; The total resultant force is obtained by vector synthesis of attraction and repulsion, and the direction of the total resultant force is used as the direction of probe movement. At the same time, combined with the data from the six-dimensional force sensor, the potential field strength is dynamically adjusted according to the actual contact force, so that the probe can maintain a stable contact force to complete the detection while avoiding obstacles.

Images