Iron and steel raw material conveying mistake proofing method and electronic equipment

Abstract

The invention discloses a steel raw material conveying mistake proofing method and electronic equipment, and relates to the field of steel raw material conveying. The method comprises the following steps: establishing a binding relationship between a raw material number and a material storage unit, a service range and an authorized material number set of feeding equipment, and a conveying process template containing a target raw material number and a process path in advance, and sequentially executing material storage availability verification, equipment capability verification and material object consistency verification before starting a conveying task each time; and plan-equipment-real object triple closed-loop verification is realized. The three-layer verification mechanism fundamentally solves the typical problems that a traditional system only depends on a single label or manual verification, so that no real material exists, equipment is misused, and the materials are not aligned to a bin, and it is ensured that conveying is allowed to be started only under the conditions that the raw material state is available, the equipment has the processing qualification, and the real material in the bin is completely consistent with a plan; therefore, the risk of wrong materials is eradicated from the source.

Description

Technical Field

[0001] This invention relates to the field of steel raw material transportation, and in particular to a method and electronic device for preventing errors in steel raw material transportation. Background Technology

[0002] In steel production, accurate feeding of raw materials (such as iron ore, sinter, and various ferroalloys) is crucial for ensuring smooth blast furnace operation and steel quality. Traditional raw material conveying systems rely heavily on manual verification of material lists or RFID / barcode-based material identification for error prevention, but these systems have significant drawbacks: Firstly, RFID tags are susceptible to interference from the metallic environment, misalignment of attachment positions, or errors in data entry, leading to discrepancies between the tags and the actual material. Secondly, a large number of similar-looking raw materials (such as different grades of ferrovanadium alloys, or lump ore and sinter with similar particle sizes) are difficult to distinguish visually, easily leading to mixed materials causing excessive composition or even production accidents. Furthermore, in large steel enterprises, due to the collaborative work involving multiple units and levels of personnel, operational errors are frequent, and once a material is mis-fed, it can cause serious quality losses and production interruptions, resulting in huge economic losses.

[0003] In recent years, although there have been attempts to introduce image recognition technology, existing solutions mostly adopt end-to-end classification models, whose output layer is strongly bound to the material number. Whenever a new alloy raw material is added, it is necessary to recollect all samples, modify the network structure, and retrain for a long time, which is difficult to adapt to the dynamic expansion needs of the raw material system in steel plants. At the same time, in practical applications, image recognition technology has limited ability to distinguish the subcategories of raw materials (such as different types of ore), especially when faced with materials with similar particle size and color. It can often only distinguish broad categories (such as "iron ore" vs. "flux"), and cannot effectively distinguish subcategories such as "powder ore", "lump ore" or "pellets", resulting in insufficient recognition accuracy and difficulty in meeting the requirements of high-precision error prevention.

[0004] Furthermore, existing systems generally lack multi-level error prevention mechanisms deeply integrated with production processes, failing to implement differentiated alarm and control strategies based on the risk level of raw material mixing. This results in either frequent false alarms impacting efficiency or missed alarms leading to significant quality risks. Therefore, there is an urgent need for a raw material conveying error prevention method that integrates reliable identification, flexible expansion, and intelligent response to upgrade from "passive error correction" to "active error prevention." Summary of the Invention

[0005] To proactively prevent and control quality and safety risks caused by incorrect part numbers, misuse of equipment, or material confusion during the transportation of steel raw materials, this invention proposes a method for error prevention in steel raw material transportation, comprising:

[0006] Based on the following pre-established foundational data:

[0007] The association between raw material part numbers and raw materials, wherein the raw material part numbers are generated according to coding rules, and each raw material part number corresponds to a steel raw material with a specific chemical composition and physical properties; the association is achieved by storing the raw material in a storage unit bound to the raw material part number;

[0008] The binding relationship between each feeding device and the storage unit it serves;

[0009] The binding relationship between each feeding device and its preset material number set;

[0010] Raw material conveying process template. This template is a preset raw material conveying task configuration data. The configuration data includes the target raw material number, the corresponding storage unit, the feeding device, and the process endpoint device. It is used to define the conveying task of conveying the raw material identified by the target raw material number stored in the storage unit to the process endpoint device through the feeding device bound to the storage unit.

[0011] When starting any raw material conveying process, perform the following error-proofing verification steps:

[0012] S1: Based on the target raw material number of the current raw material conveying process, find the storage unit that is bound to the material number and whose raw material status is available; if it does not exist, prevent the start-up and issue an alarm;

[0013] S2: Based on the found storage unit, determine the feeding device bound to it, and verify whether the preset material number set of the feeding device contains the target raw material material number; if it does not contain it, then prohibit the start-up and issue an alarm;

[0014] S3: Obtain the actual raw material number stored in the storage unit corresponding to the current raw material conveying process; and verify whether the actual raw material number is consistent with the target raw material number; if they are inconsistent, prohibit the start-up and issue an alarm;

[0015] If all the above checks pass, the current raw material delivery process will be started.

