Method, device, terminal and medium for managing function debugging procedure of probe based on internet of things

Abstract

The application discloses a kind of based on internet of things probe function debugging process management method, device, terminal and medium, the method includes: obtaining preprocessed probe function static characteristic data and preprocessed probe function dynamic characteristic data;The preprocessed probe function static characteristic data, the preprocessed probe function dynamic characteristic data and probe function historical debugging process data are extracted features and are handled, and high-dimensional coding characteristic vector is obtained;According to the high-dimensional coding characteristic vector, by preset sequence to sequence architecture model, generate target debugging process management scheme.The application aims at improving the detection accuracy of the probe after debugging and reducing the failure rate of probe operation.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, terminal and medium for managing the debugging process of detector functions based on the Internet of Things. Background Technology

[0002] Today, various detectors (gas, temperature, pressure, infrared, etc.) have become core sensing nodes for industrial safety monitoring, environmental monitoring, and intelligent manufacturing process control. The detection accuracy and operational reliability of the detectors directly determine the stability of the entire monitoring and control system. Functional debugging, as a core process before the detectors leave the factory and during regular maintenance, directly affects the final performance and service life of the detectors.

[0003] However, the shortcomings of existing detector function debugging process management techniques are mainly reflected in the following aspects: the debugging process, parameter adjustment range, and quality judgment standards are highly dependent on the individual skills of operators, the debugging quality of detectors of the same model varies significantly, it is impossible to form a replicable unified standard, and the labor cost is high and the efficiency is low; existing technologies only detect static parameters such as rated voltage and factory threshold, completely ignoring the dynamic response characteristics (response time, recovery time, power consumption fluctuation) of detectors under no-load, standard signal, and extreme signal conditions, and cannot comprehensively assess the actual operating status; the existing massive historical debugging records and fault cases are only stored as archives and have not been systematically analyzed and mined, and excellent debugging experience cannot be transformed into standardized decision-making logic; at the same time, existing technologies cannot dynamically adjust the process sequence and parameters according to the real-time collected detector data, the handling of abnormal situations is lagging behind, and it is difficult to achieve batch automated debugging of large-scale detectors. Summary of the Invention

[0004] The main purpose of this application is to provide a method, device, terminal and medium for managing the function debugging process of detectors based on the Internet of Things, which aims to improve the detection accuracy of detectors after debugging and reduce the failure rate of detector operation.

[0005] To achieve the above objectives, this application provides a method for managing the debugging process of a detector based on the Internet of Things, the method comprising: Acquire preprocessed static feature data and preprocessed dynamic feature data of the detector function; Feature extraction processing is performed on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector. The historical debugging process data of the detector function is used to characterize the status of the debugging process of the detector function in the past period. Based on the high-dimensional encoded feature vector, a target debugging process management scheme is generated through a preset sequence-to-sequence architecture model. The target debugging process management scheme includes target debugging step information and parameter adjustment data.

[0006] Specifically, the preprocessed detector function static feature data is a detector function static feature data vector, and the preprocessed detector function dynamic feature data is a detector function dynamic feature data vector. The acquisition of preprocessed static feature data and preprocessed dynamic feature data of the detector function includes: Collect basic static feature data of detector functions, wherein the basic static feature data of detector functions includes basic identification classification data and basic numerical data; The basic identifier classification data is converted into a one-hot encoded vector, and the basic numerical data is normalized to obtain a basic numerical data vector. The one-hot encoded vector and the basic numerical data vector are concatenated to obtain the detector functional static feature data vector. Synchronously acquire the time-series dynamic data of the detector under preset test commands, wherein the preset test commands include no-load test commands, standard signal input test commands, extreme signal input test commands, and continuous operation stability test commands; By using a preset sliding window, the time-series dynamic data is subjected to sliding window filtering to obtain denoised time-series dynamic data; A preset dynamic feature calculation logic is executed on the denoised time-series dynamic data to obtain dynamic feature values; The dynamic feature value is compared with the standard dynamic feature interval corresponding to the detector to obtain the dynamic feature comparison result; The dynamic feature value and the comparison result of the dynamic feature are concatenated to obtain the dynamic feature data vector of the detector function.

[0007] Specifically, the denoising timing dynamic data includes output signal amplitude, response time, recovery time, power consumption fluctuation, and communication delay; The step of performing a preset dynamic feature calculation logic on the denoised time-series dynamic data to obtain dynamic feature values ​​includes: The average value, maximum value, minimum value, and standard deviation of the output signal amplitude are calculated. The average response time and the 95th percentile of the response time were calculated. The average recovery time and the 95th percentile of the recovery time were calculated. The variance and range of power consumption fluctuations were calculated. The average communication delay and the maximum deviation of the communication delay are calculated.

[0008] Specifically, the step of performing feature extraction processing on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector includes: The static feature data vector of the detector function is first encoded by a pre-defined convolutional layer network to obtain a static feature encoding vector. By using a pre-set bidirectional long short-term memory network, the dynamic feature data vector of the detector function is subjected to a second encoding process to obtain a dynamic feature encoding vector. By using a preset attention mechanism network, the historical debugging process data of the detector function, the static feature encoding vector, and the dynamic feature encoding vector are processed to obtain the historical data encoding vector; The high-dimensional encoded feature vector is obtained by using a preset feature fusion network based on the static feature encoding vector, the dynamic feature encoding vector, and the historical data encoding vector.

