A sintering machine trolley number identification method

Abstract

The present application relates to the technical field of image processing, and more particularly to a sintering machine trolley number recognition method. The method collects trolley images through a camera, positions the trolley wheels, and positions the trolley license plate according to the relative position of the wheels and the license plate. Then, a deep neural network model is used to recognize the license plate image to obtain the corresponding number category, and the current trolley position is determined by comparing with a preset trolley position sequence table. The present application realizes model training and reasoning for trolley number recognition based on deep learning, and combines trolley number recognition based on deep learning with the trolley position sequence table to accurately and efficiently recognize the trolley number, so as to efficiently recognize the trolley number in industrial sintering machine fault detection. Moreover, the present application realizes automatic updating of the trolley position sequence table to reduce the cost of manual maintenance.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method for identifying the number of a sintering machine trolley. Background Technology

[0002] Sintering machine trolleys are widely used in the ironmaking industry. A sintering machine consists of many identical trolleys connected end-to-end, forming a chain that runs in an elliptical closed loop along the trolley's track. Sintering machine trolleys operate under continuous high-temperature, high-load conditions, frequently leading to damage and malfunctions. Therefore, online trolley status monitoring devices are commonly used to monitor the trolleys' status in real time. During online trolley status monitoring, the trolley's position needs to be confirmed in real time, and the currently monitored trolley number needs to be reported to determine which trolley the current status data belongs to. If the real-time position detection of the sintering machine trolley is inaccurate, the online monitoring data will be chaotic, resulting in inaccurate records. Furthermore, precise control of the trolley position is crucial during sintering machine startup, shutdown, and routine maintenance. Incorrect trolley position detection not only affects the sintering machine's production efficiency but also reduces equipment maintenance efficiency. Therefore, in order to ensure the accuracy of online detection data and the precision of trolley position control, it is crucial to correctly identify the trolley's running position throughout the entire operation of the sintering machine. This helps to improve production efficiency and the accuracy of equipment maintenance, and ensures the stable operation of the sintering machine.

[0003] To determine the operating position of the sintering machine trolley, a unique numerical code is typically attached to each trolley. Before the introduction of intelligent online status monitoring devices, monitoring and identifying faulty trolley numbers mainly relied on workers observing the numbers through cameras. However, this method is labor-intensive, and long-term manual monitoring cannot guarantee the accuracy of number identification. Currently, with technological advancements, many methods for identifying license plate numbers have emerged. Existing methods perform poorly in the trolley number identification task of online detection systems in two main ways: First, the harsh environment at the sintering machine operating site causes license plate deterioration and glare, resulting in blurry and interfered license plate images captured by cameras, leading to low discernibility of license plate numbers. Second, most license plate number identification methods are designed for independent images. When performing real-time online trolley number identification, two scenarios arise: real-time display and delayed display. Real-time display of the trolley number directly provides the current trolley number based on the collected trolley image, but the identification results are prone to change during the process, making it impossible to verify the correctness of the identified number. The delayed display of the trolley number involves multiple identifications of the trolley number, with the correct number obtained by statistically analyzing historical identification results. Typically, results are only returned after the trolley has passed through the detection range, making it impossible to display the trolley number in real time. Summary of the Invention

[0004] This invention aims to provide a method for identifying the trolley number of a sintering machine, solving the problem that complex environmental factors can interfere with the identification of trolley numbers during sintering machine fault detection, affecting the detection system's inability to display the currently detected trolley number in real time and leading to incorrect trolley positioning. This method uses a camera to capture images of the trolley, locates the trolley wheels, and determines the license plate number based on the relative position of the wheels and the license plate. Then, a deep neural network model is used to identify the license plate image, obtain the corresponding number category, and compare it with a preset trolley position sequence list to determine the current trolley position.

