Method and device for identifying adjustment state of underground coal mine sensor

Abstract

The invention discloses a coal mine underground sensor adjustment state identification method and device, and relates to the technical field of coal mine safety monitoring. The method comprises the following steps: selecting time window data before and after sensor alarm and preprocessing; grouping according to sensor types; generating a standardized data curve image and marking adjustment features; utilizing an image recognition model (such as YOLOv8) to train a special adjustment feature recognition model for each type of sensors; in practical application, real-time alarm data is converted into an image, and a corresponding model is called for identification, so that whether the alarm is caused by adjustment operation or not is judged. According to the method, the problem of false alarm of the coal mine safety monitoring system caused by adjustment is effectively solved, automatic and high-precision identification of the adjustment state is realized, the manual workload is reduced, and the intelligent level and the production efficiency of coal mine safety management are improved.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety production monitoring technology, specifically to a method and device for recognizing the calibration status of underground sensors in coal mines based on image recognition technology. Background Technology

[0002] Underground safety monitoring systems in coal mines are a crucial link in ensuring safe production. Their core components, various sensors (such as methane, wind speed, and temperature sensors), require calibration at prescribed intervals to ensure the accuracy and reliability of monitoring data. Current standards (such as AQ1029-2019) clearly stipulate the calibration cycles and methods for sensors based on different principles. For example, methane sensors based on carrier catalysis require calibration every 15 days, while laser-based methane sensors require calibration every 6 months. Calibration typically involves verifying parameters such as zero point, displayed value, and alarm point.

[0003] In practice, due to the complex underground environment and the difficulty in ensuring standardized operation by personnel, the calibration process itself may trigger alarms in the monitoring system, generating a large number of false alarms that do not reflect actual dangers, seriously disrupting the normal production order of the coal mine. Currently, the identification of calibration operations largely relies on manual review of data curves for judgment. This method is inefficient, highly subjective, and prone to errors, failing to meet the needs of modern intelligent management in coal mines.

[0004] Existing technologies include methods for calibration and identification based on direct analysis of sensor data features (such as statistical features and waveform features). However, such methods face significant challenges: different types of sensors (such as gas, wind speed, and temperature) have vastly different calibration specifications and data characteristics, requiring the design and adjustment of feature extraction algorithms and model parameters for each type of sensor. This results in high development and maintenance costs, poor generalization ability, and difficulty in large-scale application.

[0005] Therefore, there is an urgent need for an automated, high-precision calibration and identification method that can adapt to various sensor types, in order to effectively distinguish between real alarms and calibration alarms, reduce false alarms, and improve the intelligence level of coal mine safety monitoring systems. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for identifying the calibration status of sensors in underground coal mines based on image recognition. This method converts sensor data into images and uses a computer vision model to automatically identify the calibration patterns contained in the data curves, thereby achieving rapid, accurate, and automated calibration status identification and effectively reducing false alarms caused by calibration operations.

[0007] According to a first aspect of the present invention, a method for identifying the calibration status of sensors in underground coal mines is provided, comprising: In response to a sensor alarm event, extract data from the time window before and after the event, and generate a data curve image based on the sensor type; The system calls the model parameters corresponding to the sensor type and uses a pre-trained image recognition model to recognize the generated image. If the calibration features are recognized, it is marked as a calibration alarm; otherwise, it is marked as a normal alarm.

[0008] Furthermore, the training process of the image recognition model specifically includes: Based on alarm records, sensor status, or work logs, select data sequences within a specific time window before and after a sensor alarm event; Filter out invalid values ​​in the data sequence to obtain valid data; Based on the type of sensor, the valid data is grouped into multiple independent datasets; For each type of sensor dataset, a unified range for displaying the vertical axis is set, and each data sequence is converted into a two-dimensional curve to generate a standardized image; The generated curve images are labeled to identify the calibration feature bands, and a labeled image dataset is constructed according to sensor type; The dataset is divided into training, validation and test sets. The image recognition model is used for training and parameter optimization until the model performance reaches the preset index. Save the optimized model and parameters, and establish a mapping relationship between sensor type and model parameters.