[0009] Specifically, the preset sequence-to-sequence architecture model includes an encoder layer, a decoder layer, and a scheme optimization layer; The step of generating a target debugging process management scheme based on the high-dimensional encoded feature vector and through a preset sequence-to-sequence architecture model includes: The encoder layer performs a third encoding process on the high-dimensional encoded feature vector to obtain the encoder context vector. The decoder layer obtains the debugging procedure scheme text sequence based on the start symbol vector in the preset sequence-to-sequence architecture model and the encoder context vector. The target debugging process management scheme is obtained through the scheme optimization layer based on the debugging process scheme text sequence, the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data.

[0010] Specifically, the decoder layer includes an embedding layer, a self-attention layer, an encoder-decoder attention mechanism layer, a feedforward neural network layer, and a fully connected layer; The step of obtaining the debugging procedure text sequence through the decoder layer, based on the start symbol vector in the preset sequence-to-sequence architecture model and the encoder context vector, includes: The start symbol vector is input into the embedding layer to obtain the embedding vector; The embedding vector is input into the self-attention layer to obtain the output result of the self-attention layer; The output of the self-attention layer and the encoder context vector are input to the encoder-decoder attention mechanism layer to obtain the output of the encoder-decoder attention mechanism layer. The output of the encoder-decoder attention mechanism layer is input into the feedforward neural network layer to obtain the output of the feedforward neural network layer. The output of the feedforward neural network layer is input into the fully connected layer to obtain the probability of each word in the current position's output vocabulary. The word with the highest probability in the output vocabulary at the current position is determined as the output result corresponding to the current position, and the output result corresponding to the current position is transformed into the start symbol vector corresponding to the next time step; Repeat the above steps until the start symbol vector corresponding to the next time step is the end symbol vector, and generate the debugging procedure scheme text sequence based on the output results corresponding to all current positions in the repetition process and the end symbol vector.

[0011] Specifically, the step of obtaining the target debugging process management scheme through the scheme optimization layer, based on the debugging process scheme text sequence, the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data, includes: Through the scheme optimization layer, the text sequence of the debugging process scheme is transformed into text structured data, wherein the text structured data includes step sequence number, debugging content, parameter adjustment value, and quality inspection standard; Through the scheme optimization layer, the parameter adjustment value is adjusted according to the preprocessed detector function static feature data and the preprocessed detector function dynamic feature data to obtain the parameter adjustment data; Through the scheme optimization layer, the step ordering number is adjusted according to the historical debugging process data of the detector function to obtain the primary debugging step information; Through the scheme optimization layer, the debugging content and the primary debugging step information are adjusted according to the quality inspection standards to obtain the target debugging step information.

[0012] To achieve the above objectives, this application also provides an Internet of Things-based detector function debugging process management device, the device comprising: The first unit is used to acquire preprocessed static feature data and preprocessed dynamic feature data of the detector function. The second unit is used to perform feature extraction processing on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector. The historical debugging process data of the detector function is used to characterize the status of the debugging process of the detector function in the past period. The third unit is used to generate a target debugging process management scheme based on the high-dimensional encoded feature vector and through a preset sequence-to-sequence architecture model. The target debugging process management scheme includes target debugging step information and parameter adjustment data.

[0013] To achieve the above objectives, this application also provides a terminal, including a memory storing multiple instructions; the processor loads instructions from the memory to execute the steps in any of the methods provided in this application.

[0014] To achieve the above objectives, this application also provides a medium storing a plurality of instructions adapted for loading by a processor to execute the steps in any of the methods provided in this application.

[0015] This application provides a method, device, terminal, and medium for managing the debugging process of detector functions based on the Internet of Things. The method first acquires pre-processed static and dynamic feature data of the detector functions; then, it performs feature extraction processing on the pre-processed static and dynamic feature data, as well as historical debugging process data, to obtain a high-dimensional encoded feature vector; based on the high-dimensional encoded feature vector, a target debugging process management scheme is generated through a preset sequence-to-sequence architecture model to improve the detection accuracy of the debugged detector and reduce the detector's operational failure rate. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the method provided in the embodiments of this application; Figure 2 This is a schematic diagram illustrating the application implementation of the preset sequence-to-sequence architecture model provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the device provided in the embodiments of this application; Figure 4 This is a schematic diagram of the terminal structure provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] The shortcomings of existing detector function debugging process management techniques are mainly reflected in the following aspects: the debugging process, parameter adjustment range, and quality judgment standards are highly dependent on the individual skills of operators, resulting in significant differences in debugging quality for detectors of the same model, making it impossible to form a replicable unified standard, and also resulting in high labor costs and low efficiency; existing technologies only test static parameters such as rated voltage and factory threshold, completely ignoring the dynamic response characteristics (response time, recovery time, power consumption fluctuations) of detectors under no-load, standard signal, and extreme signal conditions, and cannot comprehensively assess the actual operating status; the existing massive historical debugging records and fault cases are only stored as archives and have not been systematically analyzed and mined, and excellent debugging experience cannot be transformed into standardized decision-making logic; at the same time, existing technologies cannot dynamically adjust the process sequence and parameters based on real-time acquired detector data, resulting in delayed handling of abnormal situations and making it difficult to achieve batch automated debugging of large-scale detectors.

[0019] Therefore, this application provides a method, device, terminal, and medium for managing the debugging process of detector functions based on the Internet of Things to solve practical technical problems.

