Sensor detection method and system

Abstract

The invention relates to the technical field of sensor detection, in particular to a sensor detection method and system. The method comprises the following steps: collecting geometric, mechanical and electrical parameters of a sensor installation environment in real time, and constructing a sensor digital twinborn model; sensor parameters are automatically identified and initialized through a local area network according to the edge computing device, and sensor initial configuration data are obtained; generating sensor simulation data according to the sensor digital twin model and the sensor initial configuration data; and performing performance detection evaluation according to the sensor simulation data to obtain sensor detection data. According to the method, real-time detection and evaluation can be carried out in a dynamically changing environment, the performance of the sensor in various working states can be accurately predicted through combination of the virtual model and the actual performance of the sensor, and efficient and stable operation of the system is ensured.

Description

Technical Field

[0001] The present invention relates to the field of sensor detection technology, and in particular to a sensor detection method and system. Background Art

[0002] With the widespread application of Internet of Things (IoT) technology and smart devices, sensors, as key hardware components, are widely used in various industries, agriculture, medical care, transportation and other fields, and undertake important functions such as environmental monitoring, data collection, and state perception. The performance of sensors directly affects the reliability of equipment and the overall effectiveness of the system, so the detection and evaluation of sensors is particularly important. Traditional sensor detection methods mostly rely on manual operation or evaluation in a single environment based on fixed test models. This method not only has certain limitations, but also leads to measurement errors and cannot truly reflect the actual performance of sensors in complex working environments. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a sensor detection method and system to solve at least one of the above technical problems.

[0004] The present application provides a sensor detection method, comprising the following steps: Step S1: Collect the geometric, mechanical and electrical parameters of the sensor installation environment in real time and build a digital twin model of the sensor; Step S2: Automatically identify and initialize sensor parameters through the local area network according to the edge computing device to obtain initial configuration data of the sensor; Step S3: Generate sensor simulation data according to the sensor digital twin model and the sensor initial configuration data; Step S4: Perform performance testing and evaluation based on the sensor simulation data to obtain sensor testing data.

[0005] In the present invention, by real-time acquisition of the geometric, mechanical and electrical parameters of the sensor installation environment and construction of the digital twin model of the sensor, the performance of the sensor in the actual working environment can be accurately simulated. The digital twin model can provide accurate environmental interaction relationships, making the simulation data more reliable, thereby providing accurate basic data for performance testing. Automated configuration reduces errors in human intervention, improves the adaptability and flexibility of sensors during deployment, and helps to improve the overall deployment efficiency and accuracy of the system. The sensor simulation data generated based on the digital twin model can reflect the working status of the sensor in real time. With the help of the efficient data processing capabilities of edge computing and local area networks, the system can monitor the performance of the sensor in real time and respond quickly to potential problems, which helps to promptly discover performance problems or failures of the sensor, reduce the response time of equipment failures, and improve the reliability and safety of the system. Through the simulation data generated by the digital twin model, performance evaluation and testing can be performed without actual sensor hardware. It can effectively reduce the number and cost of physical experiments, especially in the large-scale deployment stage of sensors, and can significantly reduce resource consumption in the testing and verification process.

[0006] Preferably, step S1 specifically includes: Step S11: deploy an environmental sensor array to collect real-time environmental parameter data through edge devices; Step S12: Generate a virtual sensor environment according to the real-time environmental parameter data to obtain virtual sensor environment data; Step S13: using laser scanning to obtain the geometric shape of the installation position, obtaining sensor geometric shape data, and performing geometric reconstruction based on the sensor geometric shape data to obtain sensor installation geometric parameter data; Step S14: calibrating the mechanical parameters of the installation environment using the pressure sensor and the accelerometer to obtain the mechanical characteristic data of the sensor; Step S15: using an oscilloscope to measure the connector impedance, voltage waveform and grounding state to obtain electrical characteristic data; Step S16: Establish a digital twin model of the sensor installation geometric parameter data, sensor mechanical characteristic data, electrical characteristic data and virtual sensor environment data to obtain a sensor digital twin model.

[0007] The present invention provides accurate real-time environmental data, which helps to make instant adjustments during sensor configuration, debugging and operation to cope with environmental changes. Generating virtual sensor environmental data based on real-time environmental parameter data helps to simulate and predict the performance of sensors in specific environments, especially when physical testing is inconvenient or impossible. The acquisition and modeling of geometric data can help to better analyze the impact of the environment on sensor signals, thereby improving the reliability of the system. The present invention can provide accurate mechanical property data, so that the working performance of the sensor can be accurately evaluated under specific mechanical conditions. By accurately measuring the electrical characteristics of the sensor (such as impedance, voltage, and grounding status), the stability of signal transmission and the electrical reliability of the sensor can be guaranteed. The digital twin model can accurately simulate the working state of the sensor and its interaction with the environment in a virtual space, providing deep insights into the behavior of the sensor.

[0008] Preferably, step S12 is specifically: Perform feature extraction based on real-time environmental parameter data to obtain environmental parameter feature data; Perform environmental geometry modeling according to environmental parameter characteristic data to obtain environmental geometry data; Perform physical attribute mapping on real-time environmental parameter data and environmental geometry data to obtain environmental physical attribute distribution map data; Generate virtual environment texture based on environment physical property distribution map data to obtain environment texture map data; A virtual sensor position interface is generated according to the environmental texture map data to obtain virtual sensor environmental data.

[0009] The feature extraction in the present invention makes the real-time monitoring of the environment more efficient, reduces the complexity of raw data processing, and can significantly improve the computing efficiency, especially when processing large-scale data sets. It provides a more accurate basis for real-time response to environmental changes and sensor adaptability evaluation. By geometrically modeling the environment, the three-dimensional structure of the environment can be accurately reproduced in the virtual space. It can help understand the impact of the environment on sensor operation, especially in complex or difficult-to-reach environments. The acquisition of environmental geometric data provides accurate geometric characteristics, such as surface morphology, spatial layout, etc. Mapping provides a realistic basis for sensor performance simulation and optimization, so that sensors can be accurately evaluated and tuned in more complex physical environments. Virtual texture generation helps to test and optimize sensors under various conditions in a virtual environment, avoiding the need for a large number of field tests in the actual environment. Through these virtual tests, potential problems can be discovered and corresponding design adjustments can be made to avoid problems in advance and improve the adaptability and accuracy of sensors.

[0010] Preferably, the feature extraction is specifically: Slice the window data according to the real-time environmental parameter data to obtain environmental parameter slice data; Calculate the environmental disturbance index according to the environmental parameter slice data to obtain environmental disturbance index data; Calculate the covariance matrix of multiple environmental variables based on the environmental parameter slice data to obtain the correlation data of multiple environmental variables; The environmental disturbance index data and multi-environmental variable correlation data are processed for abnormal environmental spatial fluctuations to obtain environmental parameter characteristic data.

[0011] In the present invention, by performing window data slicing on environmental parameter data, the method can extract representative and distinguishing time periods or data segments from the original environmental data collected in real time. The calculation of the coefficient can help detect the degree of interference of the sensor under different environmental conditions, and promptly discover abnormal conditions caused by environmental changes, thereby providing a more accurate basis for the stability and reliability evaluation of the sensor. The calculation of the covariance matrix helps to reveal the interaction between environmental factors, so as to more comprehensively evaluate the performance of the sensor under different combinations of environmental variables. A comprehensive analysis of the environmental disturbance index and the correlation data of multiple environmental variables is performed to detect potential abnormal fluctuation areas.