[0016] Furthermore, the preset material number set corresponds to the set of raw material numbers that the feeding device is authorized to process;

[0017] The feeding equipment includes fixed feeding equipment and mobile feeding equipment; the storage unit is specifically a silo or a material grid in a C-shaped material yard;

[0018] The fixed feeding device is configured to serve a silo whose preset material number set includes one or more preset raw material material numbers;

[0019] The mobile feeding device is a scraper reclaimer for retrieving materials from a C-type material yard, and its preset material number set includes the material numbers of one or more material cells in the C-type material yard.

[0020] Furthermore, the raw material status can refer to the raw materials having completed quality inspection, not being frozen, and having the quantity meeting process requirements.

[0021] Furthermore, the coding rules include the structural definition of the raw material number, which consists of a raw material category field, a raw material subcategory field, and a sequence number field arranged in a preset order.

[0022] Furthermore, when the feeding device is a mobile feeding device, after performing verification step S2 and before performing verification step S3, a position verification step is also included: obtaining its current traveling position by an encoder or UWB positioning module installed on the mobile feeding device, and determining whether the position is located within the material grid area corresponding to the target storage unit; if not, starting is prohibited; the target storage unit is a storage unit that is bound to the target raw material number of the current raw material conveying process and whose raw material status is available.

[0023] Furthermore, obtaining the actual raw material number of the raw material stored in the storage unit corresponding to the current raw material conveying process specifically involves:

[0024] The information stored in the passive RFID tag attached to the storage unit or its entrance is read by an RFID reader, and the actual raw material number is parsed from it.

[0025] The passive RFID tag is written with the material number bound to the storage unit when the raw materials are put into storage.

[0026] Furthermore, obtaining the actual raw material number of the raw material stored in the storage unit corresponding to the current raw material conveying process specifically involves:

[0027] Images of the raw material pile surface are captured by industrial cameras deployed above or to the side of the storage unit;

[0028] The image is processed using a raw material visual feature extraction model to extract multi-dimensional visual features, and the multi-dimensional visual features are matched with the raw material visual feature-part number mapping database to determine the actual raw material part number.

[0029] The multidimensional visual features include particle size distribution, surface reflectivity, and color histogram. The raw material visual feature-part number mapping database stores the correlation between each raw material part number and its corresponding multidimensional visual features. The multidimensional visual features are generated by processing the standard raw material pile surface image using the raw material visual feature extraction model.

[0030] Furthermore, the method for obtaining the visual feature extraction model of the raw materials is as follows:

[0031] Images of the surface of various steel raw materials were collected under uniform lighting and shooting conditions as image samples, and each image sample was labeled with its corresponding raw material number. The image samples included various raw materials with the same major category field but different minor category fields.

[0032] Positive and negative sample pairs are constructed based on labeled image samples to form a dataset;

[0033] A model for extracting visual features of raw materials is obtained by supervising the training of a convolutional neural network using a metric learning method with a dataset and a metric learning loss function.

[0034] Wherein: positive sample pairs are samples with the same raw material part number, and negative sample pairs are samples with the same raw material category field but different raw material subcategory fields; the supervised training drives the model through a metric learning loss function, so that the multidimensional visual features corresponding to samples with the same raw material part number are close to each other in the feature space, while the multidimensional visual features corresponding to samples with the same raw material category field but different raw material subcategory fields are far apart.

[0035] Further, in step S3, it is verified whether the actual raw material part number is consistent with the target raw material part number; if they are inconsistent, the startup is prohibited, and a multi-level alarm strategy is executed according to the difference level between the actual raw material part number and the target raw material part number in the raw material classification system:

[0036] If the raw material category fields are the same but the raw material sub-category fields are different, a level one alarm will be triggered, the corresponding feeding equipment will be locked, and staff will be notified to handle the situation on-site.

[0037] If the raw material category field is different, a level 2 alarm will be triggered, stopping the material conveying from the storage unit to the process endpoint equipment via the feeding equipment, and notifying the engineer to intervene.

[0038] To address the aforementioned problems, another aspect of the present invention provides an electronic device, including: a processor and a memory storing a program, the program including instructions that, when executed by the processor, cause the processor to perform the method described above.

[0039] To address the aforementioned problems, in another aspect of this invention, a non-transitory machine-readable medium storing computer instructions is provided, the computer instructions being used to cause the computer to perform the methods described above.