[0020] In some embodiments, the device may be integrated into an electronic device, such as a terminal or server.

[0021] In some embodiments, the server may also be implemented as a terminal.

[0022] The server can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0023] The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and the server can be connected directly or indirectly through wired or wireless communication, which is not limited herein.

[0024] The following sections provide detailed descriptions of each example. It should be noted that the sequence numbers of the following embodiments are not intended to limit the preferred order of the embodiments.

[0025] This application provides a method for managing the debugging process of a detector based on the Internet of Things. The method can improve the detection accuracy of the detector after debugging and reduce the failure rate of the detector operation.

[0026] In some embodiments, the Internet of Things (IoT) upon which the method relies is a dedicated Industrial Internet of Things (IIoT) system for the entire lifecycle debugging of detectors. This system serves as the physical foundation and data carrier for the implementation of logical algorithms and neural network models, enabling batch access of detectors to be debugged, real-time data acquisition, remote command control, and end-to-end digitalization of the debugging process. This architecture adopts an industrial-grade four-layer standard design, including: Perception layer: The perception layer serves as the interface between the Internet of Things (IoT) and the physical world. It is responsible for collecting all the raw data required for detector debugging and executing debugging commands. Its core components include: The detector cluster to be debugged: It connects to the Internet of Things through its own communication module, uploads static configuration data and dynamic response data, and receives and executes test instructions and parameter adjustment instructions issued by the platform.

[0027] Standardized debugging execution unit: including high-precision standard signal generator, programmable power supply, environmental simulation box, communication protocol converter, etc., is uniformly scheduled by the Internet of Things platform to provide the detector with a standardized test environment and input signals.

[0028] Auxiliary monitoring equipment includes temperature and humidity sensors, voltage and current monitors, network quality analyzers, etc., to monitor the debugging environment and equipment status in real time and eliminate the impact of external interference on the debugging results.

[0029] Network layer: The network layer adopts a heterogeneous communication architecture that integrates wired and wireless communication to ensure low-latency and highly reliable transmission of debugging data and control commands. Short-range communication: The equipment in the commissioning station uses industrial Ethernet, RS485 or Wi-Fi6 communication to meet the requirements of millisecond-level command issuance and high-frequency data acquisition.

[0030] Wide-area communication: For distributed detector maintenance sites, NB-IoT or LoRaWAN communication is used to achieve low-power, wide-coverage remote debugging.

[0031] Edge gateway: Deployed locally in the debugging workshop or site, it is responsible for protocol conversion, data preprocessing and local caching. It can still maintain basic debugging functions when the network is interrupted, and at the same time, it encrypts the uploaded data.

[0032] Platform layer: The platform layer is the brain of the entire IoT system, providing unified data management and basic computing capabilities for the methods described above. Equipment management platform: Enables unified registration, identity authentication, status monitoring, and lifecycle management of all detectors and equipment to be debugged.

[0033] Time Series Database: Specifically stores high-frequency time series data generated by dynamic testing of detectors, supporting millisecond-level data query and batch data export.

[0034] Data processing engine: Provides basic data cleaning, filtering, and standardization capabilities, providing pre-processed data for logical algorithms.

[0035] Edge computing nodes: Deployed on local gateways or workshop servers, they run lightweight feature extraction logic and neural network inference models to enable rapid local generation of debugging solutions.

[0036] Application layer: The application layer directly targets debugging operators and administrators, integrating all the core functional modules of the method: Debugging Task Management System: Responsible for creating, assigning, scheduling, and tracking the progress of debugging tasks.

[0037] Feature extraction and analysis module: Runs logical algorithms to automatically extract the static and dynamic features of the detector.

[0038] Intelligent debugging decision engine: Runs a neural network model to generate personalized debugging process management solutions.

[0039] Visualized monitoring and traceability system: Real-time display of debugging process data and solution execution status, supporting full-process traceability and quality analysis of debugging records.

[0040] In some embodiments, the pre-shipment functional debugging of a hydrogen sulfide (H2S) toxic gas detector in an electronic component manufacturing workshop is the specific scenario corresponding to this embodiment. The detector to be debugged is a TD-400 electrochemical hydrogen sulfide detector with a rated detection range of 0-100ppm. The debugging task corresponding to this embodiment is the automated functional debugging of the 127th detector on the batch production line.

[0041] like Figure 1 The specific process of the method can be as follows: S110. Obtain preprocessed static feature data and preprocessed dynamic feature data of the detector function.

[0042] In some embodiments, the preprocessed detector function static feature data is a detector function static feature data vector, and the preprocessed detector function dynamic feature data is a detector function dynamic feature data vector.

[0043] Specifically, the acquisition of preprocessed detector functional static feature data and preprocessed detector functional dynamic feature data includes the steps A1 to A8 as shown below: A1. Collect basic static feature data of detector functions, wherein the basic static feature data of detector functions includes basic identification classification data and basic numerical data.

[0044] In some embodiments, the basic static characteristic data of the detector can be collected via an IoT debugging platform: Basic identification classification data: unique identifier (TD202604290127), hardware model (TD-400), sensor type (electrochemical hydrogen sulfide sensor), communication protocol version (Modbus-RTUV2.1), firmware version (V1.3.2).

[0045] Basic numerical data: Rated operating voltage (24VDC), rated operating current (120mA), factory calibration zero point (0.1mV), preset detection threshold (20ppm).