[0012] Preferably, the window data slice is specifically: Perform window slicing according to the real-time environment parameter data using preset first window slicing parameter data and preset second window slicing parameter data to obtain first preliminary window slicing data and second preliminary window slicing data respectively; Performing cluster calculation on the first preliminary window slice data and the second preliminary window slice data to obtain first window slice cluster data and second window slice cluster data respectively; Calculating the cluster change rate according to the first window slice clustering data and the second window slice clustering data to obtain first cluster cluster change rate data and second cluster cluster change rate data respectively; Dynamically re-dividing the first window slice parameter data according to the first cluster change rate data to obtain first window re-slicing parameter data, and dynamically re-dividing the first window slice parameter data according to the second cluster change rate data to obtain second window re-slicing parameter data; Perform window slicing according to the real-time environment parameter data using preset first window reslicing parameter data and preset second window reslicing parameter data, to obtain first preliminary window reslicing data and second preliminary window reslicing data respectively; Performing window data feature extraction on the first preliminary window reslicing data and the second preliminary window reslicing data to obtain first window feature data and second window feature data respectively; A cross-window feature matrix is ​​constructed using the first window feature data and the second window feature data, the first preliminary window re-slicing data, and the second preliminary window re-slicing data to obtain environmental parameter slice data.

[0013] In the present invention, real-time environmental data is sliced ​​and repartitioned by presetting window slicing parameters, and large-scale environmental data can be subdivided into smaller time windows. In this way, local features of environmental changes can be better captured, the impact of noise can be reduced, and the accuracy of analysis can be improved. Clustering and cluster change rate analysis are very important for sensor systems. They can detect major changes in the environment in real time, such as drastic fluctuations in temperature, humidity or pressure, and help the system quickly adjust and optimize the operating state. Dynamic repartitioning can effectively cope with changes in uncertainty in the environment, especially under unstable or rapidly changing environmental conditions, and can adjust the slicing strategy in time to avoid the decline in data processing accuracy caused by overly static window slicing strategies. For example, in the dynamic monitoring of climate change, rapidly changing weather conditions can obtain more accurate monitoring results by adjusting the slicing strategy in real time. Feature extraction helps to analyze complex environmental data at a deeper level and helps to establish a more intelligent prediction model. For example, in intelligent manufacturing and precision equipment monitoring, valuable features can be extracted from sensor data to achieve predictive maintenance and intelligent decision-making. Cross-window feature matrix construction can integrate feature data from different time windows to provide rich context information for further data analysis and decision-making. By integrating cross-window data, we can have a more comprehensive understanding of the changing trends of the environment and further improve the accuracy and stability of the model.

[0014] Preferably, the environment geometry modeling is specifically: Get the installation environment geometry data; Perform spatial mapping according to real-time environmental parameter data and installation environment geometric data to obtain installation environment mapping data; Perform geometric estimation of the installation environment according to the real-time environmental parameter data to obtain geometric estimation data of the installation environment; Environmental parameters are fused according to the installation environment geometry estimation data and the installation environment mapping data to obtain environmental geometry data.

[0015] In the present invention, the error caused by environmental uncertainty is reduced and the accuracy of the model is ensured through spatial mapping and geometric prediction. In the spatial mapping process, the geometric form of the installation environment is combined with the real-time environmental data to create a more realistic environmental spatial mapping. By predicting the geometric form of the installation environment, the geometric form changes of the environment where the sensor is located are predicted, especially when the environment is subject to external disturbances such as temperature and humidity, the environmental geometry will change. By effectively synthesizing data from different sources (such as geometric data, real-time environmental data), a multi-dimensional and comprehensive environmental geometry data can be obtained, which provides a more accurate environmental background for the sensor, helps the sensor adjust its own parameter settings, and enhances its working stability and fault tolerance in complex environments.

[0016] Preferably, step S2 specifically includes: Step S21: Perform hardware connection verification according to the edge computing device to obtain hardware connection verification data; Step S22: performing communication channel configuration on the hardware connection verification data to obtain communication channel configuration data; Step S23: reading sensor characteristics according to the communication channel configuration data to obtain sensor characteristic data; Step S24: calibrating parameters according to the sensor characteristic data to obtain calibration parameter data; Step S25: performing gain adjustment on the parameter calibration data to obtain optimized gain value data; Step S26: Initialize parameters according to the optimized gain value data to obtain sensor initial configuration data.

[0017] The hardware connection check in the present invention ensures that the hardware connection between the sensor and the edge computing device is reliable and stable. By checking the hardware connection status, data loss or communication failure caused by poor connection is avoided. By configuring the communication channel according to the hardware connection check result, it is ensured that the sensor data can be accurately transmitted through the appropriate network channel. The communication channel configuration not only improves the stability of data transmission, but also provides a stable communication basis for sensor data acquisition. The sensor characteristic reading step can extract key information (such as sensitivity, range, error characteristics, etc.) from the sensor to ensure that the characteristics of the sensor are fully understood, so that the actual performance of the sensor can be considered during parameter initialization and optimization. By calibrating the sensor, it is ensured that its output data is closer to the true value and the accuracy and consistency of the data are improved. Gain adjustment can optimize the response sensitivity and output range of the sensor. By adjusting the gain, the response of the sensor under different working conditions can be made more in line with actual needs, especially in an environment with inconsistent signal strength or high noise, gain adjustment can significantly improve signal quality and reduce errors. By optimizing the parameter initialization after the gain value, it is ensured that the sensor has the correct working state and performance settings at startup.

[0018] Preferably, step S3 is specifically: Step S31: performing sensor parameter fitting according to the sensor digital twin model and the sensor initial configuration data to obtain a sensor fitting model; Step S32: performing noise simulation on the sensor fitting model to obtain noise interference data; Step S33: simulating the usage scenario according to the sensor fitting model and the noise interference data to obtain sensor simulation data.

[0019] In the present invention, by combining the sensor digital twin model with the initial configuration data of the sensor and fitting the sensor parameters, the various performance parameters of the sensor can be accurately estimated and adjusted. The noise simulation step generates noise interference data by performing interference simulation on the fitting model, which can comprehensively evaluate the performance of the sensor in a complex environment. Noise factors in the real environment (such as electromagnetic interference, vibration, temperature fluctuations, etc.) will affect the output of the sensor. By simulating noise interference, the anti-interference ability of the sensor can be tested and optimized. By combining the sensor fitting model and noise interference data, the use scenario simulation is performed, which can more realistically restore the working state and performance of the sensor in actual application. The scenario simulation can take into account the superposition of multiple factors, such as ambient temperature changes, pressure fluctuations, mechanical vibrations, etc., to comprehensively test the adaptability and reliability of the sensor.

[0020] Preferably, step S4 is specifically: Step S41: obtaining sensor usage target data; Step S42: performing qualified parameter mapping according to the sensor usage target data to obtain sensor qualified parameter data; Step S43: Perform in-depth performance evaluation based on the sensor qualified parameter data and the sensor simulation data to obtain sensor detection data.

[0021] In the present invention, obtaining the sensor usage target data is the first step in the sensor performance evaluation, ensuring that the evaluation work is based on the specific goals required by the sensor in actual use. The sensor usage goals are converted into quantifiable performance standards (such as accuracy, response time, operating temperature, etc.) to ensure that the sensor meets the usage requirements in various performance indicators. Through the combination of qualified parameter data and simulation data, a full range of performance testing is carried out to ensure that the sensor can meet the design standards in actual work, provide accurate and in-depth performance evaluation of the sensor, discover possible defects and optimize them.

[0022] Preferably, the present application also provides a sensor detection system for executing the sensor detection method as described above, the sensor detection system comprising: The sensor twin model building module is used to collect the geometric, mechanical and electrical parameters of the sensor installation environment in real time and build the sensor twin model; The sensor initial configuration module is used to automatically identify and initialize sensor parameters through the local area network according to the edge computing device to obtain the sensor initial configuration data; A sensor simulation module, used to generate sensor simulation data according to the sensor word twin model and the sensor initial configuration data; The performance detection and evaluation module is used to perform performance detection and evaluation based on the sensor simulation data to obtain sensor detection data.