[0040] Compared with the prior art, the present invention has at least the following beneficial effects:

[0041] (1) This invention establishes a pre-defined binding relationship between raw material part numbers and storage units, the service scope of feeding equipment and its authorized part number set, and a conveying process template containing the target raw material part number and process path. Before each conveying task is started, storage availability verification, equipment capability verification, and physical consistency verification are performed sequentially, realizing a triple closed-loop verification of plan, equipment, and physical goods. This three-layer verification mechanism fundamentally solves the typical problems caused by traditional systems relying solely on a single label or manual verification, such as "labeled but no actual material," "misuse of equipment," and "material not matching the warehouse." It ensures that conveying is only allowed to start under the conditions that the raw material is available, the equipment is qualified to handle it, and the physical goods in the warehouse are completely consistent with the plan, thereby eliminating the risk of incorrect materials at the source.

[0042] (2) The present invention clearly defines the specific conditions for the availability of raw materials (complete quality inspection, not frozen, sufficient quantity) to ensure that the storage unit selected in step S1 has the actual conditions for use, thereby improving the reliability of the system in complex storage environments.

[0043] (3) This invention introduces coding rules based on raw material category / subcategory fields and adds a position verification step for mobile feeding equipment to ensure that the scraper reclaimer only operates within the target material grid area, prevents cross-grid material reclaiming errors caused by equipment positioning deviation, and improves the error prevention accuracy in dynamic material reclaiming scenarios.

[0044] (4) This invention provides a dual-mode solution for RFID identification and visual identification based on metric learning: RFID is suitable for scenarios where tags are reliable, while visual identification does not rely on physical identification and directly extracts multi-dimensional visual features from the surface image of the raw material pile for matching, which is especially suitable for easily confused raw materials that are similar in appearance but different in category; the metric learning training strategy adopted by this invention enables the model to effectively distinguish between samples of "different subcategories within the same major category", and the addition of a new material number only requires the input of standard multi-dimensional visual features, without the need to retrain the model, which greatly improves the scalability and maintenance efficiency of the system.

[0045] (5) The multi-level alarm strategy proposed in this invention implements differentiated responses based on the difference level (whether the major categories are the same) between the actual raw material number and the target raw material number in the classification system: when the minor categories are inconsistent, only the local equipment is locked and the operators are notified to ensure production continuity; when the major categories are inconsistent, the entire material flow path is cut off and engineers are forced to intervene to prevent high-risk incorrect materials from entering the core process. This hierarchical control mechanism minimizes unnecessary downtime while ensuring safety, achieving an optimal balance between safety and efficiency. Attached Figure Description

[0046] Figure 1 A flowchart of a method for preventing errors in the transportation of steel raw materials;

[0047] Figure 2This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0048] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0049] To proactively control quality and safety risks arising from incorrect part numbers, misuse of equipment, or material mixing during the transportation of steel raw materials, such as... Figure 1 As shown, this invention proposes a method for preventing errors in the transportation of steel raw materials, comprising:

[0050] Based on the following pre-established foundational data:

[0051] The association between raw material part numbers and raw materials, wherein the raw material part numbers are generated according to coding rules, and each raw material part number corresponds to a steel raw material with a specific chemical composition and physical properties; the association is achieved by storing the raw material in a storage unit bound to the raw material part number;

[0052] The coding rules include the structural definition of the raw material number, which consists of a raw material category field, a raw material subcategory field, and a sequence number field in a preset order.

[0053] In one specific embodiment, the raw material part number is a five-digit code, the structure of which includes, in sequence: a reserved space, a raw material category field, a raw material subcategory field, and a sequence number field. Wherein:

[0054] The reserved space is fixed at "0";

[0055] The raw material category field is used to identify the main category of the raw material. For example, "0" represents coal, "1" represents iron ore, "3" represents flux, and "7" represents coke.

[0056] The raw material sub-category field is used to further distinguish material types within the major category. For example, under the iron ore major category, "1" represents iron ore fines, "2" represents lump iron ore, "4" represents pellets, and "5" represents sintered ore. Under the coal major category, "7" represents bituminous coal and "8" represents anthracite.

[0057] The sequence number field consists of two digits and is used to distinguish raw materials from different sources or specifications within the same subcategory. Through this coding rule, each type of steel raw material in the system is assigned a unique part number with semantic information, ensuring that its chemical composition and physical properties are traceable and identifiable.

[0058] The binding relationship between each feeding device and the storage unit it serves;

[0059] The binding relationship between each feeding device and its preset material number set;

[0060] Raw material conveying process template. This template is a preset raw material conveying task configuration data. The configuration data includes the target raw material number, the corresponding storage unit, the feeding device, and the process endpoint device. It is used to define the conveying task of conveying the raw material identified by the target raw material number stored in the storage unit to the process endpoint device through the feeding device bound to the storage unit.