[0046] A2. Convert the basic identifier classification data into a one-hot encoded vector, and normalize the basic numerical data to obtain a basic numerical data vector.

[0047] In some embodiments, the above 5 basic identifier classification data are converted into one-hot encoded vectors with a total dimension of 12 (e.g., the hardware model TD-400 corresponds to one-hot encoded [1,0,0]); the above 4 basic numerical data are subjected to min-max normalization and mapped to the [0,1] interval to obtain a 4-dimensional basic numerical data vector (e.g., the rated operating voltage of 24V is normalized to 0.96).

[0048] A3. The one-hot encoded vector and the basic numerical data vector are concatenated to obtain the detector functional static feature data vector.

[0049] In some embodiments, a 12-dimensional one-hot encoded vector is concatenated with a 4-dimensional basic numerical data vector in terms of dimensions to obtain a 16-dimensional detector functional static feature data vector S.

[0050] A4. Synchronously acquire the time-series dynamic data of the detector under preset test commands, wherein the preset test commands include no-load test commands, standard signal input test commands, extreme signal input test commands, and continuous operation stability test commands.

[0051] In some embodiments, four types of preset test commands are sequentially sent to the detector via an IoT platform, and five types of time-series dynamic data are simultaneously acquired at a sampling frequency of 100Hz: No-load test command: Continue for 30 seconds without introducing any gas; Standard signal input test command: Introduce H2S standard gas with a concentration of 20ppm for 30 seconds; Limit signal input test command: Introduce H2S standard gas with a concentration of 100ppm for 30 seconds; Continuous operation stability test instruction: The detector is continuously powered on for 24 hours, and data is collected every 10 minutes.

[0052] A5. The time-series dynamic data is subjected to sliding window filtering through a preset sliding window to obtain denoised time-series dynamic data.

[0053] In some embodiments, a preset sliding window with a window size of 5 and a step size of 1 can be used to filter all the collected time-series dynamic data, remove high-frequency noise, and obtain denoised time-series dynamic data.

[0054] A6. Perform a preset dynamic feature calculation logic on the denoised time-series dynamic data to obtain dynamic feature values.

[0055] In some embodiments, the denoising timing dynamic data includes output signal amplitude, response time, recovery time, power consumption fluctuation, and communication delay.

[0056] Specifically, the step of performing a preset dynamic feature calculation logic on the denoised time-series dynamic data to obtain dynamic feature values ​​includes the steps A61 to A65 shown below: A61. Calculate the average value, maximum value, minimum value, and standard deviation of the output signal amplitude.

[0057] In some embodiments, the average value (19.8mV), maximum value (20.3mV), minimum value (19.2mV), and standard deviation (0.21mV) of the output signal amplitude are calculated.

[0058] A62. Calculate the average response time and the 95th percentile of the response time.

[0059] In some embodiments, the average response time (1.2s) and the 95th percentile (1.4s) are calculated.

[0060] A63. Calculate the average recovery time and the 95th percentile of the recovery time.

[0061] In some embodiments, the average recovery time (2.1s) and the 95th percentile (2.5s) are calculated.

[0062] A64. Calculate the variance and range of power consumption fluctuations.

[0063] In some embodiments, the variance (0.8mW²) and range (3.2mW) of power consumption fluctuations are calculated.

[0064] A65. Calculate the average communication delay and the maximum deviation of the communication delay.

[0065] In some embodiments, the average communication delay (15ms) and the maximum deviation (8ms) are calculated.

[0066] A7. Compare the dynamic feature value with the standard dynamic feature interval corresponding to the detector to obtain the dynamic feature comparison result.

[0067] In some embodiments, the above 12 dynamic feature values ​​are compared with the standard dynamic feature range of the TD-400 model detector to obtain the dynamic feature comparison results: for example, all feature values ​​are within the standard range, there are no abnormal features, and the abnormality markers are all 0.

[0068] A8. The dynamic feature value and the comparison result of the dynamic feature are spliced ​​together to obtain the dynamic feature data vector of the detector function.

[0069] In some embodiments, 12 dynamic feature values ​​and 12 anomaly markers are concatenated in a dimensional manner to obtain a 24-dimensional detector functional dynamic feature data vector D.

[0070] S120. Perform feature extraction processing on the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data to obtain a high-dimensional encoded feature vector, wherein the detector function historical debugging process data is used to characterize the status of the detector function debugging process in the past time.

[0071] In some embodiments, the feature extraction processing of the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data to obtain a high-dimensional encoded feature vector includes the steps B1 to B4 shown below: B1. The static feature data vector of the detector function is first encoded by a preset convolutional layer network to obtain a static feature encoding vector.

[0072] In some embodiments, a 16-dimensional static feature data vector S is input into a two-layer one-dimensional convolutional network of a preset convolutional network. The first layer uses 16 one-dimensional convolutional kernels of size 3 for convolution operation, and after ReLU activation, a 14×16-dimensional primary feature map is obtained. The second layer uses 32 one-dimensional convolutional kernels of size 2 for convolution operation, and after ReLU activation, a 13×32-dimensional advanced feature map is obtained. Then, global average pooling is performed on the advanced feature map to obtain a 32-dimensional static feature encoding vector S_enc.

[0073] B2. By using a preset bidirectional long short-term memory network, the dynamic feature data vector of the detector function is subjected to a second encoding process to obtain a dynamic feature encoding vector.

