A gas sensor array data fusion method and system

Abstract

The present invention relates to the field of data fusion technology, and in particular to a gas sensor array data fusion method and system. The method comprises the following steps: acquiring and dividing the gas sensor array data of the monitoring area to obtain the gas sensor monitoring level data; acquiring the environmental sensor data of the monitoring area and constructing a multi-dimensional gas sensor network of the monitoring area; performing identification based on the multi-dimensional gas sensor network of the monitoring area to obtain the sensor response mode data; performing multi-sensor data fusion processing on the gas sensor array data of the monitoring area to obtain a gas concentration prediction model; performing gas concentration prediction on the monitoring area based on the gas concentration prediction model to obtain the original gas concentration prediction data of the monitoring area, and performing dynamic environmental adaptive correction on the original gas concentration prediction data of the monitoring area to obtain the gas concentration prediction data of the monitoring area. The present invention can improve the accuracy of gas concentration prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of data fusion, and in particular to a gas sensor array data fusion method and system. Background Art

[0002] As a multi-sensor system, the application scope of gas sensor array technology has gradually expanded to environmental monitoring, industrial safety, smart home, automobile emission control and other fields. Gas sensor arrays are usually composed of multiple different types of gas sensors, which can detect the concentration changes of different gases in the air in real time. However, due to the complex response characteristics of gas sensors and their susceptibility to interference from environmental factors, how to effectively extract effective information from the data obtained by multiple sensors and perform accurate analysis has become a key technical problem in the widespread application of gas sensor arrays. Traditional gas sensor array data processing methods usually rely on single sensor response data or simple signal processing algorithms. Most of these traditional methods use data interpretation directly based on the output signal of the gas sensor or gas concentration prediction based on simple mathematical models. Although these methods can provide certain effects in some applications, they have significant defects in practical applications. Traditional gas sensor array data fusion methods usually fail to fully consider the cross-interference problem between sensors. When different types of gas sensors detect the same gas, their responses differ due to factors such as sensor type, working environment, temperature and humidity changes. This response difference is often not effectively compensated and optimized in traditional methods, which can easily lead to inaccurate gas concentration predictions, and even false alarms and missed alarms. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide a gas sensor array data fusion method and system to solve at least one of the above technical problems.

[0004] To achieve the above object, a gas sensor array data fusion method comprises the following steps:

[0005] Step S1: Acquire gas sensor array data in the monitoring area, and perform spatial distribution analysis of gas sensors in the monitoring area based on the gas sensor array data in the monitoring area, thereby obtaining spatial distribution data of gas sensors in the monitoring area; divide gas sensor monitoring levels based on the spatial distribution data of gas sensors in the monitoring area, thereby obtaining gas sensor monitoring level data;

[0006] Step S2: Acquire environmental sensor data of the monitoring area, and divide the environmental sensor data of the monitoring area into sensor data grids, so as to obtain sensor grid data of the monitoring area; construct a multi-dimensional gas sensor network of the monitoring area according to the environmental sensor grid data of the monitoring area and the gas sensor monitoring level data;

[0007] Step S3: performing environmental noise denoising on the gas sensor signal based on the multi-dimensional gas sensor network in the monitoring area, thereby obtaining denoised gas sensor signal data; performing sensor response pattern recognition on the denoised gas sensor signal data, thereby obtaining sensor response pattern data;

[0008] Step S4: performing embedded feature optimization of sensor array output signals on the gas sensor array data in the monitoring area based on the sensor response pattern data, thereby obtaining a gas sensor feature vector; performing multi-sensor data fusion processing on the sensor data in the monitoring area based on the gas sensor feature vector, thereby obtaining a gas concentration prediction model;

[0009] Step S5: Predict the gas concentration in the monitoring area based on the gas concentration prediction model to obtain the original gas concentration prediction data in the monitoring area, and perform dynamic environmental adaptive correction on the original gas concentration prediction data in the monitoring area to obtain the gas concentration prediction data in the monitoring area.