[0009] Furthermore, invalid values ​​include specific numerical codes used to characterize signal interruptions.

[0010] Furthermore, a unified range for displaying the vertical coordinate is set, specifically including: increasing the set ratio as the maximum value based on the upper limit of the calibration range of this type of sensor.

[0011] According to a second aspect of the present invention, a device for identifying the calibration status of sensors in underground coal mines is provided, comprising: The real-time identification and processing module is used to respond to sensor alarm events, extract data from the time window before and after the event, and generate data curve images according to the sensor type. The calibration status discrimination module is used to call the model parameters of the corresponding sensor type and use a pre-trained image recognition model to recognize the generated image. If calibration features are recognized, it is marked as a calibration alarm; otherwise, it is marked as a normal alarm.

[0012] Furthermore, the device also includes a model training module, specifically comprising: The sample selection unit is used to select data sequences within a specific time window before and after a sensor alarm event based on alarm records, sensor status, or work logs. The data preprocessing unit is used to filter out abnormal and invalid values ​​in the data sequence to obtain valid data. Sensor type grouping unit, used to group valid data into multiple independent datasets according to the type of sensor; The curve image generation unit is used to set a uniform display range for the vertical axis for each type of sensor dataset, convert each data sequence into a two-dimensional curve graph, and generate a standardized image. The annotation and construction unit is used to annotate the generated curve images, identify the calibration feature bands, and construct the annotated image dataset according to the sensor type. The training and optimization unit is used to divide the dataset into training, validation and test sets, and to train and optimize the parameters of the image recognition model until the model performance reaches the preset index. The model deployment unit is used to save the optimized model and parameters and establish a mapping relationship between sensor types and model parameters.

[0013] Furthermore, invalid values ​​include specific numerical codes used to characterize signal interruptions.

[0014] Furthermore, the curve image generation unit sets a uniform vertical coordinate display range, specifically including: increasing the set ratio as the maximum value based on the upper limit of the calibration range of this type of sensor.

[0015] According to a third aspect of the present invention, a terminal device is provided, comprising: Processor; and A memory that stores executable code, which, when executed by the processor, causes the processor to perform the method described above.

[0016] According to a fourth aspect of the present invention, a non-transitory machine-readable storage medium is provided, on which executable code is stored, which, when executed by a processor of an electronic device, causes the processor to perform the method described above.

[0017] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: Automation and high efficiency: It enables automatic identification of sensor calibration status, greatly reducing the workload of manual curve inspection and judgment, and improving identification efficiency.

[0018] High accuracy: Utilizing a deep learning image recognition model, it has a strong ability to capture complex data waveform features during the calibration process, resulting in high recognition accuracy (high recall and precision), effectively reducing false positives and false negatives.

[0019] Strong generalization and easy scalability: By employing a "type-specific training" strategy, a dedicated model is built for each type of sensor, avoiding interference from differences in data characteristics between different sensors. The model is highly targeted and performs well. When adding a new sensor type, only a separate dataset and training model need to be built for it, without modifying the overall framework, resulting in good system scalability.

[0020] Highly practical: It can directly generate images using existing monitoring data without modifying existing hardware, resulting in low implementation costs and easy integration into existing safety monitoring systems in coal mines.

[0021] Improving safety management: By effectively filtering and adjusting false alarms, the alarm information of the coal mine safety production monitoring system is made more accurate and reliable, reducing unnecessary production interruptions and improving the intelligence level and production efficiency of coal mine safety management.

[0022] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0023] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the invention.