[0061] When starting any raw material conveying process, perform the following error-proofing verification steps:

[0062] S1: Based on the target raw material number of the current raw material conveying process, find the storage unit that is bound to the material number and whose raw material status is available; if it does not exist, prevent the start-up and issue an alarm;

[0063] The raw material status can refer to the raw materials having completed quality inspection, not being frozen, and having sufficient quantity to meet process requirements.

[0064] In actual production, if raw materials have not undergone quality inspection, have quality issues, or the quantity does not meet process requirements, the system will automatically determine them as "unavailable." For such raw materials, this invention verifies the raw material status of the storage unit in step S1 to ensure that only qualified and usable raw materials are allowed to enter the conveying process. Simultaneously, the system supports temporary permission adjustments for specific feeding devices: for example, when a batch of raw materials is frozen due to quality issues, the system can automatically remove the raw material number from the preset number set of the relevant feeding device, or directly lock the corresponding device to prevent incorrect material handling due to misoperation. This mechanism effectively avoids the risk of material mixing caused by abnormal raw materials or loss of device access control, achieving proactive error prevention across the entire chain from raw material access to equipment control.

[0065] If, in step S1, no storage unit is found that is bound to the target raw material number and has an available raw material status (e.g., no raw material with the corresponding number has been received, the raw material has not yet completed quality inspection, the inventory is insufficient, or the batch has been frozen), the system will prohibit the current raw material delivery process from starting and trigger an alarm mechanism. The alarm includes displaying structured alarm information on the operation terminal, the content of which includes at least: the target raw material number, the reason for the absence (e.g., "no available inventory," "quality inspection not completed," or "insufficient inventory"), and suggested operation guidance (e.g., "please confirm whether the raw material has been received and passed quality inspection" or "please select another available storage unit"). At the same time, the event is recorded in the system log, and notifications are pushed to relevant personnel (e.g., team leaders or dispatchers) according to the system's pre-configured notification strategy (e.g., push rules set according to job position and shift) to support rapid response and decision-making.

[0066] This invention clearly defines the specific conditions under which raw materials are usable (completed quality inspection, not frozen, sufficient quantity), ensuring that the storage unit selected in step S1 has the conditions for actual use, thereby improving the reliability of the system in complex storage environments.

[0067] S2: Based on the found storage unit, determine the feeding device bound to it, and verify whether the preset material number set of the feeding device contains the target raw material material number; if it does not contain it, then prohibit the start-up and issue an alarm;

[0068] In this embodiment, if the verification in step S2 reveals that the preset material number set of the feeding device does not include the target raw material number of the current process (e.g., the device is configured only for conveying limestone blocks, while the target material number is sintered ore), the system will prohibit the start of the conveying process and issue a device authorization error alarm. The alarm information is displayed on the operation interface, including: the target raw material number, the currently bound feeding device number, the list of material numbers that the device is allowed to process, and a conflict reason prompt (e.g., device 5-3 is not authorized to process material number 15023); at the same time, the system records the abnormal event and notifies the team leader or process engineer for verification according to the preset permission policy. If it is confirmed that the process configuration is incorrect, the device-material number binding relationship must be adjusted by personnel with the corresponding permissions before the process can be restarted.

[0069] In actual steel production environments, the physical binding relationship between storage units and feeding equipment (such as a vibrating feeder fixedly installed below a silo) only indicates the service scope of the equipment, but does not automatically grant it the authority to handle that material. For example, although a vibrating feeder is located below a limestone block silo, if it needs to be temporarily disabled due to equipment maintenance, or if process adjustments require restricting its use to specific batches of raw materials, its preset material number set must be explicitly configured through the system. Without such logical verification, equipment malfunctions may occur, such as incorrectly starting the conveying of high-volatile bituminous coal during maintenance, posing a safety risk. Therefore, this invention introduces an independent verification of the preset material number set of the feeding equipment in step S2, ensuring that even if the equipment physically serves a storage unit, it must have the authorization to process that material number, thus achieving dual verification from physical ownership to operational authority.

[0070] As a typical application scenario, when a certain type of raw material (such as high-volatile bituminous coal) is temporarily disabled for safety reasons (both disabling and unlocking operations are performed by engineers with management privileges, thus requiring only routine maintenance by management personnel to effectively prevent the risk of incorrect or mixed materials due to misoperation during 24-hour continuous operation in large steel enterprises), the system can exclude it by modifying the preset material number set of the relevant feeding equipment. At this time, even if the equipment is physically connected to the silo storing the raw material, the system will still detect in step S2 that the preset material number set does not contain the target raw material number and will prohibit the start of the conveying process, avoiding incorrect material selection due to human error. This mechanism effectively supports dynamic permission management requirements and significantly improves the system's security and controllability under complex operating conditions.