[0010] Optionally, step S1 specifically includes:

[0011] Step S11: acquiring gas sensor array data in the monitoring area, and classifying the gas sensor array data in the monitoring area by sensor type, thereby obtaining MEMS micro-nano gas sensor array data and non-MEMS micro-nano gas sensor array data;

[0012] Step S12: extracting gas sensor spatial coordinates from the MEMS micro-nano gas sensor array data and the non-MEMS micro-nano gas sensor array data, respectively, so as to obtain MEMS micro-nano gas sensor spatial coordinate data and non-MEMS micro-nano gas sensor data;

[0013] Step S13: performing a gas sensor spatial distribution analysis based on the MEMS micro-nano gas sensor spatial coordinate data and the non-MEMS micro-nano gas sensor data, thereby obtaining the gas sensor spatial distribution data of the monitoring area;

[0014] Step S14: evaluating the monitoring capability of the sensor based on the spatial distribution data of the gas sensors in the monitoring area, thereby obtaining the monitoring capability data of the gas sensors;

[0015] Step S15: Divide the gas sensor monitoring level according to the gas sensor monitoring capability data and the spatial distribution data of the gas sensors in the monitoring area, so as to obtain the gas sensor monitoring level data.

[0016] Optionally, step S14 is specifically:

[0017] Step S141: classifying the spatial distribution data of gas sensors in the monitoring area by sensor type, thereby obtaining the spatial distribution data of MEMS micro-nano gas sensors and the spatial distribution data of non-MEMS micro-nano gas sensors;

[0018] Step S142: constructing a MEMS micro-nano gas sensor coverage model based on the MEMS micro-nano gas sensor spatial distribution data; constructing a non-MEMS micro-nano gas sensor coverage model based on the non-MEMS micro-nano gas sensor spatial distribution data;

[0019] Step S143: extracting gas sensing features from the gas sensor array data in the monitoring area, thereby obtaining gas sensing data in the monitoring area, and constructing a gas diffusion model in the monitoring area according to the gas sensing data in the monitoring area and the spatial distribution data of the gas sensors in the monitoring area;

[0020] Step S144: evaluating the gas monitoring capability of the MEMS micro-nano gas sensor according to the gas diffusion model of the monitoring area and the coverage model of the MEMS micro-nano gas sensor, thereby obtaining the gas monitoring capability data of the MEMS micro-nano gas sensor; evaluating the gas monitoring capability of the non-MEMS micro-nano gas sensor according to the gas diffusion model of the monitoring area and the coverage model of the non-MEMS micro-nano gas sensor, thereby obtaining the gas monitoring capability data of the non-MEMS micro-nano gas sensor;

[0021] Step S145: performing sensor cross-interference evaluation on the gas monitoring capability data of the MEMS micro-nano gas sensor and the gas monitoring capability data of the non-MEMS micro-nano gas sensor, thereby obtaining a regional gas sensor cross-interference factor;

[0022] Step S146: calibrate the actual monitoring capability of the gas sensor to the gas monitoring capability data of the MEMS micro-nano gas sensor and the gas monitoring capability data of the non-MEMS micro-nano gas sensor according to the regional gas sensor cross-interference factor, so as to obtain the gas sensor monitoring capability data.

[0023] Optionally, step S15 is specifically:

[0024] Step S151: performing sensor redundancy evaluation according to the gas sensor monitoring capability data and the spatial distribution data of the gas sensors in the monitoring area, thereby obtaining regional sensor deployment redundancy data;

[0025] Step S152: performing sensor measurement accuracy and sensitivity level classification based on the gas sensor monitoring capability data, thereby obtaining sensor monitoring capability level data;

[0026] Step S153: constructing a gas sensor monitoring level division model according to the gas sensor spatial distribution data of the monitoring area and the sensor monitoring capability level data, and performing monitoring level division based on the gas sensor monitoring level division model, thereby obtaining initial gas sensor monitoring level data;

[0027] Step S154: classifying the initial gas sensor monitoring layer data and the regional sensor deployment redundancy data into abnormal monitoring layer areas through the gas sensor monitoring layer division model, thereby obtaining layer overlapping area data and layer blank area data;

[0028] Step S155: Perform dynamic weighted optimization adjustment on the initial gas sensor monitoring layer data according to the layer overlapping area data and the layer blank area data, so as to obtain the gas sensor monitoring layer data.