[0023] The beneficial effect of the present invention is that by collecting the geometric, mechanical and electrical parameters of the sensor installation environment in real time and establishing a digital twin model, the present invention provides an accurate and dynamic digital representation for sensor detection. The digital twin model can simulate the physical environment and sensor behavior in the real world, and help track the working status of the sensor and environmental changes in real time. The introduction of edge computing devices can push computing and data processing from the cloud to the vicinity of the sensor, reducing data transmission delays and improving real-time performance and data processing efficiency. Automatic identification and initialization of sensors are completed through the local area network, ensuring rapid configuration and seamless integration of equipment. By combining the digital twin model with the initial configuration data of the sensor, simulation data is generated, and the simulation data can more accurately predict the behavior and performance of the sensor under different working environments. Not only is it limited to the testing of the basic performance of the sensor (such as sensitivity, response time, etc.), it also takes into account the influence of external disturbances, environmental factors and system interactions, and can perform all-round and multi-dimensional performance testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings: Figure 1 A flow chart showing the steps of a sensor detection method according to an embodiment is shown; Figure 2 A flowchart of a method for constructing a sensor word twin model according to an embodiment is shown; Figure 3 A flowchart showing the steps of a sensor initial configuration method according to an embodiment is shown; Figure 4 A flow chart showing the steps of a sensor simulation method according to an embodiment is shown; Figure 5 A flowchart of the steps of a performance detection and evaluation method according to an embodiment is shown. DETAILED DESCRIPTION

[0025] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0027] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0028] See also Figures 1 to 5 , the present application provides a sensor detection method, comprising the following steps: Step S1: Collect the geometric, mechanical and electrical parameters of the sensor installation environment in real time and build a digital twin model of the sensor; Specifically, a group of environmental sensing devices (such as temperature sensors, humidity sensors, light sensors, etc.) are installed to collect the geometric, mechanical and electrical parameters of the environment where the sensor is located. Geometric parameters include the physical dimensions of the installation location, spatial position relationships, etc.; mechanical parameters include stress and vibration in the environment; electrical parameters include voltage, current, power, etc. A digital twin model is established based on the geometric shape, working conditions and mechanical properties of the sensor in the actual environment. The model is based on the data of actual sensors and environmental parameters, and is modeled using digital 3D modeling tools and simulation software (such as CAD software, ANSYS, MATLAB, etc.), including simulation data of the sensor's physical structure, external environmental changes, mechanical response and electrical performance.

[0029] Step S2: Automatically identify and initialize sensor parameters through the local area network according to the edge computing device to obtain initial configuration data of the sensor; Specifically, edge computing devices are used to communicate with sensors through local area networks (such as Wi-Fi or Ethernet) to automatically identify the sensor's model, specifications, production information, etc. The edge device can identify the sensor through its own ID or QR code and read the basic configuration data of the sensor (such as sensitivity, range, operating voltage, etc.). The identified sensor is initialized and set through the edge computing device, including setting the sensor's initial working state, the sensor's sampling frequency, the data transmission method (such as serial communication, wireless transmission, etc.) and the working mode (such as continuous working mode, periodic detection mode, etc.). The edge computing device stores the configuration data of the sensor after initialization, and uploads this data to the cloud platform or local server through the local area network.

[0030] Step S3: Generate sensor simulation data according to the sensor digital twin model and the sensor initial configuration data; Specifically, the digital twin model and the initial configuration data of the sensor are used to generate the working state simulation of the sensor under different environments (such as temperature, humidity, current fluctuation, etc.) through simulation software. The simulation data includes the output signal, error, response time, etc. of the sensor. The simulation process needs to take into account various environmental changes and the physical characteristics of the sensor. Based on the initial configuration data and the environmental model, the response characteristics and performance of the sensor are calculated. For example, based on the geometric and mechanical characteristics of the sensor, its output error under different vibration, temperature, and humidity conditions is simulated; based on electrical parameters, the fluctuation of the sensor voltage and current response is simulated. In the simulation, the digital twin model is combined with the known mathematical model and sensor characteristics to generate a complete simulation data set, including the expected output of the sensor under various environments. The accuracy of the simulation data directly affects the effectiveness of the performance evaluation, so multiple simulations are required and the consistency with the actual data is verified.

[0031] Step S4: Perform performance testing and evaluation based on the sensor simulation data to obtain sensor testing data.

[0032] Specifically, according to the application requirements and expected performance standards of the sensor, specific evaluation indicators are set, such as sensitivity, response time, stability, linearity, etc. For each performance parameter, a performance threshold is determined. The performance of the sensor is evaluated by comparing the simulated data with the performance indicators (performance thresholds) set by the standard. The detection process compares the simulated data with the actual sensor measurement results to check the difference between the two. The difference is used to identify sensor failures or deviations. By analyzing the deviation between the simulated data and the actual detection data, it can be further determined whether the sensor has a fault or performance degradation. For example, if there is a significant difference between the simulated data and the actual output, the sensor has problems such as drift, aging, or improper installation. Based on the performance evaluation and fault diagnosis results, the sensor detection data is generated and provided to the user for subsequent maintenance, calibration or optimization measures.

[0033] Preferably, step S1 specifically includes: Step S11: deploy an environmental sensor array to collect real-time environmental parameter data through edge devices; Specifically, multiple environmental sensors are arranged in the sensor installation environment to monitor environmental parameters in real time, such as temperature, humidity, air pressure, light intensity, etc., such as DHT22 temperature and humidity sensor, BME680 air pressure and air quality sensor, etc. In the temperature sensor installation environment of a factory, temperature sensors (such as DS18B20), humidity sensors (such as DHT22), and air pressure sensors (such as BME280) are arranged. All sensors are connected to an edge computing device (such as Raspberry Pi or industrial edge computing gateway). The edge device collects sensor data in real time through an interface (such as I2C, SPI or GPIO) and stores it. The data includes real-time information such as ambient temperature, humidity, and air pressure. The edge device reads the temperature, humidity, and air pressure data once a second, and uploads the data to the cloud platform through a wireless network (such as Wi-Fi or Ethernet) for further processing.

[0034] Step S12: Generate a virtual sensor environment according to the real-time environmental parameter data to obtain virtual sensor environment data; Specifically, based on the environmental parameter data collected in real time, a computational model or simulation software is used to generate a virtual sensor working environment to simulate the response and performance of the sensor in a real environment. The actual working environment of the temperature sensor has a temperature range of 20°C to 40°C, a humidity range of 30% to 80%, and a pressure range of 950hPa to 1050hPa. On this basis, a virtual simulation tool (such as ANSYS, MATLAB or Simulink) is used to generate simulated environmental data to deduce the response of the sensor under different temperatures, humidity and air pressures. The model is used to calculate the virtual output data of the sensor under different environmental conditions (such as the output voltage of the temperature sensor). Through simulation, it is found that when the temperature is 30°C, the humidity is 50%, and the air pressure is 1010hPa, the output voltage of the temperature sensor is 2.5V.

[0035] Step S13: using laser scanning to obtain the geometric shape of the installation position, obtaining sensor geometric shape data, and performing geometric reconstruction based on the sensor geometric shape data to obtain sensor installation geometric parameter data; Specifically, a laser scanning device (such as a Leica or FARO laser scanner) is used to scan the installation location of the sensor. The laser scanner can accurately obtain the three-dimensional geometric data of the installation location, including the size and position of the wall, bracket, etc. Use Leica ScanStation P40 for laser scanning to scan the geometric shape of the sensor installation location and accurately obtain the three-dimensional coordinate data of each installation point. The laser scanning data is input into a three-dimensional modeling software (such as AutoCAD, SolidWorks or MeshLab) for geometric reconstruction to generate a three-dimensional model of the sensor installation environment. The model will display geometric parameters such as the distance and relative position between the sensor and other devices. Through the geometric reconstruction software, the point cloud data obtained by the laser scanning is converted into a three-dimensional CAD model to obtain the precise geometric data of the sensor installation location (such as the distance between the sensor and the wall, the angle of the bracket, the spatial coordinates of the installation point, etc.).