[0024] Figure 1 This is a flowchart illustrating a method for identifying the calibration status of sensors in underground coal mines according to an exemplary embodiment of the present invention. Figure 2 This is a schematic diagram of data curves generated during a sensor calibration process in the embodiment. Figure 3 This is a structural block diagram of a coal mine underground sensor calibration status identification device according to an exemplary embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a computing device according to an exemplary embodiment of the present invention. Detailed Implementation

[0025] Preferred embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0026] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0027] It should be understood that although the terms "first," "second," "third," etc., may be used in this invention to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this invention, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, features defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0028] The technical solutions of the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0029] Figure 1 This is a flowchart illustrating a method for identifying the calibration status of sensors in underground coal mines according to an exemplary embodiment of the present invention.

[0030] See Figure 1 The method includes: S1. Based on alarm records, sensor status, or work logs, select the data sequence within a specific time window before and after the sensor alarm event.

[0031] Specifically, from historical monitoring data, raw data sequences within a specific time window (e.g., half an hour) before and after the time a sensor alarm occurs are selected as candidate samples; the sample sources include: a) directly extracted from the alarm record time point; b) extracted from the sensor status identification information; c) extracted from the time point recorded in the downhole worker's calibration work log.

[0032] S2. Filter out invalid values ​​in the data sequence to obtain valid data.

[0033] Specifically, step S2 cleans the candidate sample data selected in step S1, filtering out invalid data points that represent signal interruptions or abnormalities (such as -9999, -9998, etc.), and retaining valid normal monitoring data.

[0034] S3. Based on the type of sensor, the valid data is grouped into multiple independent datasets.

[0035] Specifically, step S3 groups the cleaned data samples according to the type of sensor (such as methane sensor, wind speed sensor, temperature sensor, etc.) to form multiple independent datasets corresponding to the sensor type.

[0036] S4. For each type of sensor dataset, set a unified display range for the vertical axis, convert each data sequence into a two-dimensional curve, and generate a standardized image.

[0037] Specifically, in step S4, for each type of sensor dataset in step S3, the display range of the image's vertical axis (Y-axis) is set according to the upper limit of the calibration range of that type of sensor (for example, the upper limit is increased by 20% as the maximum value of the vertical axis). The data sequence of each sample is converted into a continuous two-dimensional curve, and the curves of all similar sensors are uniformly placed under the same coordinate system standard to generate a standardized data curve image before and after the event.

[0038] S5. Label the generated curve images, identify the calibration feature bands, and construct a labeled image dataset according to sensor type.

[0039] Specifically, step S5 involves manually or semi-automatically annotating the curve images of various sensors generated in step S4, identifying typical waveform segments in the images that represent the transition from "calibration start" to "calibration end"; and constructing datasets containing the annotated images according to sensor type.

[0040] S6. Divide the dataset into training, validation and test sets, and use the image recognition model for training and parameter optimization until the model performance reaches the preset index.

[0041] Specifically, the dataset obtained in step S5 is divided into training, validation, and test sets (e.g., 70%, 15%, 15%) according to the proportions; an image recognition model (such as YOLOv8) is selected for training; the training objective is to enable the model to identify the calibration feature bands from the curve image; the model hyperparameters are optimized through the validation set, and the models for different sensor types are independently fine-tuned; the model performance must meet the preset indicators, such as recall ≥98%, precision ≥95%, F1 score ≥0.96, and the performance fluctuation should be less than 2% after multiple consecutive training sessions.

[0042] S7. Save the optimized model and parameters, and establish a mapping relationship between sensor type and model parameters.

[0043] S8. In response to a sensor alarm event, extract the data before and after the event and generate a data curve image based on the sensor type.

[0044] Specifically, for real-time or newly generated sensor alarm events, data from the same time window before and after the alarm time are extracted, preprocessed, and then the corresponding model parameters from step S7 are called according to the type of the sensor to generate a data curve image.

[0045] S9. Call the model parameters of the corresponding sensor type, use the pre-trained image recognition model to recognize the generated image, and mark it as a calibration alarm if the calibration feature is recognized, otherwise mark it as a normal alarm.