[0005] The technical solution of the present invention is as follows:

[0006] A method for identifying the number of a sintering machine trolley includes the following steps:

[0007] Step 1: Training method for license plate recognition model

[0008] Images of the trolley are captured by a fixed-position camera. Hough circle detection is performed on the trolley images to locate the wheel positions. Based on the relative positional relationship between the wheels and the parking space numbers, the parking space number image is obtained.

[0009] During the initial training, the parking space number images were manually labeled as either "trolley" or "transitional". Based on the parking space number images and their corresponding labels, a deep neural network was used to train a binary classification model for pre-classification of the parking space number images.

[0010] During the second training iteration, the category labels are corrected based on the pre-classification results. At this stage, the labeled categories include two special numbered labels representing "transitional" and "no license plate," as well as a numerical label representing the vehicle number. Based on the parking space number images and the corresponding pre-classified labeled data, a deep neural network is used for model learning and training to obtain the parking space number recognition model.

[0011] Step 2: License Plate Recognition Method

[0012] (2.1) Establish a trolley parking space sequence table. The trolley parking space sequence table is a list storing trolley numbers. The trolley parking space sequence table reflects the parking space where the trolley is located and its corresponding number. For example, the first element of the list corresponds to the number of the first parking space for the trolley.

[0013] (2.2) Real-time images of the trolley are captured by a fixed camera and images of parking space numbers are obtained. Then, the parking space number recognition model is used to obtain the recognition result of the parking space number image.

[0014] (2.3) The identification results are saved to the numbered short-term storage unit and subjected to comprehensive logic processing. In the comprehensive logic processing, candidate duplicate data of the numbered short-term storage unit are obtained.

[0015] (2.4) If the candidate duplicate data of the numbered short-term storage cell is determined to have not entered the transition region, then return to step (2.2) until the candidate duplicate data of the obtained numbered short-term storage cell is determined to have entered the transition region.

[0016] (2.5) After the candidate duplicate data of the short-term storage unit is determined to have entered the transition area, the car that has captured the trolley image through the camera is taken as the previous car in the two cars corresponding to the transition area; the car that has not captured the trolley image through the camera is taken as the next car in the two cars corresponding to the transition area.

[0017] The target duplicate data of the candidate duplicate data storage unit of the previous car is counted, and this target duplicate data is identified as the number of the previous car. The position of the number is searched in the car position sequence table to determine the position of the previous car. At the same time, the number of the next car is inferred from the car position sequence table. This number will be used for the numbering output of the next car in the cycle that has not entered the transition area.

[0018] Step 3: Automatic update method for the trolley parking space sequence table

[0019] A trolley counter is set up. Each time a trolley passes the camera and the target duplicate data in the candidate duplicate data storage unit is counted, the trolley counter is incremented by 1. When the trolley counter value equals the preset total number of trolleys, the trolley parking space sequence table update method is executed. The specific update method is as follows: First, the historical parking space information data is aligned with the trolley parking space sequence table. Then, the parking space numbers are compared one by one to see if they are the same, and the number of parking spaces with different numbers is counted. If the number of parking spaces is less than a preset threshold, the update is deemed invalid; if the number of parking spaces is greater than the preset threshold, historical parking space information data correction processing is performed. The corrected result will be used as the updated trolley parking space sequence table.

[0020] The beneficial effects of this invention are:

[0021] 1. This invention utilizes deep learning to train and infer a model for trolley number recognition, enabling accurate and efficient identification of trolley numbers. By employing a lightweight deep neural network for model training, the model is made more compact in terms of size and computational complexity, thus achieving efficient trolley number identification in industrial sintering machine fault detection.

[0022] 2. This invention combines deep learning-based trolley number recognition with a trolley parking space sequence list to accurately and stably identify trolley numbers. The model trained through a deep neural network can more effectively handle various factors affecting license plate feature extraction and recognition, such as complex industrial environments, license plate obstruction, and corrosion, thus improving the model's generalization performance in recognizing trolley license plates. Combining the model's trolley number recognition results with the trolley parking space sequence list effectively locates the trolley position, ensuring that only one trolley number appears during the recognition process for each carriage.