[0036] Step S14: calibrating the mechanical parameters of the installation environment using the pressure sensor and the accelerometer to obtain the mechanical characteristic data of the sensor; Specifically, pressure sensors (such as Honeywell's differential pressure sensor) and accelerometers (such as ADXL345) are installed in the sensor installation environment to monitor mechanical changes in the environment, such as pressure fluctuations, vibrations, and accelerations. Accelerometers and pressure sensors are arranged in the sensor installation environment to monitor vibration and pressure changes in the environment in real time. For example, an accelerometer is used to measure the vibration frequency of the installation environment, and a pressure sensor is used to monitor air pressure fluctuations. The mechanical properties of the sensor under different environmental conditions are calibrated through the environmental data collected by the sensor. For example, the response characteristics of the sensor in such an environment are calculated based on the vibration frequency measured by the accelerometer and the pressure fluctuation measured by the pressure sensor. The vibration frequency in the environment is 50Hz, and the pressure fluctuation is ±5Pa. The sensitivity of the sensor to these mechanical changes is calibrated through the response data of the sensor to obtain the mechanical property data of the sensor.

[0037] Step S15: using an oscilloscope to measure the connector impedance, voltage waveform and grounding state to obtain electrical characteristic data; Specifically, use an oscilloscope (such as Tektronix or Keysight) to measure the electrical characteristics of the sensor connector. First, measure the impedance of the sensor power pin to ensure that the electrical connection is stable. Observe the sensor voltage waveform to ensure that the voltage output is stable and meets the design requirements. Finally, test the grounding status to ensure that the grounding is good. Use an oscilloscope to measure the voltage signal of the sensor. Assume that the measured voltage waveform is a 1kHz sine wave with an amplitude of 5V. Next, test the grounding status and confirm that the grounding voltage is 0V and the grounding is good. Based on the measured data, calibrate the electrical characteristics of the sensor, including the resistance of the connector, the stability of the output signal, and the grounding status. The measured electrical characteristics data include an impedance of 50Ω, a voltage waveform of a sine wave (1kHz, 5V), and a good grounding.

[0038] Step S16: Establish a digital twin model of the sensor installation geometric parameter data, sensor mechanical characteristic data, electrical characteristic data and virtual sensor environment data to obtain a sensor digital twin model.

[0039] Specifically, using software such as MATLAB / Simulink or ANSYS, the geometric data, mechanical data, electrical data, and environmental data of the sensor are input into the simulation platform to generate a digital twin model. The model can simulate the response and performance of the sensor under different environmental conditions. Use the above integrated data to establish a digital twin model to simulate the performance of the sensor under different environmental conditions. Through this model, the status of the sensor in the actual working environment can be monitored in real time, and predictive maintenance can be performed. In MATLAB / Simulink, a virtual model containing sensor geometric parameters, environmental data, electrical characteristics, etc. is established to simulate the output of the sensor under temperature and humidity changes, and the performance of the sensor is monitored in real time.

[0040] Preferably, step S12 is specifically: Perform feature extraction based on real-time environmental parameter data to obtain environmental parameter feature data; Specifically, real-time environmental parameter data includes multiple sensing data, such as temperature, humidity, air pressure, light, etc. The data comes from sensor arrays or other environmental monitoring devices. The goal of environmental parameter feature extraction is to extract meaningful features from these raw data, such as change trends, fluctuation amplitudes, periodicity, etc. The extraction methods include statistical analysis (such as mean, variance, kurtosis, skewness, etc.), frequency domain analysis (such as Fourier transform), and time series analysis. For each sensor data collected, data preprocessing (denoising, normalization, etc.) is first performed. Then, the data is sliced ​​and features are extracted using a filtering algorithm or sliding window technology. Environmental features include: temperature change rate, humidity fluctuation range, air pressure periodicity, etc.

[0041] Perform environmental geometry modeling according to environmental parameter characteristic data to obtain environmental geometry data; Specifically, environmental geometric modeling is to model the spatial layout of the environment, which requires the collection of geometric data of the environment, such as space dimensions, location of obstacles, shape of walls, etc. If the environment has collected geometric data through laser scanning, 3D cameras and other devices, these data can be used directly for modeling. Otherwise, simulation or mathematical modeling methods can be used to generate a geometric model of the environment. Environmental geometric modeling usually includes the following methods: obtaining point cloud data of the environmental space through a laser rangefinder, and then converting the point cloud data into a three-dimensional model. Alternatively, environmental modeling is performed through a camera and a depth sensor (such as Kinect) to obtain the geometric information of the space. Alternatively, if there are ready-made architectural design drawings or CAD models, these data can be used directly for modeling.

[0042] Perform physical attribute mapping on real-time environmental parameter data and environmental geometry data to obtain environmental physical attribute distribution map data; Specifically, real-time environmental parameter data (such as temperature, humidity, etc.) are mapped with environmental geometric data (such as the walls, floors, ceilings, etc. of the room). The corresponding physical properties (such as temperature, humidity, etc.) are calibrated at each spatial position to generate an environmental physical property distribution map. The mapping process uses an interpolation algorithm (such as nearest neighbor interpolation, bilinear interpolation, cubic interpolation, etc.) to assign environmental data to each point in the space. According to the spatial layout of the environmental geometric data, the space is divided into grids, and each grid point is assigned a physical property (such as temperature or humidity). Use a three-dimensional visualization tool (such as OpenGL or Unity3D) to generate an environmental physical property distribution map, which represents the physical states of temperature, humidity, etc. at different spatial locations.

[0043] Generate virtual environment texture based on environment physical property distribution map data to obtain environment texture map data; Specifically, the purpose of texture generation is to visually express physical properties (such as temperature, humidity, etc.) so that they can be reflected in the virtual environment in the form of images or textures. This includes mapping physical property data to corresponding texture layers, such as using color gradients on a temperature distribution map to represent different temperature ranges, or using different textures on a humidity distribution map to represent changes in humidity. The generated physical property distribution map is combined with a three-dimensional environment model using texture mapping technology in computer graphics. For example, color bands or texture maps are used to express temperature changes (such as red for high temperatures and blue for low temperatures), and these textures are generated and applied with the help of tools such as Photoshop, Blender, or Unity. For temperature distribution maps, a temperature map is generated in which areas with higher temperatures (such as near windows) are represented by red and areas with lower temperatures (such as near walls) are represented by blue. By applying these textures to the surface of the virtual environment model, the temperature distribution of the entire room is intuitively displayed.

[0044] A virtual sensor position interface is generated according to the environmental texture map data to obtain virtual sensor environmental data.

[0045] Specifically, based on the environmental texture map data, the position of the sensor in the virtual environment is determined, and the sensor is associated with a specific location in the environment (such as a wall, window, etc.). Each virtual sensor will collect physical property data (such as temperature, humidity, etc.) from the location and generate corresponding environmental data. The process of generating the location interface is usually implemented through a virtual simulation tool, where the "location" of the sensor can be specified based on the physical space layout. A location is specified for each sensor in the virtual environment, and the actual working environment data of the sensor is generated based on the texture map data of the location. Each sensor location can be mapped to a specific physical property (such as temperature or humidity at a specific location) and its behavior is simulated in the virtual environment. The virtual sensor location is in the center of the room, and the sensor environmental data is generated based on the temperature map data of the location (such as 21°C). The virtual sensor obtains the required environmental parameters based on the location data in the room.

[0046] Preferably, the feature extraction is specifically: Slice the window data according to the real-time environmental parameter data to obtain environmental parameter slice data; Specifically, in order to analyze the dynamic changes of environmental parameters, the original time series data is divided into several windows, each of which contains a fixed length or a fixed amount of data. For example, a time window (such as 5 seconds or 10 seconds) is selected, and then the data slices within the time period are extracted for analysis.