[0074] In some embodiments, the 24-dimensional dynamic feature data vector D is input into a pre-defined bidirectional long short-term memory network (with 64 hidden layer units), including: a forward LSTM network processes the forward-order dynamic feature vector and outputs a 64-dimensional forward hidden state vector; a backward LSTM network processes the reverse-order dynamic feature vector and outputs a 64-dimensional backward hidden state vector; then, the two hidden state vectors are concatenated to obtain a 128-dimensional dynamic feature encoding vector D_enc.

[0075] B3. The historical data encoding vector is obtained by processing the historical debugging process data of the detector function, the static feature encoding vector and the dynamic feature encoding vector through a preset attention mechanism network.

[0076] In some embodiments, the static feature encoding vector S_enc and the dynamic feature encoding vector D_enc are concatenated and mapped to a 64-dimensional query vector Q through a fully connected layer in a preset attention mechanism network. 1000 historical debugging process data entries (1000×64 dimensions) of the TD-400 detector stored in the database are mapped to 1000×64-dimensional key vector K and value vector V through two fully connected layers in the preset attention mechanism network. The dot product of Q and the transpose of K is calculated and divided by a preset parameter for scaling. Then, the attention weights are obtained through the softmax function. Finally, the weights are multiplied by V and summed to obtain the 64-dimensional historical data encoding vector H_enc.

[0077] B4. The high-dimensional encoded feature vector is obtained by using a preset feature fusion network based on the static feature encoding vector, the dynamic feature encoding vector, and the historical data encoding vector.

[0078] In some embodiments, the 32-dimensional S_enc, 128-dimensional D_enc, and 64-dimensional H_enc are concatenated in terms of dimensions to obtain a 224-dimensional fused feature vector; a nonlinear transformation is performed through a fully connected layer containing 256 neurons in a preset feature fusion network to obtain a 256-dimensional high-dimensional encoded feature vector E.

[0079] S130. Based on the high-dimensional encoded feature vector, a target debugging process management scheme is generated through a preset sequence-to-sequence architecture model, wherein the target debugging process management scheme includes target debugging step information and parameter adjustment data.

[0080] In some embodiments, the preset sequence-to-sequence architecture model includes an encoder layer, a decoder layer, and a scheme optimization layer.

[0081] Specifically, the step of generating a target debugging process management scheme based on the high-dimensional encoded feature vector and through a preset sequence-to-sequence architecture model includes the following steps C1 to C3: C1. The high-dimensional encoded feature vector is subjected to a third encoding process through the encoder layer to obtain the encoder context vector.

[0082] In some embodiments, a 256-dimensional high-dimensional encoded feature vector E is input into two Transformer encoder layers of the encoder layer, and after processing by a multi-head self-attention mechanism and a feedforward neural network, a 256-dimensional encoder context vector C is obtained.

[0083] C2. Through the decoder layer, the text sequence of the debugging procedure scheme is obtained based on the start symbol vector in the preset sequence to sequence architecture model and the encoder context vector.

[0084] In some embodiments, such as Figure 2 The decoder layer includes an embedding layer, a self-attention layer, an encoder-decoder attention mechanism layer, a feedforward neural network layer, and a fully connected layer.

[0085] Specifically, the vocabulary of the preset sequence-to-sequence architecture model includes 200 debugging-specific terms and 3 special control characters ( <start>Corresponding index 0, <end>Corresponding index 1, <pad>(Corresponding to index 2), the embedding vector dimension is 256.

[0086] Specifically, the step of obtaining the debugging procedure scheme text sequence through the decoder layer based on the start symbol vector in the preset sequence-to-sequence architecture model and the encoder context vector includes the following steps C21 to C27: C21. Input the start symbol vector into the embedding layer to obtain the embedding vector.

[0087] In some embodiments, the pre-generated start symbol vector <start>(256-dimensional) Input embedding layer to obtain 256-dimensional embedding vector.

[0088] C22. Input the embedding vector into the self-attention layer to obtain the output result of the self-attention layer.

[0089] In some embodiments, the embedded vector is input into the self-attention layer, future position information is masked, and the output result of the self-attention layer is obtained.

[0090] C23. Input the output of the self-attention layer and the encoder context vector into the encoder-decoder attention mechanism layer to obtain the output of the encoder-decoder attention mechanism layer.

[0091] In some embodiments, the output of the self-attention layer and the encoder context vector C are input into the encoder-decoder attention mechanism layer to obtain the attention output.

[0092] C24. Input the output of the encoder-decoder attention mechanism layer into the feedforward neural network layer to obtain the output of the feedforward neural network layer.

[0093] In some embodiments, the attention output is input into a feedforward neural network layer (containing 512 neurons, ReLU activated) to obtain the feedforward network output.

[0094] C25. Input the output of the feedforward neural network layer into the fully connected layer to obtain the probability of each word in the current position output vocabulary.

[0095] In some embodiments, the output of the feedforward network is input into the fully connected layer, and the probability of each word in the vocabulary at the current position is calculated using the softmax function.

[0096] C26. Determine the word with the highest probability in the output vocabulary at the current position as the output result corresponding to the current position, and convert the output result corresponding to the current position into the start symbol vector corresponding to the next time step.

[0097] In some embodiments, the word with the highest probability, "Step 1", is selected as the output result of the current position and converted into a corresponding 256-dimensional vector as the input of the next time step.