[0023] 3. This invention enables automatic updating of the trolley parking space sequence list, reducing the cost of manual maintenance. By storing the numbers identified during trolley operation, the latest trolley number corresponding to each parking space can be obtained. After a trolley completes one cycle of operation, the stored parking space number information is compared with the trolley parking space sequence list, thereby automatically maintaining and updating the trolley parking space sequence list to prevent inconsistencies between the trolley parking space sequence list and the actual situation due to reasons such as license plates falling off. Attached Figure Description

[0024] Figure 1 Flowchart of the method for identifying the serial number of a sintering machine.

[0025] Figure 2 The flowchart for the first stage of the integrated logic processing for sintering machine number identification.

[0026] Figure 3 The flowchart for the second stage of the integrated logic processing for sintering machine number identification.

[0027] Figure 4 A flowchart of a statistical method for candidate duplicate data storage units. Detailed Implementation

[0028] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and technical solutions.

[0029] The flowchart of the sintering machine trolley number identification method provided by the present invention is as follows: Figure 1 As shown, the specific content is as follows:

[0030] (1) Image acquisition and pre-classification of trolley parking space numbers

[0031] To train a deep neural network-based trolley number recognition model, it is necessary to collect images of trolley parking space numbers under different time periods and environmental conditions. Simultaneously, to reduce the workload of manually labeling the number images for model training, image pre-classification is required during image acquisition. Therefore, this project will proceed in two parts: acquiring trolley parking space number images and automatically pre-classifying trolley parking space number images.

[0032] Image acquisition of parking space numbers for trolleys. First, an image of the trolley corresponding to its number is captured from the side using a fixed-position camera. Then, based on the wheel contour information, Hough circle detection is used to detect the wheels in the trolley image, thereby locating the wheel positions. Finally, based on the relative positional relationship between the wheel positions and the parking space numbers, the parking space number image is obtained.

[0033] Specifically, since license plate numbers are uniformly set at specific positions on the trolley, the coordinates of the midpoint representing the parking space number image can be obtained by calculating the midpoint of two adjacent wheels. For example, assuming the center coordinates of the parking space number image in the trolley image are (x, y), the horizontal axis coordinate x can be obtained by averaging the two horizontal axis values ​​of the midpoint coordinates of adjacent wheels. The vertical axis coordinate y can be obtained by averaging the two vertical axis values ​​of the midpoint coordinates of adjacent wheels, and then subtracting the manually preset relative distance between the wheel and the parking space number along the vertical axis. Based on the coordinates of the midpoint of the parking space number image and the pre-set size of the trolley parking space number image, the trolley parking space number image is extracted from the trolley image. In particular, the trolley parking space number image is obtained from the relative positional relationship between the wheels and the license plate. In this example, there are two types of trolley parking space number images: one is an image containing the number collected from the middle position of the trolley, and the other is an image without the number collected from the transition position of adjacent trolleys. Both types of trolley parking space number images are used in the pre-classification of trolley parking space number images.

[0034] Pre-classification of trolley parking space number images is performed as follows: The training task of the trolley number recognition model based on a deep neural network requires the support of trolley parking space number images and corresponding number labels. During the data preparation phase, a large number of collected trolley parking space number images need to be categorized. Pre-classification of trolley parking space number images involves judging two acquisition scenarios and pre-classifying images with the same label for the same scenario. In this example, the two scenarios of trolley parking space number images alternate, meaning the acquired images repeatedly appear in a "trolley-transition-trolley" pattern, assigning a separate label to each location stage. For example, a binary classification model can identify images of the "trolley" or "transition" category. In this example, the images pass through the camera in the order of "trolley-transition-trolley-transition-trolley," and the classification model labels the acquired images with pre-classification labels of "trolley 1-transition 1-trolley 2-transition 2-trolley 3." Ultimately, images in the pre-classification labels representing "transition" are all grouped into one category, while the pre-classification labels representing "cart" are merged periodically based on the total number of carts. Specifically, the pre-classification labels representing "cart" include cases without license plates. The pre-classification processing of the cart parking space number images ultimately yields cart parking space number images and their corresponding labels for training the license plate recognition model. The labels for the cart parking space number images include numeric labels (N classes) that match the numbers in the cart parking space number images, and special number labels (2 classes) representing no license plate and transitional phases.