[0046] Specifically, the image recognition model trained in step S7 is used to load the corresponding parameters and recognize the curve image generated in step S8. If the model recognizes that the image contains calibration features, it is determined that the alarm was caused by calibration operation and is marked and filtered. If no calibration features are recognized, it is determined to be a normal alarm and needs to be processed according to the procedure.

[0047] The following explanation uses the calibration of a methane sensor in a coal mine as an example.

[0048] 1. Data Acquisition and Preprocessing: Extract all methane sensor alarm records from the monitoring database for the past year, taking 30 minutes of data samples before and after each alarm. Use a program script to automatically filter out abnormal and invalid values ​​(-9999, -9998, etc.) from the data.

[0049] 2. Grouping and Image Generation: Label all sample data as "Methane Sensor". Set the upper limit of the ordinate of the image to 1.2 times the commonly used upper limit for methane calibration (e.g., 2.5% CH4) (i.e., 3.0% CH4). Use a plotting library (such as Matplotlib) to automatically plot the data sequence of each sample into a line graph and save it as an image file (e.g., PNG format). Figure 2 As shown.

[0050] 3. Dataset Construction: Technicians familiar with the calibration process manually annotated the typical calibration waveform segments that showed a significant increase, stabilization, and subsequent decrease in concentration due to the introduction of a standard gas sample. Positive samples were images containing this waveform segment, while negative samples were images of common faults or interference alarms that did not contain this waveform segment. The positive and negative samples together constituted the training dataset for the methane sensor.

[0051] 4. Model Training: The YOLOv8 model architecture was adopted. The methane sensor dataset was randomly divided into training, validation, and test sets in a 70:15:15 ratio. The model was trained using the training set, and hyperparameters (such as learning rate and batch size) were adjusted based on the performance on the validation set. The final model achieved a recall of 98.5%, precision of 96.2%, and an F1 score of 0.973 on the test set, with performance fluctuations of less than 1.5% over five consecutive training cycles, meeting the preset requirements.

[0052] 5. Deployment and Application: Deploy the trained optimal methane sensor calibration status recognition model to the coal mine monitoring system server. When a methane sensor alarm occurs in the system, this recognition program is automatically triggered: extracting data from 30 minutes before and after the alarm, cleaning the data, generating a graph, and calling the methane sensor calibration status recognition model for identification. If calibration characteristics are identified, the alarm is marked as "caused by calibration" in the alarm management system, and no emergency response procedure is pushed; if no characteristics are identified, it is treated as a normal alarm.

[0053] The present invention provides a method for identifying the calibration status of sensors in underground coal mines, which can achieve the following beneficial effects. Automation and high efficiency: It enables automatic identification of sensor calibration status, greatly reducing the workload of manual curve inspection and judgment, and improving identification efficiency.

[0054] High accuracy: Utilizing a deep learning image recognition model, it has a strong ability to capture complex data waveform features during the calibration process, resulting in high recognition accuracy (high recall and precision), effectively reducing false positives and false negatives.

[0055] Strong generalization and easy scalability: By employing a "type-specific training" strategy, a dedicated model is built for each type of sensor, avoiding interference from differences in data characteristics between different sensors. The model is highly targeted and performs well. When adding a new sensor type, only a separate dataset and training model need to be built for it, without modifying the overall framework, resulting in good system scalability.

[0056] Highly practical: It can directly generate images using existing monitoring data without modifying existing hardware, resulting in low implementation costs and easy integration into existing safety monitoring systems in coal mines.

[0057] Improving safety management: By effectively filtering and adjusting false alarms, the alarm information of the coal mine safety production monitoring system is made more accurate and reliable, reducing unnecessary production interruptions and improving the intelligence level and production efficiency of coal mine safety management.