[0047] Calculate the environmental disturbance index according to the environmental parameter slice data to obtain environmental disturbance index data; Specifically, the disturbance index is an indicator used to quantify the magnitude and frequency of changes in environmental parameters. The disturbance index is calculated using the volatility, standard deviation, or rate of change of the data. For example, when the temperature, humidity, or air pressure changes greatly, the environmental disturbance index will be higher. For each data slice, the standard deviation or rate of change (such as the amount of change per second) of each environmental parameter (such as temperature, humidity, and air pressure) is calculated as a component of the disturbance index. By calculating the disturbance index of each window, the environmental disturbance index data for each time window can be obtained.

[0048] Calculate the covariance matrix of multiple environmental variables based on the environmental parameter slice data to obtain the correlation data of multiple environmental variables; Specifically, the covariance matrix is ​​an important tool for measuring the correlation between multiple environmental variables. By calculating the covariance matrix between environmental parameters, we can understand the relationship between different environmental parameters (such as temperature, humidity, air pressure, etc.). For each slice data of environmental parameters (such as temperature, humidity and air pressure), calculate the covariance matrix between these variables. The diagonal elements of the covariance matrix represent the variance of each environmental parameter, and the off-diagonal elements represent the correlation between different environmental parameters. Through this matrix, we can understand which environmental parameters have a strong correlation and which have a weak correlation.

[0049] The environmental disturbance index data and multi-environmental variable correlation data are processed for abnormal environmental spatial fluctuations to obtain environmental parameter characteristic data.

[0050] Specifically, the processing goal of abnormal fluctuations is to remove noise and abnormal fluctuations from environmental data to ensure that the obtained feature data can accurately reflect the actual state of the environment. Abnormal fluctuations include short-term fluctuations caused by sudden climate change, equipment failure, etc. These fluctuations are accidental and unrepresentative. Anomaly detection is performed on the environmental disturbance index and the correlation data of multiple environmental variables to identify abnormal fluctuations in the environment. Statistical methods (such as Z-score detection) or machine learning methods (such as outlier detection algorithms) are used to discover abnormal fluctuations. The environmental disturbance index data and the data in the covariance matrix are Z-score standardized to calculate the Z-score of each data point. If the Z-score of a data point is greater than a certain threshold (such as 3), it is considered an abnormal fluctuation. The disturbance index of a time window is 3.5 and the Z-score is 3, indicating that this is an abnormal fluctuation and needs further processing. After detecting abnormal fluctuations, the data can be smoothed or denoised to remove interference factors and obtain more accurate environmental parameter characteristics. For the identified abnormal fluctuations, the sliding average method is used for smoothing. For example, the disturbance index data of the last five windows are averaged to eliminate short-term abnormal fluctuations.

[0051] Preferably, the window data slice is specifically: Perform window slicing according to the real-time environment parameter data using preset first window slicing parameter data and preset second window slicing parameter data to obtain first preliminary window slicing data and second preliminary window slicing data respectively; Specifically, the window size is set, for example, the "first window slice parameter" is defined as 5 seconds, and the "second window slice parameter" is defined as 10 seconds. The environmental parameter data is collected once every second. According to the preset first window slice parameter (5 seconds) and second window slice parameter (10 seconds), the first preliminary window slice data and the second preliminary window slice data are cut out from the real-time environmental parameter data.

[0052] Performing cluster calculation on the first preliminary window slice data and the second preliminary window slice data to obtain first window slice cluster data and second window slice cluster data respectively; Specifically, clustering calculation is performed on the first preliminary window slice data and the second preliminary window slice data. Clustering algorithms such as K-means can divide the data into multiple clusters according to similarity. Each cluster represents a data set with similar environmental parameter change patterns. The first window slice data (5-second window) is processed using the K-means clustering algorithm and divided into two clusters: the first cluster: [25°C, 50%, 1013hPa], [25.1°C, 50.1%, 1013.1hPa], [25.2°C, 50.2%, 1013.2hPa], the second cluster: [25.3°C, 50.3%, 1013.3hPa], [25.4°C, 50.4%,1013.4hPa]. Similar processing is performed on the second window slice data (10-second window), resulting in: first cluster: [25°C, 50%, 1013hPa], ..., [25.4°C, 50.4%, 1013.4hPa], second cluster: [25.5°C, 50.5%, 1013.5hPa], ..., [25.9°C, 50.9%, 1013.9hPa].

[0053] Calculating the cluster change rate according to the first window slice clustering data and the second window slice clustering data to obtain first cluster cluster change rate data and second cluster cluster change rate data respectively; Specifically, the rate of change of each cluster is calculated, and the rate of change can be obtained by calculating the time series change of the cluster center. The rate of change of the cluster describes how the cluster changes in the time dimension and reflects the degree of dynamic change of the environment. For the first cluster of the first window slice data, the rate of change of temperature and humidity is calculated: temperature change rate = (25.2°C - 25°C) / (2s - 0s) = 0.1°C / s, humidity change rate = (50.2% - 50%) / (2s - 0s) = 0.1% / s. For the first cluster of the second window slice data, the rate of change of temperature and humidity within the cluster is calculated: temperature change rate = (25.4°C - 25°C) / (4s - 0s) = 0.1°C / s, humidity change rate = (50.4% - 50%) / (4s - 0s) = 0.1% / s. The change rate data of the first cluster is obtained: [0.1°C / s, 0.1% / s], and the change rate data of the second cluster is obtained: [0.1°C / s, 0.1% / s].

[0054] Dynamically re-dividing the first window slice parameter data according to the first cluster change rate data to obtain first window re-slicing parameter data, and dynamically re-dividing the first window slice parameter data according to the second cluster change rate data to obtain second window re-slicing parameter data; Specifically, the window is dynamically repartitioned according to the cluster change rate data. If the cluster change rate is large, it means that the environment changes more drastically, and the window needs to be repartitioned to better capture the volatility of the environment. The length of the slice is dynamically adjusted according to the cluster change rate, or it is divided in time according to the change rate. For example, if the cluster change rate of the first window is greater than 0.1°C / s, the time interval of the slice can be shortened from 5 seconds to 3 seconds.

[0055] Perform window slicing according to the real-time environment parameter data using preset first window reslicing parameter data and preset second window reslicing parameter data, to obtain first preliminary window reslicing data and second preliminary window reslicing data respectively; Specifically, based on the adjusted re-slicing parameter data (such as the adjusted window time), the real-time environment parameter data is re-sliced.

[0056] Performing window data feature extraction on the first preliminary window reslicing data and the second preliminary window reslicing data to obtain first window feature data and second window feature data respectively; Specifically, feature extraction is performed on each window slice data (whether original or re-sliced ​​data), and the features include mean, standard deviation, maximum value, minimum value, amplitude, volatility, etc.

[0057] Specifically, more importantly, matrix decomposition is performed on the first preliminary window reslicing data and the second preliminary window reslicing data to obtain first matrix decomposition data and second matrix decomposition data respectively; main feature extraction is performed on the first matrix decomposition data and the second matrix decomposition data to obtain first main feature data and second main feature data respectively; clustering calculation is performed based on the first main feature data and the second main feature data to obtain first main feature clustering data and second main feature clustering data respectively; cross-window aggregation is performed on the first main feature data and the second main feature clustering data to obtain cross-window aggregation data; feature extraction is performed based on the cross-window aggregation data to obtain first window feature data; time series difference accumulation calculation is performed on the first preliminary window reslicing data and the second preliminary window reslicing data to obtain first window change rate feature data and second window change rate feature data respectively. feature data; performing spatial dimension mapping according to the first window change rate feature data and the second window change rate feature data, respectively obtaining the first window virtual space grid data and the second window virtual space grid data; performing spatial characteristic stratification according to the first window virtual space grid data and the second window virtual space grid data, respectively obtaining the first space characteristic stratification data and the second space characteristic stratification data; performing multi-scale recursive aggregation analysis according to the first space characteristic stratification data and the second space characteristic stratification data, obtaining multi-scale recursive aggregation data; performing graph feature extraction according to the multi-scale recursive aggregation data, obtaining the second window feature data; wherein the cross-window aggregation is specifically as follows: performing similarity calculation according to the first main feature clustering data and the second main feature clustering data, obtaining the main feature similarity data; aggregating the first main feature data and the second main feature data according to the main feature similarity data, obtaining the cross-window aggregation data.