[0035] (2) License plate recognition model training

[0036] A lightweight convolutional neural network, ResNet18, was used to train the license plate recognition model. To ensure the generalization ability of the recognition model, data augmentation strategies such as random image blurring, random displacement, and random brightness variation were employed to increase the diversity of training samples.

[0037] (3) Comprehensive logic processing

[0038] The comprehensive logic processing method is based on the parking space number list and the historical license plate number recognition results. The recognition results obtained by the license plate recognition model are processed through logical relationships to obtain the final output license plate number.

[0039] The flowchart of the first stage of the synthesis logic processing is as follows: Figure 2As shown, the system's runtime can be divided into three parts. The first part, when the runtime is less than a preset value, involves the integrated logic processing directly using the input as the output. The second part, when the runtime equals the preset value, uses the value with the most repetitions in the short-term number storage unit as the candidate duplicate data and outputs it as the current vehicle's number. The short-term storage unit is a doubly linked list; when the list is full, data is stored using a first-in, first-out (FIFO) rule. The number storage unit stores number data using a FIFO rule, used to store the numbering results identified within the most recent time period. The third part, when the runtime exceeds the preset value, if the candidate duplicate data is not equal to the input license plate number, uses the current vehicle's number from the previous frame as the output; if the candidate duplicate data equals the input license plate number change, the system enters the second stage of integrated logic processing.

[0040] The second stage of synthetic logic processing, such as Figure 3 As shown. The goal of the second stage of the integrated logic processing is to determine the number of the previous carriage by statistically analyzing the target duplicate data in the candidate duplicate data storage unit of the previous carriage. The candidate duplicate data storage unit is a list that stores the carriage number recognition results (candidate duplicate data). Then, by querying the carriage number list, the next carriage's license plate number is obtained, and this number will be output as the current carriage number for the license plate integrated logic processing. Specifically, the second stage of the integrated logic processing determines whether the recognition of the previous carriage is complete by checking whether the candidate duplicate data in the short-term number storage unit is the transition label "00". If the candidate duplicate data is not equal to the transition label, the carriage operation status is set to "normal", indicating that the currently recognized number belongs to the same carriage as the number in the previous frame. In this case, the number is not updated, and the output current carriage number is the current carriage number obtained in the previous frame. If the candidate duplicate data is equal to the transition label, it is determined whether the carriage operation status is "normal" and whether the license plate position meets the preset criteria. When the conditions are met, the carriage operation status is set to "transition", and the number of the previous carriage is statistically analyzed according to the rules. Then, the current carriage number is retrieved as the output.

[0041] The trolley operation status indicates whether the target to be processed belongs to a carriage or is in a transition between carriages, represented by "normal" and "transition" statuses respectively. During trolley license plate detection, the trolley operation status will repeatedly switch between "normal-transition-normal" patterns as the trolley moves. The trolley operation status prevents repeated updates of the current trolley number during the same transition phase.

[0042] The preset license plate position is to ensure that when the trolley is in the "transition" state, the position of the current target to be processed passes through the center line, and at this time it is determined that the target being detected is a new trolley.