[0058] Figure 3 This is a structural block diagram of a sensor calibration status identification device in a coal mine, as illustrated in an exemplary embodiment of the present invention.

[0059] See Figure 3 The device includes: The real-time identification and processing module is used to respond to sensor alarm events, extract data from the time window before and after the event, and generate data curve images according to the sensor type. The calibration status discrimination module is used to call the model parameters of the corresponding sensor type and use a pre-trained image recognition model to recognize the generated image. If calibration features are recognized, it is marked as a calibration alarm; otherwise, it is marked as a normal alarm.

[0060] Optionally, such as Figure 3 As shown, in this embodiment, the device further includes: a model training module, specifically comprising: The sample selection unit is used to select data sequences within a specific time window before and after a sensor alarm event based on alarm records, sensor status, or work logs. The data preprocessing unit is used to filter out abnormal and invalid values ​​in the data sequence to obtain valid data. Sensor type grouping unit, used to group valid data into multiple independent datasets according to the type of sensor; The curve image generation unit is used to set a uniform display range for the vertical axis for each type of sensor dataset, convert each data sequence into a two-dimensional curve graph, and generate a standardized image. The annotation and construction unit is used to annotate the generated curve images, identify the calibration feature bands, and construct the annotated image dataset according to the sensor type. The training and optimization unit is used to divide the dataset into training, validation and test sets, and to train and optimize the parameters of the image recognition model until the model performance reaches the preset index. The model deployment unit is used to save the optimized model and parameters and establish a mapping relationship between sensor types and model parameters.

[0061] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated further here.

[0062] Figure 4 This is a schematic diagram of the structure of a computing device according to an exemplary embodiment of the present invention.

[0063] See Figure 4 The computing device 400 includes a memory 410 and a processor 420.

[0064] The processor 420 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0065] Memory 410 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. ROM may store static data or instructions required by processor 420 or other modules of the computer. Permanent storage devices may be read-write storage devices. Permanent storage devices may be non-volatile storage devices that retain stored instructions and data even when the computer is powered off. In some embodiments, permanent storage devices use mass storage devices (e.g., magnetic or optical disks, flash memory) as permanent storage devices. In other embodiments, permanent storage devices may be removable storage devices (e.g., floppy disks, optical drives). System memory may be a read-write storage device or a volatile read-write storage device, such as dynamic random access memory. System memory may store some or all of the instructions and data required by the processor during operation. Furthermore, memory 410 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, memory 410 may include a removable storage device that is readable and / or writable, such as a laser disc (CD), a read-only digital multifunction optical disc (e.g., DVD-ROM, dual-layer DVD-ROM), a read-only Blu-ray disc, an ultra-high density optical disc, a flash memory card (e.g., SD card, mini SD card, Micro-SD card, etc.), a magnetic floppy disk, etc. Computer-readable storage media do not contain carrier waves or transient electronic signals transmitted wirelessly or via wired connections.

[0066] The memory 410 stores executable code, which, when processed by the processor 420, can cause the processor 420 to execute part or all of the methods described above.

[0067] Furthermore, the method according to the present invention can also be implemented as a computer program or computer program product, which includes computer program code instructions for performing some or all of the steps in the above-described method of the present invention.

[0068] Alternatively, the present invention can also be implemented as a non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) storing executable code (or computer program, or computer instruction code) thereon, which, when executed by a processor of an electronic device (or computing device, server, etc.), causes the processor to perform part or all of the steps of the method described above according to the present invention.

[0069] The present invention has been described in detail above with reference to the accompanying drawings. In the above embodiments, the descriptions of each embodiment have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. Those skilled in the art should also understand that the actions and modules involved in the specification are not necessarily essential to the present invention. Furthermore, it is understood that the steps in the method of the embodiments of the present invention can be adjusted, combined, and deleted according to actual needs, and the modules in the device of the embodiments of the present invention can be combined, divided, and deleted according to actual needs.