[0058] The first preliminary window re-slice data and the second preliminary window re-slice data are respectively constructed as matrices, where the rows represent time points and the columns represent different environmental variables (such as temperature, humidity, and air pressure). The matrix is ​​decomposed into two low-dimensional matrices. The first two main modes are selected from the decomposition, and their corresponding principal components are extracted. For example, the first principal component reflects the main changes in temperature fluctuations; the second principal component reflects the coordinated changes in humidity and air pressure. According to the matrix decomposition results, the main features are extracted by singular value decomposition (SVD) or principal component analysis (PCA) to obtain the first principal feature data and the second principal feature data, respectively. The first principal feature data and the second principal feature data are clustered respectively to identify the distribution of feature patterns. For example, the first principal feature data [0.9, 0.8, 0.7] are clustered into two categories, and the result is [1, 1, 2], indicating that the first and second time points belong to the same category, and the third time point belongs to another category. The clustering process of the second window is similar, and the second principal feature cluster data is output. The first principal feature cluster data and the second principal feature cluster data are compared across windows to identify the change pattern and obtain the cluster label sequence alignment sequence. Generate aggregation indicators such as consistency and drift degree according to the sequence alignment of clustering label sequences. For example, the first window cluster is [1,1,2], and the second window cluster is [1,2,2]. Cross-window consistency indicator: the first two time points are consistent (1 and 1), and the next time point changes (2 and 2). According to the aggregation indicator and the cross-window consistency indicator, the first main feature clustering data and the second main feature clustering data are aggregated through weight calculation (that is, weight generation based on the aggregation indicator and the cross-window consistency indicator. If the consistency is high (such as >0.8), it depends more on the main feature data of the first window. If the drift degree is high, consider increasing the feature weight of the second window to highlight the time change) to obtain cross-window aggregated data. The cross-window aggregated data is fused with the main feature data to generate a feature vector.

[0059] Calculate the difference and cumulative change rate for the time series in the first preliminary window reslicing data and the second preliminary window reslicing data. The difference represents the change value of adjacent data points in the time series. The cumulative change rate represents the cumulative change from the initial moment to the current moment. Map the change rate feature data to the spatial dimension to generate a virtual space grid. The spatial grid points are composed of the characteristic values ​​of the position and environmental variables. Use the interpolation algorithm to assign the change rate values ​​of non-grid points to the grid. Layer the virtual space grid data by space and time to extract the layered characteristics. Group the grid points by position and calculate the mean and gradient of each group of grid points. Aggregate the change rate of the grid points by time to generate the characteristic data of each layer. The first spatial characteristic layered data: mean characteristic: μ=[1.5,2.5], gradient characteristic: =[1,0.5]; Second spatial characteristic layered data: mean characteristic: μ=[0.75,1.5], gradient characteristic: =[0.5,0.25].

[0060] The spatial characteristic layered data is aggregated and analyzed using a multi-scale recursive method to extract global features. Recursive analysis is used to capture short-term, medium-term and long-term trends: , for The recursive characteristic value corresponding to the moment, is the weight data of the first time scale, is the recursive characteristic value at the previous moment, and the initial moment (t=1) Set to the initial value of the first spatial feature layer data, such as the mean or gradient at the first time point, is the second time scale weight data, It is the characteristic value at the current time point, which comes from the spatial characteristic layered data. Short-term trend: A smaller value, such as 0.2, makes the recursive feature more dependent on the current feature value. A larger value, such as 0.8, highlights the contribution of the features at the current time point. Medium-term trend: A medium value, such as 0.5, balances historical trends with current features. Medium, such as 0.5. Long-term trend: Larger values, such as 0.8, emphasize historical trends. A smaller value, such as 0.2, reduces the impact of fluctuations in the current data.

[0061] Use spatial grid points as nodes of the graph, and the distance or gradient between nodes as edge weights. Extract global features of the graph (such as centrality and connectivity) to obtain graph features. Fuse recursively aggregated data with graph features to generate the second window feature vector.

[0062] A cross-window feature matrix is ​​constructed using the first window feature data and the second window feature data, the first preliminary window re-slicing data, and the second preliminary window re-slicing data to obtain environmental parameter slice data.

[0063] Specifically, the feature data of the first window and the second window are combined to form a cross-window feature matrix. This matrix can contain the feature data of each window and is arranged in chronological order. Window 1 feature data: [25.1°C, 50.1%, 1013.1hPa], Window 2 feature data: [25.55°C, 50.55%, 1013.55hPa]. Cross-window feature matrix: [ [25.1, 50.1, 1013.1], [25.55, 50.55, 1013.55]].

[0064] Preferably, the environment geometry modeling is specifically: Get the installation environment geometry data; Specifically, the geometric data of the installation environment is obtained through a sensor array (such as a laser scanner, RGB-D camera, depth camera, etc.) or a 3D scanning device. The device can collect information such as the spatial layout of the scene, the surface of the object, and the spatial coordinates. The collected geometric data includes the three-dimensional point cloud data of the scene, the boundary of the object, the surface curvature, and other information. These data can be used to model the three-dimensional form of the environment.

[0065] Perform spatial mapping according to real-time environmental parameter data and installation environment geometric data to obtain installation environment mapping data; Specifically, the geometric data of the environment is spatially mapped with real-time environmental parameters (such as temperature, humidity, air pressure, etc.) through spatial interpolation, registration, coordinate transformation and other technologies. The goal is to bind each collected environmental parameter to the corresponding geometric position to generate a three-dimensional spatial model with environmental parameters. According to the change law of environmental parameters, interpolation is performed between known geometric data position points to infer the environmental parameter values ​​at other spatial positions. Align the real-time collected environmental data with the geometric model to ensure that the spatial coordinates match the environmental parameter data. Certain environmental parameter data (such as temperature, humidity, etc.) are known, and there is a corresponding three-dimensional point cloud model. The environmental parameters are combined with the point cloud data through interpolation methods (such as Kriging interpolation, inverse distance weighted method, etc.). For example, the temperature data corresponds to 25°C with the coordinates (x1, y1, z1), and the humidity data corresponds to 50% with the coordinates (x2, y2, z2). Through spatial mapping, a three-dimensional data set containing environmental parameters can be obtained. Perform geometric estimation of the installation environment according to the real-time environmental parameter data to obtain geometric estimation data of the installation environment; Specifically, based on the existing geometric data and environmental parameter data, use prediction algorithms or modeling methods (such as machine learning, data fitting, etc.) to estimate the environmental geometry. Through the correlation between historical data and real-time environmental parameters, predict the changing trend of environmental geometry in the future. Time series prediction methods (such as ARIMA, LSTM networks, etc.) or machine learning methods based on environmental characteristics (such as regression analysis, support vector machines, etc.) can be used to predict the geometric changes of the environment. For example, predict the position change of a device or object in a certain spatial area. Through historical data analysis, it is found that temperature changes will cause some objects to deform. For example, due to temperature changes, the length or position of some equipment may change slightly. Through prediction models (such as regression analysis), the changes in device position (x, y, z) in the next 10 minutes can be predicted. Environmental parameters are fused according to the installation environment geometry estimation data and the installation environment mapping data to obtain environmental geometry data.