[0029] Optionally, step S2 specifically includes:

[0030] Step S21: Acquire the environmental sensor data of the monitoring area, and perform environmental sensor spatial distribution statistics on the environmental sensor data of the monitoring area, so as to obtain environmental sensor spatial distribution data;

[0031] Step S22: performing sensor data grid division based on the spatial distribution data of the environmental sensors, thereby obtaining sensor grid data of the monitoring area;

[0032] Step S23: mapping the sensor grid sensor data of the monitoring area to the sensor grid data of the monitoring area according to the environmental sensor data of the monitoring area, thereby obtaining the environmental sensor data of the grid of the monitoring area;

[0033] Step S24: Calculate the gas concentration influencing factors of the monitoring area grid environmental sensor data and the monitoring area gas sensor data, thereby obtaining a gas concentration influencing factor set, and screen the monitoring area grid environmental sensor data according to the gas concentration influencing factor set, thereby obtaining the monitoring area grid main component environmental sensor data;

[0034] Step S25: Perform spatial multi-dimensional data fusion on the main component environmental sensing data of the monitoring area grid and the gas sensor monitoring level data, so as to obtain a multi-dimensional gas sensor network in the monitoring area.

[0035] Optionally, step S3 specifically includes:

[0036] Step S31: collecting real-time sensing data of the monitoring area based on the multi-dimensional gas sensing network of the monitoring area, thereby obtaining real-time gas sensing data of the monitoring area and real-time environmental sensing data of the monitoring area;

[0037] Step S32: performing gas sensor signal noise point identification on the real-time gas sensor data of the monitoring area, thereby obtaining gas sensor signal noise point data;

[0038] Step S33: performing sensor signal noise time series correlation according to the gas sensor signal noise data and the real-time environmental sensor data of the monitoring area, thereby obtaining environmental noise source data;

[0039] Step S34: selecting a gas concentration influencing factor from a gas concentration influencing factor set according to the environmental noise source data, thereby obtaining a noise source gas concentration influencing factor, and performing signal environmental noise denoising on the real-time gas sensor data in the monitoring area according to the noise source gas concentration influencing factor, thereby obtaining gas sensor denoised signal data;

[0040] Step S35: performing sensor response pattern recognition on the gas sensor denoised signal data to obtain sensor response pattern data.

[0041] Optionally, step S35 is specifically:

[0042] Step S351: extracting the signal time domain features of the gas sensor denoised signal data to obtain the gas sensor signal time domain data, and performing frequency domain conversion on the gas sensor signal time domain data to obtain the gas sensor signal spectrum;

[0043] Step S352: obtaining a gas property expert experimental database, and extracting gas property from the gas property expert experimental database, thereby obtaining gas property data;

[0044] Step S353: performing gas sensor response simulation according to the gas characteristic data and the gas sensor denoised signal data, thereby obtaining gas sensor simulated response data;

[0045] Step S354: performing structural integration of gas sensor response feature mapping based on the gas sensor simulated response data, thereby obtaining gas sensor simulated response pattern data;

[0046] Step S355: performing response pattern similarity calculation on the gas sensor simulated response pattern data and the gas sensor signal spectrum, thereby obtaining simulated response pattern similarity data;

[0047] Step S356: Perform sensor response pattern recognition based on the simulated response pattern similarity data to obtain sensor response pattern data.

[0048] Optionally, step S353 is specifically:

[0049] The gas sensor denoising signal data is classified by sensor type, thereby obtaining MEMS micro-nano gas sensor signal data and non-MEMS micro-nano gas sensor signal data;

[0050] Initialize the response simulation parameters according to the gas characteristic data and the MEMS micro-nano gas sensor signal data, so as to obtain the MEMS micro-nano type sensor initialization response simulation parameters;

[0051] Initializing response simulation parameters according to non-MEMS micro-nano gas sensor signal data and gas characteristic data, thereby obtaining initialization response simulation parameters of non-MEMS micro-nano type sensors;

[0052] Perform sensor physical response modeling on the MEMS micro-nano gas sensor signal data and the non-MEMS micro-nano gas sensor signal data, so as to obtain the MEMS micro-nano sensor response model and the non-MEMS micro-nano sensor response model;

[0053] Through the MEMS micro-nano sensor response model, and using the MEMS micro-nano type sensor initialization response simulation parameters, the gas sensor response simulation of the multi-dimensional gas sensor network in the monitoring area is performed to obtain the MEMS micro-nano type sensor response simulation data;

[0054] Through the non-MEMS micro-nano sensor response model and using the non-MEMS micro-nano sensor initialization response simulation parameters, the gas sensor response simulation of the multi-dimensional gas sensor network in the monitoring area is performed to obtain the non-MEMS micro-nano sensor response simulation data;

[0055] According to the gas sensor monitoring level data, the sensor response simulation coupling is performed on the MEMS micro-nano type sensor response simulation data and the non-MEMS micro-nano type sensor response simulation data, so as to obtain the gas sensor simulation response mode data.