[0043] The method for counting the previous trolley number according to the rules is as follows: Figure 4 As shown. First, it is necessary to statistically analyze the most frequently repeated value (target duplicate data), the proportion of the target duplicate data among all values ​​(frequency of occurrence of the target duplicate data), and the number of non-duplicate values ​​(number of number types) in the candidate duplicate data storage unit. Based on the statistical data, the number of the previous carriage will be obtained according to priority rules. The first priority stipulates that when the target duplicate data is consecutive to the latest value of the license plate recording unit, the target duplicate data is the number of the previous carriage. The second priority stipulates that when the frequency of occurrence of the target duplicate data is greater than a preset value, and the number of number types is less than a preset value, the target duplicate data is the number of the previous carriage. If the above conditions are not met, the latest valid value is directly retrieved from the parking space recording unit, and the parking space number of the previous carriage is inferred based on the number of carriages in the interval, and the parking space number is used as the number of the previous carriage for output.

[0044] After the statistical method for the previous trolley number is completed, the license plate number and parking space number of the previous trolley will be saved to the license plate record unit and parking space record unit for subsequent trolley number identification.

[0045] (4) Automatic update of the trolley parking space sequence table

[0046] Because license plates can easily fall off or be manually moved during sintering machine operation, the trolley position sequence list must always be updated to the latest version. To reduce the manual cost of correcting the trolley position sequence list, it is automatically updated based on historical trolley number identification results.

[0047] Specifically, a trolley counter is set up. Each time a trolley passes the camera and the target duplicate data in the candidate duplicate data storage unit is counted, the trolley counter is incremented by 1. When the trolley counter value equals the preset total number of trolleys, historical parking space information is used as input to execute the trolley parking space sequence table update method. First, the historical parking space information data is aligned with the trolley parking space sequence table. Then, the parking space numbers are compared one by one to see if they are the same, and the number of parking spaces with different numbers is counted. If the number of parking spaces is less than a preset threshold, the update is deemed invalid; if the number of parking spaces is greater than the preset threshold, historical parking space information data correction processing is performed. The corrected result will be used as the updated trolley parking space sequence table.

[0048] Historical parking space information: Parking space information is obtained each time the target duplicate data of the historical identification number results are counted. The parking space information will be stored in the historical parking space information list, and its storage data and format are the same as the trolley parking space sequence table, used to indicate the position of adjacent trolley numbers.

[0049] Aligning historical parking space information data with the trolley parking space sequence list: Since the first number in the historical parking space information data may be any value of the trolley number and may not be the same as the first number in the trolley parking space sequence list, it is necessary to align the historical parking space information data with the trolley parking space sequence list for information matching. Specifically, find the position of the first number in the trolley parking space sequence list in the historical parking space information data. If the number does not exist, find the position of the second number in the trolley parking space sequence list, and so on, until the Nth number can be found in the historical parking space information data. After finding the position of the Nth number in the historical parking space information list, find the position N-1 positions before the Nth position. Cut the historical parking space information list into two parts before this position. Align the part containing the Nth number with the trolley parking space sequence list from the first position, and append the remaining part to it. For example, assume the historical parking space information list is 6,5,4,3,2,1,10,7; and the trolley parking space sequence list is 9,8,7,6,5,4,3,2. Perform alignment according to the following steps:

[0050] a. The third number "7" in the trolley parking space sequence list can be found in the historical parking space information list;

[0051] b. Locate position "1" two positions before the position of "7" in the historical parking space information list;

[0052] c. Truncate the historical parking space information list before “1” into two parts: 1, 10, 7 and 6, 5, 4, 3, 2;

[0053] d.1,10,7 are aligned with the first position of the trolley parking space sequence list, and the remaining parts are appended to it.

[0054] e. The aligned list of historical parking space information is 1, 10, 7, 6, 5, 4, 3, 2.

[0055] Parking space number comparison: If the number of a historical parking space is the same as the number in the trolley parking space sequence table, or if the number of a historical parking space is the same as the number in the trolley parking space sequence table, then the parking space numbers are considered to be the same; otherwise, the parking space numbers are considered to be different.