[0098] C27. Repeat the above steps until the start symbol vector corresponding to the next time step is the end symbol vector, and generate the debugging process scheme text sequence based on the output results corresponding to all current positions in the repetition process and the end symbol vector.

[0099] In some embodiments, the above steps are repeated, generating the following in sequence: Step 1: Check detector power connection; Step 2: Introduce 20ppm H2S standard gas; Step 3: Zero point calibration; Step 4: Range calibration; Step 5: Introduce 100ppm H2S standard gas for limit test; Step 6: Continuous operation stability test; Step 7: Data recording and pass / fail determination. <end>Finally, a complete text sequence of debugging procedures is obtained.

[0100] C3. Through the scheme optimization layer, the target debugging process management scheme is obtained based on the debugging process scheme text sequence, the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data.

[0101] In some embodiments, the step of obtaining the target debugging process management scheme through the scheme optimization layer based on the debugging process scheme text sequence, the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data includes the following steps C31 to C34: C31. Through the scheme optimization layer, the text sequence of the debugging process scheme is converted into text structured data, wherein the text structured data includes step sequence number, debugging content, parameter adjustment value, and quality inspection standard.

[0102] In some embodiments, the text sequence of debugging procedure schemes is converted into structured data, and the step sequence number, debugging content, default parameter adjustment value, and basic quality inspection standards are extracted.

[0103] C32. Through the scheme optimization layer, the parameter adjustment value is adjusted according to the preprocessed detector function static feature data and the preprocessed detector function dynamic feature data to obtain the parameter adjustment data.

[0104] In some embodiments, based on the factory calibration zero point (0.1mV) in the static characteristics and the average amplitude of the output signal (19.8mV) in the dynamic characteristics, the zero point calibration parameter adjustment value is optimized from the default 0.05mV to 0.1mV, and the range calibration parameter adjustment value is optimized from the default 0.2mV to 0.1mV to obtain parameter adjustment data.

[0105] C33. Through the scheme optimization layer, the step ordering number is adjusted according to the historical debugging process data of the detector function to obtain the primary debugging step information.

[0106] In some embodiments, based on the fact that the success rate of "calibrating the zero point first and then calibrating the range" in historical debugging process data is higher than that of the reverse order (92.3%), the order of steps 3 and 4 is kept unchanged to obtain the primary debugging process information.

[0107] C34. Through the scheme optimization layer, the debugging content and the primary debugging step information are adjusted according to the quality inspection standard to obtain the target debugging step information.

[0108] In some embodiments, targeted quality inspection standards are added after each key debugging step. For example, after step 3, "the output signal should be between -0.1mV and 0.1mV" is added, and after step 4, "the output signal should be between 19.5mV and 20.5mV" is added, thus obtaining the target debugging step information.

[0109] Ultimately, the generated target debugging process management plan includes: Step 1: Check the detector power connection and confirm that the voltage is between 22.8V and 25.2V; Step 2: Introduce 20ppm H2S standard gas and stabilize for 30 seconds; Step 3: Zero point calibration, parameter adjustment value is 0.1mV, quality standard: output signal between -0.1mV and 0.1mV; Step 4: Calibrate the range, adjust the parameter value to 0.1mV, quality standard: output signal between 19.5mV and 20.5mV; Step 5: Introduce 100ppm H2S standard gas for extreme testing. Quality standards: response time ≤1.5s, recovery time ≤3s; Step 6: Continuous working stability test for 2 hours, quality standard: output signal fluctuation ≤ ±2%FS; Step 7: Data recording and qualification assessment, upload debugging data to the IoT platform.

[0110] In summary, this application provides a method for managing the debugging process of a detector based on the Internet of Things. The personalized debugging process management scheme generated by this method reduces the debugging time of the detector from the traditional 45 minutes to 22 minutes, achieves a post-debugging detection accuracy of 99.2%, and reduces the failure rate to 0.7% over the following six months. This application can improve the detection accuracy of the detector after debugging and reduce the detector's operational failure rate.

[0111] To better implement the above methods, this application also provides an IoT-based detector function debugging process management device. This device can be integrated into an electronic device, such as a terminal or server. The terminal can be a mobile phone, tablet, smart Bluetooth device, laptop, or personal computer; the server can be a single server or a server cluster consisting of multiple servers.

[0112] For example, in this embodiment, the method of this application embodiment will be described in detail by taking the Internet of Things-based detector function debugging process management device specifically integrated into the terminal as an example.

[0113] For example, such as Figure 3 As shown, the IoT-based detector function debugging process management device 300 may include a first unit 301, a second unit 302 and a third unit 303. The device includes: The first unit is used to acquire preprocessed static feature data and preprocessed dynamic feature data of the detector function. The second unit is used to perform feature extraction processing on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector. The historical debugging process data of the detector function is used to characterize the status of the debugging process of the detector function in the past period. The third unit is used to generate a target debugging process management scheme based on the high-dimensional encoded feature vector and through a preset sequence-to-sequence architecture model. The target debugging process management scheme includes target debugging step information and parameter adjustment data.

[0114] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0115] As can be seen from the above, the embodiments of this application can improve the detection accuracy of the detector after debugging and reduce the detector's failure rate.

[0116] This application also provides an electronic device, which can be a terminal, a server, or other similar device. The terminal can be a mobile phone, tablet computer, smart Bluetooth device, laptop computer, personal computer, etc.; the server can be a single server or a server cluster composed of multiple servers, etc.