[0066] Specifically, the estimated installation environment geometry data and the environment mapping data are fused, and the advantages of both are combined to obtain a more accurate and complete environment geometry data. Environment parameter fusion often relies on data fusion algorithms (such as Kalman filtering, particle filtering, etc.), which can weightedly fuse data from different sources (such as geometry data, environment parameter data) to obtain an optimal environment geometry model. Use weighted averaging, Kalman filtering and other methods to deal with uncertainty and obtain a more accurate environment geometry model. For example, Kalman filtering can dynamically fuse real-time collected geometry data with environmental parameters to predict and correct changes in environment geometry data. There are the following data: Environmental parameter data obtained through spatial mapping, such as the correspondence between temperature, humidity and spatial position. Changes in the position of the environment geometry obtained by the prediction model at future moments. The two are fused through the Kalman filtering algorithm to obtain a more accurate environment geometry data. For example, at a certain location in the installation environment, changes in temperature and humidity cause changes in the device position. Kalman filtering can combine these changes and ultimately give more accurate location information and environmental parameters for the device: Preferably, step S2 specifically includes: Step S21: Perform hardware connection verification according to the edge computing device to obtain hardware connection verification data; Specifically, a physical connection is made between the edge computing device and the sensor to verify the normal connection of the sensor's power supply and signal transmission channel (such as I2C, SPI, UART, etc.). Check the stability of the connection to see if there are problems such as unstable power supply and poor contact. You can confirm whether the connection is successful by reading the sensor's ID or sending back a self-test signal. Use a simple "heartbeat" signal or self-test signal to confirm the hardware connection. Verify whether the connection is stable by detecting the sensor's response time and data integrity. The connected sensor is a temperature sensor. The edge computing device will send a connection request (such as an initialization signal or a query command) to the sensor and wait for the sensor to return its unique ID and response code. If the returned data is complete and there is no error code, the connection verification is successful; if the verification fails, a hardware failure prompt is returned.

[0067] Step S22: performing communication channel configuration on the hardware connection verification data to obtain communication channel configuration data; Specifically, according to the hardware connection verification results, configure the communication protocol between devices. Different sensors support different communication methods, such as I2C, SPI, UART, etc. Ensure that the edge computing device and the sensor can exchange data correctly. During the configuration process, you need to set the transmission rate, data format, communication parameters, etc. to ensure that the data can be transmitted correctly. According to the communication interface standard of the sensor, set the communication port parameters of the edge device (such as baud rate, clock frequency, etc.). Use the I2C protocol for communication. The edge computing device configures the I2C bus frequency to 100kHz and the device address to 0x48.

[0068] Step S23: reading sensor characteristics according to the communication channel configuration data to obtain sensor characteristic data; Specifically, use the configured communication channel to interact with the sensor and read the basic characteristics of the sensor, such as sensor model, firmware version, sampling rate, sensor range, etc. Obtain basic information about the sensor by issuing specific commands (such as "read device ID" or "read sensor characteristics" commands). Read the characteristic data of the temperature sensor, and the returned characteristic data includes model, sampling rate, range, etc.: Sensor model: DHT22, firmware version: V1.2, sampling rate: 1Hz, range: -40°C ~ 80°C, accuracy: ±0.5°C.

[0069] Step S24: calibrating parameters according to the sensor characteristic data to obtain calibration parameter data; Specifically, the sensor is calibrated based on the read sensor characteristic data. The calibration process usually involves measuring the error between the output of the sensor and the standard reference value and making corrections. Calibration parameters include offset correction, sensitivity correction, zero calibration, etc. A known reference standard is used to correct the output value of the sensor to make it more consistent with the actual value. For example, if the sensor is a temperature sensor, the deviation of the temperature sensor can be calibrated by comparing it with a standard thermometer. The offset and sensitivity are calculated by linear regression or other methods.

[0070] Step S25: performing gain adjustment on the parameter calibration data to obtain optimized gain value data; Specifically, the gain of the sensor is adjusted according to the calibration results. Gain adjustment is to amplify or reduce the output signal of the sensor to make its output value more accurate and close to the actual measured value. The purpose of gain adjustment is to minimize the error and improve the accuracy and stability of the sensor. The gain factor is used to correct the output of the sensor. For example, if the actual output of the sensor is found to be lower than the standard value after calibration, the gain can be increased, otherwise the gain can be reduced.

[0071] Step S26: Initialize parameters according to the optimized gain value data to obtain sensor initial configuration data.

[0072] Specifically, the sensor is initialized using the optimized gain value and calibration data. The initialization process usually involves applying the sensor's setting parameters (such as gain, bias, sampling frequency, etc.) to the sensor's firmware or hardware configuration. Based on the previous calibration and gain adjustment results, the sensor's operating parameters are set and saved. These settings will serve as the sensor's initial configuration data.

[0073] Preferably, step S3 is specifically: Step S31: performing sensor parameter fitting according to the sensor digital twin model and the sensor initial configuration data to obtain a sensor fitting model; Specifically, the digital twin model of the sensor includes the physical properties of the sensor (such as temperature, humidity, pressure, etc.), as well as the response characteristics under different environmental conditions. The initial configuration data includes the parameter settings of the sensor, such as gain, sampling rate, working range, etc. A sensor fitting model is established by fitting the mathematical formula or physical model in the sensor digital twin model with the initial configuration data. The sensor fitting model is a mathematical model that describes the relationship between the output and input of the sensor. The sensor model parameters are optimized through least squares method, nonlinear regression, support vector machine, etc., so that the output of the model is as close as possible to the output of the real sensor. According to different environmental conditions, the working environment data of the sensor (such as temperature, humidity, etc.) is input, the predicted output is calculated, and the sensor model parameters are adjusted to minimize the error of the fitting result.

[0074] Step S32: performing noise simulation on the sensor fitting model to obtain noise interference data; Specifically, the sensor may be interfered by various noises in actual use, which come from electronic noise, environmental fluctuations, power supply noise, etc. In order to simulate the working state of the sensor more accurately, a noise term is added to the fitting model. Noise simulation can be achieved through different noise models (such as Gaussian noise, white noise, etc.). Noise is usually random and has certain statistical properties (mean, standard deviation, etc.). Based on the output of the sensor fitting model, a noise term is added to generate simulated noisy data. This can be achieved by generating random numbers that obey a specific distribution (such as Gaussian distribution) and adding them to the fitting results.

[0075] Step S33: simulating the usage scenario according to the sensor fitting model and the noise interference data to obtain sensor simulation data.

[0076] Specifically, the use scenarios of sensors usually include multiple factors, such as environmental conditions such as temperature, humidity, and pressure, which affect the performance of the sensor. During the simulation process, simulation data is generated based on the actual use scenarios (such as temperature changes, equipment vibration, environmental humidity, etc.). The key to the simulation scenario is to combine the sensor fitting model and noise data with the specific use scenario data to simulate the performance of the sensor in various working environments. According to different use scenarios (such as ambient temperature changes, sensor force, etc.), by inputting different environmental parameters, calculate the response of the sensor under these conditions. Use simulation tools (such as MATLAB / Simulink, Python, etc.) to simulate environmental changes, input various environmental data (such as changing temperature or humidity), and output sensor simulation data.

[0077] Preferably, step S4 is specifically: Step S41: obtaining sensor usage target data; Specifically, sensor usage target data refers to the performance goals or standards that the sensor should meet in actual applications, such as measurement accuracy, response time, range, error tolerance, etc. The data comes from multiple sources, such as product design requirements, industry standards, actual test results, etc. The usage target data helps define the "expected behavior" or "qualified standards" of the sensor. The process of obtaining usage target data requires extracting relevant standards from the system design phase or engineering specifications. The maximum error range of the sensor (such as ±0.5°C), response time requirements (such as less than 1 second), operating temperature range (such as -40°C to 85°C), etc.

[0078] Step S42: performing qualified parameter mapping according to the sensor usage target data to obtain sensor qualified parameter data; Specifically, qualified parameter mapping is to compare the target data with the actual performance of the sensor to determine which sensor parameters meet the requirements within the target range. The mapping process can be achieved by comparing the simulated output data of the sensor with the target data. Identify whether the sensor meets the design requirements and whether it meets the requirements in terms of accuracy, response speed, stability, etc. The core of the mapping is to match the actual measured data (such as simulated data) with the target data to determine whether it meets the qualified standards. The target accuracy of the temperature sensor is ±0.5°C, and qualified parameter mapping requires comparing the simulated data with the target range. For example, if the simulated data indicates that the temperature change is within the range of ±0.4°C, the target accuracy is met.