[0056] Historical parking space information data correction and processing: To address the issues of duplicate and incorrect parking space numbers in historical parking space information data, a data correction method is used for rectification. Specifically, for duplicate numbers, if the duplicate number is the same as the corresponding position number in the trolley parking space sequence table, the number is retained; if the duplicate number is different from the corresponding position number in the trolley parking space sequence table, it is corrected to a parking space number. For incorrect parking space numbers, all parking space number data in the historical parking space information is corrected to prevent inconsistencies between the data and the actual parking space numbers.

Claims

1. A method for identifying the number of a sintering machine trolley, characterized in that, Includes the following steps: Step 1: Training method for license plate recognition model Images of the trolley are captured by a fixed-position camera. Hough circle detection is performed on the trolley images to locate the wheel positions. Based on the relative positional relationship between the wheels and the parking space number, the parking space number image is obtained. During the first training, the parking space number images were manually labeled as either "trolley" or "transitional". Based on the parking space number images and the corresponding labeled data, a binary classification model was trained using a deep neural network for pre-classification processing of the parking space number images. During the second training, the category labels are corrected based on the results of the pre-classification. At this time, the labeled categories are two special number labels representing "transition" and "no license plate", as well as a number label representing the trolley number. Based on the parking space number image and the corresponding labeled data after pre-classification, a deep neural network is used for model learning and training to obtain the parking space number recognition model. Step 2: License Plate Recognition Method (2.1) Establish a trolley parking space sequence table; wherein, the trolley parking space sequence table is a list that stores trolley numbers; the trolley parking space sequence table can reflect the parking space where the trolley is located and its corresponding number; (2.2) Real-time images of the trolley are captured by a fixed camera, and images of parking space numbers are obtained. Then, the parking space number recognition model is used to obtain the recognition result of the parking space number images. (2.3) The identification results are saved to the numbered short-term storage unit and subjected to comprehensive logic processing. In the comprehensive logic processing, the candidate duplicate data of the numbered short-term storage unit are obtained. (2.4) If the candidate duplicate data of the numbered short-term storage cell is determined to have not entered the transition region, then return to step (2.2) until the candidate duplicate data of the obtained numbered short-term storage cell is determined to have entered the transition region; (2.5) After the candidate duplicate data of the short-term storage unit is determined to have entered the transition area, the car that has captured the trolley image through the camera is taken as the previous car in the two cars corresponding to the transition area; the car that has not captured the trolley image through the camera is taken as the next car in the two cars corresponding to the transition area. The target duplicate data of the candidate duplicate data storage unit of the previous car is counted and identified as the number of the previous car; the position of the number is searched from the car position sequence table to determine the position of the previous car; at the same time, the number of the next car is inferred from the car position sequence table. This number will be used for the output of the number of the next car in the cycle that has not entered the transition area. The synthesis logic processing described in step 2 includes two stages, as follows: Phase 1: When the running time is less than the preset value, the integrated logic processing directly uses the input as the output; when the running time equals the preset value, the value with the most repetitions in the short-term storage unit is taken as the candidate duplicate data and output as the current vehicle's number; the short-term storage unit is a doubly linked list, and when the list is full, data is stored according to the first-in-first-out rule; the number short-term storage unit saves number data according to the first-in-first-out rule, used to store the numbering results identified within a certain period of time; when the running time exceeds the preset value, if the candidate duplicate data is not equal to the input license plate number, the number of the current vehicle in the previous frame is used as the output; if the candidate duplicate data is equal to the input license plate change, then the second phase of integrated logic processing begins. The second stage aims to determine the number of the previous carriage by statistically analyzing the target duplicate data in the candidate duplicate data storage unit of the previous carriage. The candidate duplicate data storage unit is a list that stores the carriage number recognition results. Then, by querying the carriage number list, the next carriage's license plate number is obtained, and this number will be output as the current carriage number for the license plate comprehensive logic processing. Specifically, the second stage of the comprehensive logic processing uses whether the candidate duplicate data in the short-term number storage unit is the transition label "00" to determine if the recognition of the previous carriage is complete. If the candidate duplicate data is not equal to the transition label, the carriage operation status is set to "normal," indicating that the currently recognized number belongs to the same carriage as the previous frame's number. In this case, the number is not updated, and the output current carriage number is the current carriage number obtained in the previous frame. If the candidate duplicate data is equal to the transition label, it is determined whether the carriage operation status is "normal" and whether the license plate position meets the preset criteria. When the conditions are met, the carriage operation status is set to "transition," and the number of the previous carriage is statistically analyzed according to the rules. Then, the current carriage number is retrieved as the output. Step 3: Automatic update method for the trolley parking space sequence table A trolley counter is set up. Each time a trolley passes the camera and the target duplicate data in the candidate duplicate data storage unit is counted, the trolley counter is incremented by 1. When the trolley counter value equals the preset total number of trolleys, the trolley parking space sequence table update method is executed. The specific update method is as follows: First, the historical parking space information data is aligned with the trolley parking space sequence table. Then, the numbers of each parking space are compared one by one to see if they are the same, and the number of parking spaces with different numbers is counted. If the number of parking spaces is less than a preset threshold, the update is deemed invalid. If the number of parking spaces is greater than the preset threshold, historical parking space information data correction processing is performed. The corrected result will be used as the updated trolley parking space sequence table. The method for calculating the previous car number according to the rules is as follows: First, it is necessary to calculate the value with the most repetitions in the candidate duplicate data storage unit, the proportion of the target duplicate data among all values, and the number of non-repetitive values. Based on the statistical data, the number of the previous car will be obtained according to priority rules. The first priority stipulates that when the target duplicate data is consecutive to the latest value of the license plate recording unit, the target duplicate data is the number of the previous car. The second priority stipulates that when the frequency of the target duplicate data is greater than a preset value and the number of number types is less than a preset value, the target duplicate data is the number of the previous car. If the above conditions are not met, the latest valid value is directly retrieved from the parking space recording unit, and the parking space number of the previous car is inferred based on the number of cars in the interval, and the parking space number is used as the number of the previous car for output. After the statistical method for the previous trolley number is completed, the license plate number and parking space number of the previous trolley will be saved to the license plate record unit and parking space record unit for subsequent trolley number identification.