[0071] The preset material number set corresponds to the set of raw material numbers that the feeding equipment is authorized to process.

[0072] The feeding equipment includes fixed feeding equipment and mobile feeding equipment; the storage unit is specifically a silo or a material grid in a C-shaped material yard;

[0073] The fixed feeding device is configured to serve a silo whose preset material number set includes one or more preset raw material material numbers;

[0074] The mobile feeding device is a scraper reclaimer for retrieving materials from a C-type material yard, and its preset material number set includes the material numbers of one or more material cells in the C-type material yard.

[0075] The C-type stockyard is a common type of enclosed, strip-shaped raw material storage yard. Its cross-section is C-shaped, and its interior is divided into multiple independent compartments by retaining walls for the classified storage of different types of bulk raw materials (such as iron ore powder, pellets, flux, etc.). The compartments are arranged longitudinally along tracks, and a scraper reclaimer travels on these tracks, reclaiming material from designated compartments using a telescopic scraper. Because the materials in adjacent compartments may look similar but have different compositions, misalignment of the reclaiming position or operation across compartments can easily lead to material mixing accidents. Therefore, precise location and material number verification are required for the mobile reclaiming equipment.

[0076] When the feeding device is a mobile feeding device, a position verification step is also included after verification step S2 and before verification step S3:

[0077] The current travel position is obtained by an encoder or UWB positioning module installed on the mobile feeding device, and it is determined whether the position is located within the material grid area corresponding to the target storage unit; if not, the start is prohibited; the target storage unit is a storage unit that is bound to the target raw material number of the current raw material conveying process and whose raw material status is available.

[0078] For mobile feeding equipment (such as scraper reclaimers for C-type material yards), its preset material number set includes the material numbers of the raw materials bound to one or more material grids it serves. When executing the raw material conveying process, the system first determines the target raw material number and the corresponding storage unit (i.e., the target material grid area) according to the process template, and verifies whether the preset material number set of the mobile feeding equipment contains the target raw material number; if it does, it further verifies whether its current physical position is within the target material grid area through its traveling positioning device (such as an encoder or UWB module) to ensure that the equipment is actually in the planned operating position and prevents accidental retrieval across material grids.

[0079] This invention introduces a coding rule based on the raw material category / subcategory field and adds a position verification step for mobile feeding equipment to ensure that the scraper reclaimer only operates within the target material grid area, preventing cross-grid material reclaiming errors caused by equipment positioning deviations and improving the error prevention accuracy in dynamic material reclaiming scenarios.

[0080] S3: Obtain the actual raw material number stored in the storage unit corresponding to the current raw material conveying process; and verify whether the actual raw material number is consistent with the target raw material number; if they are inconsistent, prohibit the start-up and issue an alarm;

[0081] If all the above checks pass, the current raw material delivery process will be started.

[0082] The step of obtaining the actual raw material number stored in the storage unit corresponding to the current raw material conveying process includes two implementation methods:

[0083] The first implementation method is as follows:

[0084] The information stored in the passive RFID tag attached to the storage unit or its entrance is read by an RFID reader, and the actual raw material number is parsed from it.

[0085] The passive RFID tag is written with the material number bound to the storage unit when the raw materials are put into storage.

[0086] In the steel production process, there are many types of raw materials. Among them, many raw materials (such as sintered ore and blast furnace lump ore, different grades of ferroalloys, etc.) have very different chemical compositions and process uses, but they are highly similar in appearance—they all appear as reddish-brown or black granular masses with overlapping particle sizes and similar surface reflective properties. This makes it difficult for traditional image classification-based visual recognition systems to accurately distinguish them, which can easily lead to material mixing accidents. At the same time, existing AI recognition solutions mostly use end-to-end classification models, where the number of neurons in the output layer is tied to the type of material. Whenever a new alloy or auxiliary material (such as special vanadium-nitrogen alloy) is added, all samples must be collected again, the network structure modified, and joint training conducted for several hours. This not only results in high operation and maintenance costs, but also makes it difficult to meet the real-time requirements of steel plants for the dynamic expansion of the raw material system.

[0087] To address this, the present invention provides a second embodiment.

[0088] The second implementation method is as follows:

[0089] Images of the raw material pile surface are captured by industrial cameras deployed above or to the side of the storage unit;

[0090] The image is processed using a raw material visual feature extraction model to extract multi-dimensional visual features, and the multi-dimensional visual features are matched with the raw material visual feature-part number mapping database to determine the actual raw material part number.

[0091] The multidimensional visual features include, but are not limited to, particle size distribution, surface reflectivity, and color histogram. The raw material visual feature-part number mapping database stores the correlation between each raw material part number and its corresponding multidimensional visual features. The multidimensional visual features are generated by the raw material visual feature extraction model by processing the surface image of the standard raw material pile.