[0056] Optionally, step S4 is specifically:

[0057] Step S41: extracting sensor output signal data from the gas sensor array data in the monitoring area to obtain sensor output signal data, and performing multi-dimensional signal matrix conversion on the sensor output signal data to obtain a sensor output signal matrix;

[0058] Step S42: performing response pattern feature matching on the sensor output signal matrix based on the sensor response pattern data, thereby obtaining output signal matching response pattern data;

[0059] Step S43: performing signal feature extraction on the output signal matching response pattern data to obtain output signal feature data, and performing response pattern embedded feature selection on the output signal feature data to obtain a gas sensor feature vector;

[0060] Step S44: uniformly encoding the gas sensor feature vectors to obtain the gas sensor feature coding vectors, and performing feature vector fusion on the gas sensor feature coding vectors to obtain a vector fusion feature matrix;

[0061] Step S45: construct a gas concentration prediction model based on the vector fusion feature matrix and the monitoring area sensor data.

[0062] The present invention solves the challenges encountered by gas sensor array technology in practical applications by systematically introducing a multidimensional gas sensor network and a data fusion algorithm, especially in terms of the accuracy of gas concentration prediction, sensor cross-interference and environmental noise suppression. By analyzing the spatial distribution of the gas sensor array data in the monitoring area, the sensors in the monitoring area can be efficiently laid out and spatial distribution data can be generated. This analysis can accurately determine the coverage and monitoring capability of each sensor, so that the deployment of the sensor not only has the comprehensiveness of the coverage area, but also avoids redundancy and insufficiency, thereby effectively optimizing the use efficiency of the sensor. By introducing the MEMS micro-nano gas sensor array and combining multidimensional data fusion, denoising, response pattern recognition and other processing methods, the accuracy, stability and adaptability of gas monitoring are significantly improved. The use of the MEMS micro-nano gas sensor array can provide accurate gas concentration data at multiple monitoring points, and optimize the sensor layout of the monitoring area through spatial distribution analysis, thereby improving the coverage capability of gas monitoring. The data provided by different types of gas sensors in the same area are effectively integrated. The MEMS sensor can provide efficient data collection in a complex environment through its miniaturization and low power consumption, while reducing the interference and errors that may occur in traditional sensors. By further dividing the monitoring level, multiple independent monitoring levels can be established in the region, and each level can be accurately controlled according to the response characteristics, working environment and spatial distribution of the sensor. This division helps to dynamically track and predict the changes in gas concentration in different areas in real time, and improve the response speed and accuracy of the monitoring system. After obtaining the environmental sensor data of the monitoring area and dividing the sensor grid, the present invention can further integrate the data of the environmental sensor and the gas sensor through the establishment of the sensor grid. The division of grid data enables environmental influencing factors (such as temperature and humidity, etc.) to be effectively associated with gas concentration changes, thereby enhancing the accuracy of the gas concentration prediction model. The construction of a multidimensional gas sensor network not only provides spatial structured analysis of gas sensor data, but also can fuse sensor data from multiple dimensions. Through multi-level algorithm processing, the detection of different gas components is more accurate and comprehensive. The data quality is significantly improved by signal denoising processing and identification of sensor response patterns. The identification and noise removal of environmental noise sources can effectively filter out unnecessary interference information, reduce the noise components in the sensor response signal, and ensure the accuracy of the sensor output signal. The recognition of sensor response patterns can accurately match the output characteristics of each sensor according to its characteristics and response rules, thereby generating a feature vector representing the change in gas concentration. The multi-dimensional fusion processing of gas sensor data not only improves the validity of the data, but also avoids the problem of false alarms or missed alarms caused by cross-interference of sensors.Through in-depth analysis and optimization of sensor response patterns, especially the unique response characteristics of MEMS micro-nano sensors, it is further ensured that the output signal of each sensor can be accurately extracted and optimized, and the accuracy of gas concentration prediction is improved. At the same time, the signal denoising technology is used to effectively reduce the influence of environmental noise, ensure the stability of the sensor, and enable it to provide reliable data support in various complex environments. This vectorized feature data can not only reflect the changing trend of gas concentration, but also provide accurate input data for subsequent data fusion and prediction models. Based on the embedded feature optimization of gas sensor output signals, the deep extraction and effective fusion of gas sensor signals are realized. By optimizing the output signal matrix of different gas sensor arrays, the detection ability of sensors in complex environments can be improved, and the redundant information between multi-sensor data can be reduced. This processing process fully considers the diversity and complexity of gas sensors, ensures the effective coordination of different types of sensors, and significantly improves the effect of multi-sensor data fusion processing. MEMS micro-nano gas sensors play a particularly prominent role in sensor response pattern recognition. They can not only provide more refined response data, but also significantly enhance the reliability of gas concentration prediction models through embedded feature optimization and multi-sensor data fusion. Through the conversion of multi-dimensional signal matrix and the deep extraction of signal features, the high sensitivity and high precision of MEMS sensors make the final gas concentration prediction more timely and accurate. In addition, the deployment of MEMS micro-nano sensors not only improves the redundancy and monitoring capabilities of the sensor network, but also effectively reduces errors through sensor cross-interference evaluation, further optimizing the overall performance of the gas monitoring system. The constructed gas concentration prediction model can not only perform efficient gas concentration prediction based on multi-dimensional sensor data, but also adaptively adjust and optimize model parameters so that it can respond to changes in environmental conditions and differences in sensor response in real time. Through dynamic environmental adaptive correction technology, the drift effect of the sensor itself and the influence of changes in the external environment can be eliminated, thereby ensuring the stability and reliability of the prediction results. Based on this high-precision prediction model, the system can accurately evaluate the changes in gas concentration and provide real-time feedback on the environmental status of the monitoring area, greatly improving the real-time response and early warning capabilities of the gas monitoring system. Through the synergistic effect of the above-mentioned technical steps, the present invention can significantly improve the monitoring accuracy, data quality and system adaptability of the gas sensor array, solve a series of problems existing in the traditional gas sensor array system, such as multi-sensor data fusion, signal noise interference, sensor response difference, etc., and finally realize more efficient and accurate gas concentration monitoring and prediction functions.