[0117] In some embodiments, the product processing device may also be integrated into multiple electronic devices, such as multiple servers, with multiple servers implementing the IoT-based detector function debugging process management method of this application.

[0118] In this embodiment, the electronic device will be described in detail as a terminal, for example, such as... Figure 4 As shown, it illustrates a structural schematic diagram of the terminal 400 involved in an embodiment of this application. Specifically: The terminal 400 may include components such as a processor 401 with one or more processing cores, a memory 402 with one or more media, a power supply 403, an input module 404, and a communication module 405. Those skilled in the art will understand that... Figure 4 The terminal 400 structure shown does not constitute a limitation on the terminal 400, and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein: The processor 401 is the control center of the terminal 400. It connects various parts of the terminal 400 via various interfaces and lines, and performs various functions and processes data by running or executing software programs and / or modules stored in the memory 402, and by calling data stored in the memory 402, thereby providing overall monitoring of the terminal 400. In some embodiments, the processor 401 may include one or more processing cores; in some embodiments, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 401.

[0119] The memory 402 can be used to store software programs and modules. The processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the terminal 400, etc. In addition, the memory 402 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

[0120] The terminal 400 also includes a power supply 403 that supplies power to the various components. In some embodiments, the power supply 403 can be logically connected to the processor 401 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 403 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0121] The terminal 400 may also include an input module 404, which can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0122] The terminal 400 may also include a communication module 405. In some embodiments, the communication module 405 may include a wireless module. The terminal 400 can perform short-range wireless transmission through the wireless module of the communication module 405, thereby providing users with wireless broadband Internet access. For example, the communication module 405 can be used to help users send and receive emails, browse web pages, and access streaming media.

[0123] Although not shown, terminal 400 may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, processor 401 in terminal 400 loads the executable files corresponding to the processes of one or more applications into memory 402 according to the following instructions, and processor 401 runs the applications stored in memory 402 to realize various functions, as follows: Acquire preprocessed static feature data and preprocessed dynamic feature data of the detector function; Feature extraction processing is performed on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector. The historical debugging process data of the detector function is used to characterize the status of the debugging process of the detector function in the past period. Based on the high-dimensional encoded feature vector, a target debugging process management scheme is generated through a preset sequence-to-sequence architecture model. The target debugging process management scheme includes target debugging step information and parameter adjustment data.

[0124] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0125] As can be seen from the above, the embodiments of this application can improve the detection accuracy of the detector after debugging and reduce the detector's failure rate.

[0126] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be accomplished by instructions, or by instructions controlling related hardware. These instructions can be stored in a medium and loaded and executed by a processor.

[0127] To this end, embodiments of this application provide a medium storing multiple instructions that can be loaded by a processor to execute steps in any of the IoT-based detector function debugging process management methods provided in embodiments of this application. For example, the instructions can execute the following steps: Acquire preprocessed static feature data and preprocessed dynamic feature data of the detector function; Feature extraction processing is performed on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector. The historical debugging process data of the detector function is used to characterize the status of the debugging process of the detector function in the past period. Based on the high-dimensional encoded feature vector, a target debugging process management scheme is generated through a preset sequence-to-sequence architecture model. The target debugging process management scheme includes target debugging step information and parameter adjustment data.

[0128] The medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0129] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a medium. A processor of a computer device reads the computer instructions from the medium and executes the computer instructions, causing the computer device to perform the methods provided in the various optional implementations of the above embodiments.

[0130] Since the instructions stored in the medium can execute the steps in any of the IoT-based detector function debugging process management methods provided in the embodiments of this application, the beneficial effects that any of the IoT-based detector function debugging process management methods provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0131] The foregoing has provided a detailed description of a method, apparatus, terminal, and medium for managing the debugging process of a detector based on the Internet of Things, as provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.< / end> < / start> < / pad> < / end> < / start>

Claims

1. A method for managing the debugging process of a detector based on the Internet of Things, characterized in that, The method includes: Acquire preprocessed static feature data and preprocessed dynamic feature data of the detector function; Feature extraction processing is performed on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector. The historical debugging process data of the detector function is used to characterize the status of the debugging process of the detector function in the past period. Based on the high-dimensional encoded feature vector, a target debugging process management scheme is generated through a preset sequence-to-sequence architecture model. The target debugging process management scheme includes target debugging step information and parameter adjustment data.

2. The method as described in claim 1, characterized in that, The preprocessed detector function static feature data is a detector function static feature data vector, and the preprocessed detector function dynamic feature data is a detector function dynamic feature data vector. The acquisition of preprocessed static feature data and preprocessed dynamic feature data of the detector function includes: Collect basic static feature data of detector functions, wherein the basic static feature data of detector functions includes basic identification classification data and basic numerical data; The basic identifier classification data is converted into a one-hot encoded vector, and the basic numerical data is normalized to obtain a basic numerical data vector. The one-hot encoded vector and the basic numerical data vector are concatenated to obtain the detector functional static feature data vector. Synchronously acquire the time-series dynamic data of the detector under preset test commands, wherein the preset test commands include no-load test commands, standard signal input test commands, extreme signal input test commands, and continuous operation stability test commands; By using a preset sliding window, the time-series dynamic data is subjected to sliding window filtering to obtain denoised time-series dynamic data; A preset dynamic feature calculation logic is executed on the denoised time-series dynamic data to obtain dynamic feature values; The dynamic feature value is compared with the standard dynamic feature interval corresponding to the detector to obtain the dynamic feature comparison result; The dynamic feature values ​​and the comparison results of the dynamic features are concatenated to obtain the dynamic feature data vector of the detector function.