[0092] The method for obtaining the visual feature extraction model of the raw materials is as follows:

[0093] Images of the surface of various steel raw materials were collected under uniform lighting and shooting conditions as image samples, and each image sample was labeled with its corresponding raw material number. The image samples included various raw materials with the same major category field but different minor category fields.

[0094] Positive and negative sample pairs are constructed based on labeled image samples to form a dataset;

[0095] A model for extracting visual features of raw materials is obtained by supervising the training of a convolutional neural network using a metric learning method with a dataset and a metric learning loss function.

[0096] Wherein: positive sample pairs are samples with the same raw material part number, and negative sample pairs are samples with the same raw material category field but different raw material subcategory fields; the supervised training drives the model through a metric learning loss function, so that the multidimensional visual features corresponding to samples with the same raw material part number are close to each other in the feature space, while the multidimensional visual features corresponding to samples with the same raw material category field but different raw material subcategory fields are far apart.

[0097] In other words, addressing the problems of the aforementioned end-to-end classification models, this invention proposes a raw material visual feature recognition architecture based on metric learning. First, a general feature extractor is constructed using a convolutional neural network. During the training phase, this network adjusts model parameters by constructing positive and negative sample pairs (positive samples are images with the same part number, and negative samples are easily confused images of raw materials with the same major category but different minor categories). This ensures that feature vectors of similar raw materials are close to each other, while feature vectors of easily confused dissimilar raw materials are far apart. This allows the model to automatically extract multi-dimensional visual features strongly correlated with physical properties (such as particle size distribution, surface reflectivity, and deep representations of color histograms) from raw material images. Crucially, the model outputs fixed-dimensional feature vectors (i.e., multi-dimensional visual features) rather than direct part number labels. Based on this, a "raw material visual feature-part number mapping database" is established to store the standard multi-dimensional visual features corresponding to each raw material part number. When adding new raw materials, only a small number of standard images need to be collected. The features are extracted through forward inference using the deployed model, and the features are associated with the new part number and stored in the database to complete the system expansion. There is no need to modify the model structure or retrain the parameters, thereby achieving accurate error-proof identification of highly similar raw materials and flexible dynamic management of part numbers.

[0098] In the model training process of this invention, although the final deployed raw material visual feature extraction model only outputs multi-dimensional feature vectors representing the visual characteristics of raw materials, rather than directly predicting the raw material part numbers, labeling the training samples with standard raw material part numbers remains an indispensable supervisory prerequisite. Specifically, the part number labels are used to guide the construction of positive and negative sample pairs: the system constructs positive sample pairs based on images with the same part number, allowing the model to map these samples to the neighborhood region in the feature space (i.e., the mathematical space composed of high-dimensional vectors output by the convolutional neural network); simultaneously, negative sample pairs are constructed based on part numbers with the same major category field but different minor category fields, allowing the model to explicitly separate these raw materials that are similar in appearance but different in process use in the feature space. By using a metric learning loss function (such as triplet loss), driven by a large number of labeled samples, the model gradually learns discriminative feature representations that can effectively characterize physical differences such as particle size distribution, surface reflectivity, and color histograms. The convolutional neural network can be a backbone network commonly used in the field (such as ResNet or EfficientNet), and its output is global pooled to form a fixed-dimensional embedding vector (i.e., multi-dimensional visual features) for subsequent similarity matching.

[0099] This invention provides a dual-mode solution for RFID identification and visual recognition based on metric learning: RFID is suitable for scenarios where tags are reliable, while visual recognition does not rely on physical identifiers and directly extracts multi-dimensional visual features from images of raw material piles for matching, which is especially suitable for easily confused raw materials that are similar in appearance but different in category; the metric learning training strategy adopted in this invention enables the model to effectively distinguish between samples of "different subcategories within the same major category", and adding new material numbers only requires inputting standard multi-dimensional visual features without retraining the model, which greatly improves the scalability and maintenance efficiency of the system.

[0100] In step S3, it is verified whether the actual raw material part number is consistent with the target raw material part number; if they are inconsistent, the process is prohibited from starting, and a multi-level alarm strategy is executed based on the difference level between the actual raw material part number and the target raw material part number in the raw material classification system.

[0101] If the raw material category fields are the same but the raw material sub-category fields are different, a level one alarm will be triggered, the corresponding feeding equipment will be locked, and staff will be notified to handle the situation on-site.

[0102] If the raw material category field is different, a level 2 alarm will be triggered, stopping the material conveying from the storage unit to the process endpoint equipment via the feeding equipment, and notifying the engineer to intervene.