[0070] Those skilled in the art will also understand that the various exemplary logic blocks, modules, circuits, and algorithm steps described in connection with the disclosure herein can be implemented as electronic hardware, computer software, or a combination of both.

[0071] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0072] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A coal mine underground sensor calibration state identification method, characterized in that, The method comprises the steps of: In response to a sensor alarm event, extracting data in a time window before and after the event, and generating a data curve image according to the type of the sensor; Calling the model parameters corresponding to the type of the sensor, using a pre-trained image recognition model to recognize the generated image, and marking it as a calibration alarm if the calibration feature is recognized, otherwise marking it as a normal alarm.

2. The method of claim 1, wherein, The training process of the image recognition model specifically comprises: Based on the alarm record, sensor state or work log, selecting a data sequence in a specific time window before and after the sensor alarm event; Filtering the abnormal invalid values in the data sequence to obtain valid data; According to the type of the sensor, the valid data is grouped to form a plurality of independent data sets; For each type of sensor data set, set a uniform vertical coordinate display range, convert each data sequence into a two-dimensional curve graph, and generate a standardized image; Labeling the generated curve image and identifying the calibration feature band, and constructing a labeled image data set according to the type of the sensor; Divide the data set into training set, validation set and test set, and use the image recognition model to train and optimize the parameters until the model performance reaches the preset index; Save the optimized model and parameters, and establish a mapping relationship between the sensor type and the model parameters.

3. The method of claim 2, wherein, The abnormal invalid values include specific numerical codes for representing signal interruption.

4. The method according to claim 2 or 3, characterized in that, The setting of the uniform vertical coordinate display range specifically comprises: increasing the maximum value by a certain percentage according to the upper limit of the calibration range of the sensor.

5. A coal mine underground sensor calibration status recognition device, characterized in that, The method comprises the steps of: A real-time recognition processing module is configured to, in response to a sensor alarm event, extract data in a time window before and after the event, and generate a data curve image according to the type of the sensor; A calibration state discrimination module is configured to call the model parameters corresponding to the type of the sensor, use a pre-trained image recognition model to recognize the generated image, and mark it as a calibration alarm if the calibration feature is recognized, otherwise mark it as a normal alarm.

6. The apparatus of claim 5, wherein, Further comprising: A model training module specifically comprises: A sample selection unit is configured to select a data sequence in a specific time window before and after a sensor alarm event based on an alarm record, a sensor state or a work log; A data preprocessing unit is configured to filter abnormal invalid values in the data sequence to obtain valid data; A sensor type grouping unit is configured to group the valid data according to the type of the sensor to form a plurality of independent data sets; A curve image generation unit is configured to, for each type of sensor data set, set a uniform vertical coordinate display range, convert each data sequence into a two-dimensional curve graph, and generate a standardized image; A labeling and construction unit is configured to label the generated curve image and identify the calibration feature band, and construct a labeled image data set according to the type of the sensor; A training and optimization unit is configured to divide the data set into a training set, a validation set and a test set, and use the image recognition model to train and optimize the parameters until the model performance reaches the preset index; A model deployment unit is configured to save the optimized model and parameters, and establish a mapping relationship between the sensor type and the model parameters.

7. The apparatus of claim 6, wherein, The abnormal invalid values include specific numerical codes for representing signal interruption.

8. The apparatus of claim 6 or 7, wherein, The curve image generating unit sets a uniform vertical coordinate display range, specifically including: setting a proportion as a maximum value according to the upper limit of the calibration range of the sensor.

9. A terminal device, comprising: Comprise: a processor; and a memory having stored thereon executable code that, when executed by the processor, causes the processor to perform the method of any of claims 1-4. 10.A non-transitory machine-readable storage medium having stored thereon executable code to: When the executable code is executed by the processor of the electronic device, the processor performs the method of any of claims 1-4. When the executable code is executed by the processor of the electronic device, the processor performs the method of any of claims 1-4.

Images