3. The method as described in claim 2, characterized in that, The denoising timing dynamic data includes output signal amplitude, response time, recovery time, power consumption fluctuation, and communication delay; The step of performing a preset dynamic feature calculation logic on the denoised time-series dynamic data to obtain dynamic feature values ​​includes: The average value, maximum value, minimum value, and standard deviation of the output signal amplitude are calculated. The average response time and the 95th percentile of the response time were calculated. The average recovery time and the 95th percentile of the recovery time were calculated. The variance and range of power consumption fluctuations were calculated. The average communication delay and the maximum deviation of the communication delay are calculated.

4. The method as described in claim 2, characterized in that, The process of extracting features from the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function yields a high-dimensional encoded feature vector, including: The static feature data vector of the detector function is first encoded by a pre-defined convolutional layer network to obtain a static feature encoding vector. By using a pre-set bidirectional long short-term memory network, the dynamic feature data vector of the detector function is subjected to a second encoding process to obtain a dynamic feature encoding vector. By using a preset attention mechanism network, the historical debugging process data of the detector function, the static feature encoding vector, and the dynamic feature encoding vector are processed to obtain the historical data encoding vector; The high-dimensional encoded feature vector is obtained by using a preset feature fusion network based on the static feature encoding vector, the dynamic feature encoding vector, and the historical data encoding vector.

5. The method as described in claim 1, characterized in that, The preset sequence-to-sequence architecture model includes an encoder layer, a decoder layer, and a scheme optimization layer; The step of generating a target debugging process management scheme based on the high-dimensional encoded feature vector and through a preset sequence-to-sequence architecture model includes: The encoder layer performs a third encoding process on the high-dimensional encoded feature vector to obtain the encoder context vector. The decoder layer obtains the debugging procedure scheme text sequence based on the start symbol vector in the preset sequence-to-sequence architecture model and the encoder context vector. The target debugging process management scheme is obtained through the scheme optimization layer based on the debugging process scheme text sequence, the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data.

6. The method as described in claim 5, characterized in that, The decoder layer includes an embedding layer, a self-attention layer, an encoder-decoder attention mechanism layer, a feedforward neural network layer, and a fully connected layer; The step of obtaining the debugging procedure text sequence through the decoder layer, based on the start symbol vector in the preset sequence-to-sequence architecture model and the encoder context vector, includes: The start symbol vector is input into the embedding layer to obtain the embedding vector; The embedding vector is input into the self-attention layer to obtain the output result of the self-attention layer; The output of the self-attention layer and the encoder context vector are input to the encoder-decoder attention mechanism layer to obtain the output of the encoder-decoder attention mechanism layer. The output of the encoder-decoder attention mechanism layer is input into the feedforward neural network layer to obtain the output of the feedforward neural network layer. The output of the feedforward neural network layer is input into the fully connected layer to obtain the probability of each word in the current position's output vocabulary. The word with the highest probability in the output vocabulary at the current position is determined as the output result corresponding to the current position, and the output result corresponding to the current position is transformed into the start symbol vector corresponding to the next time step; Repeat the above steps until the start symbol vector corresponding to the next time step is the end symbol vector, and generate the debugging procedure scheme text sequence based on the output results corresponding to all current positions in the repetition process and the end symbol vector.

7. The method as described in claim 5, characterized in that, The step of obtaining the target debugging process management scheme through the scheme optimization layer, based on the debugging process scheme text sequence, the preprocessed detector function static feature data, the preprocessed detector function dynamic feature data, and the detector function historical debugging process data, includes: Through the scheme optimization layer, the text sequence of the debugging process scheme is transformed into text structured data, wherein the text structured data includes step sequence number, debugging content, parameter adjustment value, and quality inspection standard; Through the scheme optimization layer, the parameter adjustment value is adjusted according to the preprocessed detector function static feature data and the preprocessed detector function dynamic feature data to obtain the parameter adjustment data; Through the scheme optimization layer, the step ordering number is adjusted according to the historical debugging process data of the detector function to obtain the primary debugging step information; Through the scheme optimization layer, the debugging content and the primary debugging step information are adjusted according to the quality inspection standards to obtain the target debugging step information.

8. A device for managing the debugging process of a detector based on the Internet of Things, characterized in that, The device includes: The first unit is used to acquire preprocessed static feature data and preprocessed dynamic feature data of the detector function. The second unit is used to perform feature extraction processing on the preprocessed static feature data of the detector function, the preprocessed dynamic feature data of the detector function, and the historical debugging process data of the detector function to obtain a high-dimensional encoded feature vector. The historical debugging process data of the detector function is used to characterize the status of the debugging process of the detector function in the past period. The third unit is used to generate a target debugging process management scheme based on the high-dimensional encoded feature vector and through a preset sequence-to-sequence architecture model. The target debugging process management scheme includes target debugging step information and parameter adjustment data.

9. A terminal, characterized in that, The method includes a processor and a memory, the memory storing multiple instructions; the processor loads instructions from the memory to perform the steps of the method as described in any one of claims 1 to 7.

10. A medium, characterized in that, The medium stores a plurality of instructions adapted for loading by a processor to execute the steps of the method according to any one of claims 1 to 7.

Images