[0103] In practice, when it is necessary to change the type of raw material stored in a storage unit (such as a silo or a C-type material yard), operators must first ensure that the original raw materials in the storage unit have been completely emptied. Subsequently, an engineer with management authority must confirm in the system that the storage unit is suitable for receiving the new type of raw material and assign a corresponding material number to the new batch of raw materials to be stored. Only after the above authorization confirmation is completed will the system allow the batch of raw materials bound to the new material number to be stored in the storage unit, and simultaneously update the preset material number set of the relevant feeding equipment. This mechanism effectively prevents the risk of material mixing caused by failure to empty the storage unit or unauthorized material number switching during storage capacity reuse, ensuring that the raw material replacement process is controlled and traceable.

[0104] The multi-level alarm strategy proposed in this invention implements differentiated responses based on the difference level between the actual raw material part number and the target raw material part number in the classification system (whether the major categories are the same): when the minor categories are inconsistent, only the local equipment is locked and the operators are notified to ensure production continuity; when the major categories are inconsistent, the entire material flow path is cut off and engineers are forced to intervene to prevent high-risk incorrect materials from entering the core process. This hierarchical control mechanism minimizes unnecessary downtime while ensuring safety, achieving an optimal balance between safety and efficiency.

[0105] This invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, which, when executed by the at least one processor, causes the electronic device to perform the method of this invention.

[0106] The present invention also provides a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform the method of the present invention.

[0107] This invention also provides a computer program product, including a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform the method of this invention.

[0108] refer to Figure 2 The present invention will now describe a structural block diagram of an electronic device that can serve as a server or client in embodiments of the present invention, which is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0109] like Figure 2 As shown, the electronic device includes a computing unit 401, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 into a random access memory (RAM) 403. The RAM 403 may also store various programs and data required for the operation of the electronic device. The computing unit 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0110] Multiple components in the electronic device are connected to I / O interface 405, including: input unit 406, output unit 407, storage unit 408, and communication unit 409. Input unit 406 can be any type of device capable of inputting information into the electronic device. Input unit 406 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of the electronic device. Output unit 407 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 408 may include, but is not limited to, disks and optical discs. Communication unit 409 allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0111] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, CPUs, graphics processing units (GPUs), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the methods and processes described above. For example, in some embodiments, the method embodiments of the present invention can be implemented as a computer program tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed on an electronic device via ROM 402 and / or communication unit 409. In some embodiments, the computing unit 401 can be configured to perform the methods described above by any other suitable means (e.g., by means of firmware).

[0112] Computer programs for implementing the methods of embodiments of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0113] In the context of embodiments of the present invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable signal medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0114] This invention establishes a pre-defined binding relationship between raw material part numbers and storage units, the service scope of feeding equipment and its authorized part number set, and a conveying process template containing the target raw material part number and process path. Before each conveying task is initiated, it sequentially performs storage availability verification, equipment capability verification, and physical consistency verification, achieving a triple closed-loop verification of plan, equipment, and physical inventory. This three-layer verification mechanism fundamentally solves the typical problems caused by traditional systems relying solely on a single label or manual verification, such as "labeled but no actual material," "equipment misuse," and "material not matching the warehouse." It ensures that conveying is only allowed to start under the conditions that the raw material is available, the equipment is qualified to handle it, and the physical inventory in the warehouse is completely consistent with the plan, thereby eliminating the risk of incorrect material delivery at the source.

[0115] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0116] Furthermore, in this invention, descriptions involving terms such as "first," "second," and "a" are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0117] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0118] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

Claims

1. A method for preventing errors in the conveying of steel raw materials, characterized in that, include: Based on the following pre-established foundational data: The association between raw material part numbers and raw materials, wherein the raw material part numbers are generated according to coding rules, and each raw material part number corresponds to a steel raw material with a specific chemical composition and physical properties; the association is achieved by storing the raw material in a storage unit bound to the raw material part number; The binding relationship between each feeding device and the storage unit it serves; The binding relationship between each feeding device and its preset material number set; Raw material conveying process template. This template is a preset raw material conveying task configuration data. The configuration data includes the target raw material number, the corresponding storage unit, the feeding device, and the process endpoint device. It is used to define the conveying task of conveying the raw material identified by the target raw material number stored in the storage unit to the process endpoint device through the feeding device bound to the storage unit. When starting any raw material conveying process, perform the following error-proofing verification steps: S1: Based on the target raw material number of the current raw material conveying process, find the storage unit that is bound to the material number and whose raw material status is available; if it does not exist, prevent the start-up and issue an alarm; S2: Based on the found storage unit, determine the feeding device bound to it, and verify whether the preset material number set of the feeding device contains the target raw material material number; if it does not contain it, then prohibit the start-up and issue an alarm; S3: Obtain the actual raw material number stored in the storage unit corresponding to the current raw material conveying process; and verify whether the actual raw material number is consistent with the target raw material number; if they are inconsistent, prohibit the start-up and issue an alarm; If all the above checks pass, the current raw material delivery process will be started.