[0079] Step S43: Perform in-depth performance evaluation based on the sensor qualified parameter data and the sensor simulation data to obtain sensor detection data.

[0080] Specifically, deep performance evaluation is a comprehensive analysis and evaluation of the performance of the sensor under actual or simulated conditions, taking into account various characteristics of the sensor, such as accuracy, response time, stability, anti-interference ability, etc. The evaluation results will determine whether the sensor meets the expected working standards. Deep performance evaluation uses statistical methods, regression analysis, error analysis, etc. to quantify performance, and simulates data to verify the reliability of the sensor under various environmental conditions. The simulated data is compared with the qualified parameter data to analyze the performance of the sensor. The evaluation is performed by calculating error values, response time, stability indicators, etc. For example, the accuracy of the sensor is evaluated by calculating the error range, and the response time is evaluated by simulating the response time under different environmental conditions. For example, error analysis is performed using sensor simulation data and target data to evaluate whether the sensor meets the accuracy and response speed requirements. If the error of the simulated data exceeds the target error range, the sensor needs to be adjusted or optimized.

[0081] Preferably, the present application also provides a sensor detection system for executing the sensor detection method as described above, the sensor detection system comprising: The sensor twin model building module is used to collect the geometric, mechanical and electrical parameters of the sensor installation environment in real time and build the sensor twin model; The sensor initial configuration module is used to automatically identify and initialize sensor parameters through the local area network according to the edge computing device to obtain the sensor initial configuration data; A sensor simulation module, used to generate sensor simulation data according to the sensor word twin model and the sensor initial configuration data; The performance detection and evaluation module is used to perform performance detection and evaluation based on the sensor simulation data to obtain sensor detection data.

[0082] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0083] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A sensor detection method, characterized in that: The following steps are involved: Step S1: Collect the geometric, mechanical and electrical parameters of the sensor installation environment in real time and build a digital twin model of the sensor; Step S2: Automatically identify and initialize sensor parameters through the local area network according to the edge computing device to obtain initial configuration data of the sensor; Step S3: Generate sensor simulation data according to the sensor digital twin model and the sensor initial configuration data; Step S4: Perform performance testing and evaluation based on the sensor simulation data to obtain sensor testing data.

2. The method according to claim 1, characterized in that Step S1 is specifically as follows: Step S11: deploy an environmental sensor array to collect real-time environmental parameter data through edge devices; Step S12: Generate a virtual sensor environment according to the real-time environmental parameter data to obtain virtual sensor environment data; Step S13: using laser scanning to obtain the geometric shape of the installation position, obtaining sensor geometric shape data, and performing geometric reconstruction based on the sensor geometric shape data to obtain sensor installation geometric parameter data; Step S14: calibrating the mechanical parameters of the installation environment using the pressure sensor and the accelerometer to obtain the mechanical characteristic data of the sensor; Step S15: using an oscilloscope to measure the connector impedance, voltage waveform and grounding state to obtain electrical characteristic data; Step S16: Establish a digital twin model of the sensor installation geometric parameter data, sensor mechanical characteristic data, electrical characteristic data and virtual sensor environment data to obtain a sensor digital twin model.

3. The method according to claim 2, characterized in that Step S12 is specifically as follows: Perform feature extraction based on real-time environmental parameter data to obtain environmental parameter feature data; Perform environmental geometry modeling according to environmental parameter characteristic data to obtain environmental geometry data; Perform physical attribute mapping on real-time environmental parameter data and environmental geometry data to obtain environmental physical attribute distribution map data; Generate virtual environment texture based on environment physical property distribution map data to obtain environment texture map data; A virtual sensor position interface is generated according to the environmental texture map data to obtain virtual sensor environmental data.

4. The method according to claim 3, characterized in that The feature extraction is as follows: Slice the window data according to the real-time environmental parameter data to obtain environmental parameter slice data; Calculate the environmental disturbance index according to the environmental parameter slice data to obtain environmental disturbance index data; Calculate the covariance matrix of multiple environmental variables based on the environmental parameter slice data to obtain the correlation data of multiple environmental variables; The environmental disturbance index data and multi-environmental variable correlation data are processed for abnormal environmental spatial fluctuations to obtain environmental parameter characteristic data.

5. The method according to claim 4, characterized in that The window data slices are as follows: Perform window slicing according to the real-time environment parameter data using preset first window slicing parameter data and preset second window slicing parameter data to obtain first preliminary window slicing data and second preliminary window slicing data respectively; Performing cluster calculation on the first preliminary window slice data and the second preliminary window slice data to obtain first window slice cluster data and second window slice cluster data respectively; Calculating the cluster change rate according to the first window slice clustering data and the second window slice clustering data to obtain first cluster cluster change rate data and second cluster cluster change rate data respectively; Dynamically re-dividing the first window slice parameter data according to the first cluster change rate data to obtain first window re-slicing parameter data, and dynamically re-dividing the first window slice parameter data according to the second cluster change rate data to obtain second window re-slicing parameter data; Perform window slicing according to the real-time environment parameter data using preset first window reslicing parameter data and preset second window reslicing parameter data, to obtain first preliminary window reslicing data and second preliminary window reslicing data respectively; Performing window data feature extraction on the first preliminary window reslicing data and the second preliminary window reslicing data to obtain first window feature data and second window feature data respectively; A cross-window feature matrix is ​​constructed using the first window feature data and the second window feature data, the first preliminary window re-slicing data, and the second preliminary window re-slicing data to obtain environmental parameter slice data.

6. The method according to claim 2, characterized in that The specific environment geometry modeling is as follows: Get the installation environment geometry data; Perform spatial mapping according to real-time environmental parameter data and installation environment geometric data to obtain installation environment mapping data; Perform geometric estimation of the installation environment according to the real-time environmental parameter data to obtain geometric estimation data of the installation environment; Environmental parameters are fused according to the installation environment geometry estimation data and the installation environment mapping data to obtain environmental geometry data.

7. The method according to claim 1, characterized in that Step S2 is specifically as follows: Step S21: Perform hardware connection verification according to the edge computing device to obtain hardware connection verification data; Step S22: performing communication channel configuration on the hardware connection verification data to obtain communication channel configuration data; Step S23: reading sensor characteristics according to the communication channel configuration data to obtain sensor characteristic data; Step S24: calibrating parameters according to the sensor characteristic data to obtain calibration parameter data; Step S25: performing gain adjustment on the parameter calibration data to obtain optimized gain value data; Step S26: Initialize parameters according to the optimized gain value data to obtain sensor initial configuration data.

8. The method according to claim 1, characterized in that Step S3 is specifically as follows: Step S31: performing sensor parameter fitting according to the sensor digital twin model and the sensor initial configuration data to obtain a sensor fitting model; Step S32: performing noise simulation on the sensor fitting model to obtain noise interference data; Step S33: simulating the usage scenario according to the sensor fitting model and the noise interference data to obtain sensor simulation data.

9. The method according to claim 1, characterized in that: Step S4 is specifically as follows: Step S41: obtaining sensor usage target data; Step S42: performing qualified parameter mapping according to the sensor usage target data to obtain sensor qualified parameter data; Step S43: Perform in-depth performance evaluation based on the sensor qualified parameter data and the sensor simulation data to obtain sensor detection data.

10. A sensor detection system, characterized in that: For executing the sensor detection method according to claim 1, the sensor detection system comprises: The sensor twin model building module is used to collect the geometric, mechanical and electrical parameters of the sensor installation environment in real time and build the sensor twin model; The sensor initial configuration module is used to automatically identify and initialize sensor parameters through the local area network according to the edge computing device to obtain the sensor initial configuration data; A sensor simulation module, used to generate sensor simulation data according to the sensor word twin model and the sensor initial configuration data; The performance detection and evaluation module is used to perform performance detection and evaluation based on the sensor simulation data to obtain sensor detection data.

Images