2. The sintering machine trolley number identification method according to claim 1, characterized in that, In step 3, Aligning historical parking space information data with the trolley parking space sequence list: Since the first number in the historical parking space information data may be any value of the trolley number and may not be the same as the first number in the trolley parking space sequence list, it is necessary to align the historical parking space information data with the trolley parking space sequence list for information matching. Specifically, find the position of the first number in the trolley parking space sequence list in the historical parking space information data. If the number does not exist, find the position of the second number in the trolley parking space sequence list, and so on, until the Nth number can be found in the historical parking space information data. Find the position of the Nth number in the historical parking space information list, find the position N-1 positions before the Nth position, and cut the historical parking space information list into two parts before this position. Align the part with the Nth number with the trolley parking space sequence list from the first position, and append the remaining part to the end. Parking space number comparison: If the number of a historical parking space is the same as the number in the trolley parking space sequence table, or if the number of a historical parking space is the same as the number in the trolley parking space sequence table, then the parking space numbers are considered to be the same; otherwise, the parking space numbers are considered to be different. Historical parking space information data correction processing: To address the issues of duplicate and incorrect parking space numbers in historical parking space information data, a data correction method is used for rectification. Specifically, for duplicate numbers, if the duplicate number is the same as the number in the same position in the trolley parking space sequence table, the number is retained; if the duplicate number is different from the number in the same position in the trolley parking space sequence table, it is corrected to a parking space number. For incorrect parking space numbers, all parking space number data in the historical parking space information is corrected to prevent inconsistencies between the parking space number data and the actual parking space numbers.

Images