2. The method for preventing errors in the conveying of steel raw materials according to claim 1, characterized in that, The preset material number set corresponds to the set of raw material numbers that the feeding equipment is authorized to process. The feeding equipment includes fixed feeding equipment and mobile feeding equipment; the storage unit is specifically a silo or a material grid in a C-shaped material yard; The fixed feeding device is configured to serve a silo whose preset material number set includes one or more preset raw material material numbers; The mobile feeding device is a scraper reclaimer for retrieving materials from a C-type material yard, and its preset material number set includes the material numbers of one or more material cells in the C-type material yard.

3. The method for preventing errors in the conveying of steel raw materials according to claim 1, characterized in that, The raw material status can refer to the raw materials having completed quality inspection, not being frozen, and having sufficient quantity to meet process requirements.

4. The method for preventing errors in the conveying of steel raw materials according to claim 1, characterized in that, The coding rules include the structural definition of the raw material number, which consists of a raw material category field, a raw material subcategory field, and a sequence number field in a preset order.

5. The method for preventing errors in the conveying of steel raw materials according to claim 2, characterized in that, When the feeding device is a mobile feeding device, after performing verification step S2 and before performing verification step S3, a position verification step is also included: obtaining its current traveling position by an encoder or UWB positioning module installed on the mobile feeding device, and determining whether the position is located within the material grid area corresponding to the target storage unit; if not, starting is prohibited; the target storage unit is a storage unit that is bound to the target raw material number of the current raw material conveying process and whose raw material status is available.

6. The method for preventing errors in the conveying of steel raw materials according to claim 1, characterized in that, The step of obtaining the actual raw material number of the raw material stored in the storage unit corresponding to the current raw material conveying process is as follows: The information stored in the passive RFID tag attached to the storage unit or its entrance is read by an RFID reader, and the actual raw material number is parsed from it. The passive RFID tag is written with the material number bound to the storage unit when the raw materials are put into storage.

7. The method for preventing errors in the conveying of steel raw materials according to claim 4, characterized in that, The step of obtaining the actual raw material number of the raw material stored in the storage unit corresponding to the current raw material conveying process is as follows: Images of the raw material pile surface are captured by industrial cameras deployed above or to the side of the storage unit; The image is processed using a raw material visual feature extraction model to extract multi-dimensional visual features, and the multi-dimensional visual features are matched with the raw material visual feature-part number mapping database to determine the actual raw material part number. The multidimensional visual features include particle size distribution, surface reflectivity, and color histogram. The raw material visual feature-part number mapping database stores the correlation between each raw material part number and its corresponding multidimensional visual features. The multidimensional visual features are generated by processing the standard raw material pile surface image using the raw material visual feature extraction model.

8. The method for preventing errors in the conveying of steel raw materials according to claim 7, characterized in that, The method for obtaining the visual feature extraction model of the raw materials is as follows: Images of the surface of various steel raw materials were collected under uniform lighting and shooting conditions as image samples, and each image sample was labeled with its corresponding raw material number. The image samples included various raw materials with the same major category field but different minor category fields. Positive and negative sample pairs are constructed based on labeled image samples to form a dataset; A model for extracting visual features of raw materials is obtained by supervising the training of a convolutional neural network using a metric learning method with a dataset and a metric learning loss function. Wherein: positive sample pairs are samples with the same raw material part number, and negative sample pairs are samples with the same raw material category field but different raw material subcategory fields; the supervised training drives the model through a metric learning loss function, so that the multidimensional visual features corresponding to samples with the same raw material part number are close to each other in the feature space, while the multidimensional visual features corresponding to samples with the same raw material category field but different raw material subcategory fields are far apart.

9. The method for preventing errors in the conveying of steel raw materials according to claim 1, characterized in that, In step S3, it is verified whether the actual raw material part number is consistent with the target raw material part number; if they are inconsistent, the process is prohibited from starting, and a multi-level alarm strategy is executed based on the difference level between the actual raw material part number and the target raw material part number in the raw material classification system. If the raw material category fields are the same but the raw material sub-category fields are different, a level one alarm will be triggered, the corresponding feeding equipment will be locked, and staff will be notified to handle the situation on-site. If the raw material category field is different, a level 2 alarm will be triggered, stopping the material conveying from the storage unit to the process endpoint equipment via the feeding equipment, and notifying the engineer to intervene.

10. An electronic device, comprising: A processor and a memory storing a program, characterized in that the program includes instructions that, when executed by the processor, cause the processor to perform the error-proofing method for transporting steel raw materials according to any one of claims 1 to 9.

Images