[0063] Optionally, the present specification also provides a gas sensor array data fusion system for executing the gas sensor array data fusion method as described above, the gas sensor array data fusion system comprising:

[0064] The gas sensor monitoring level division module is used to obtain the gas sensor array data of the monitoring area, and perform spatial distribution analysis of the gas sensors in the monitoring area based on the gas sensor array data of the monitoring area, so as to obtain the spatial distribution data of the gas sensors in the monitoring area; and divide the gas sensor monitoring level based on the spatial distribution data of the gas sensors in the monitoring area, so as to obtain the gas sensor monitoring level data;

[0065] A multi-dimensional gas sensor network construction module is used to obtain the environmental sensor data of the monitoring area and divide the environmental sensor data of the monitoring area into sensor data grids, thereby obtaining the sensor grid data of the monitoring area; a multi-dimensional gas sensor network of the monitoring area is constructed according to the environmental sensor grid data of the monitoring area and the gas sensor monitoring level data;

[0066] The sensor response pattern recognition module is used to perform environmental noise denoising of the gas sensor signal based on the multi-dimensional gas sensor network in the monitoring area, thereby obtaining gas sensor denoised signal data; perform sensor response pattern recognition on the gas sensor denoised signal data, thereby obtaining sensor response pattern data;

[0067] The sensor data fusion module is used to optimize the embedded features of the sensor array output signal of the gas sensor array data in the monitoring area based on the sensor response mode data, so as to obtain the gas sensor feature vector; based on the gas sensor feature vector, multi-sensor data fusion processing is performed on the sensor data in the monitoring area to obtain the gas concentration prediction model;

[0068] The environmental adaptive correction module is used to predict the gas concentration in the monitoring area based on the gas concentration prediction model, so as to obtain the original gas concentration prediction data in the monitoring area, and to perform dynamic environmental adaptive correction on the original gas concentration prediction data in the monitoring area, so as to obtain the gas concentration prediction data in the monitoring area.

[0069] The gas sensor array data fusion system of the present invention can implement any one of the gas sensor array data fusion methods of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the gas sensor array data fusion method. The internal modules of the system cooperate with each other to improve the accuracy of gas concentration prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0071] Figure 1 It is a schematic diagram of the steps of the gas sensor array data fusion method of the present invention;

[0072] Figure 2 Detailed step flow diagram of step S1 in the present invention;

[0073] Figure 3 Detailed step flow diagram of step S2 in the present invention;

[0074] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0075] The technical method of the present invention is described clearly and completely below 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.

[0076] 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.

[0077] 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.

[0078] To achieve this, please refer to Figures 1 to 3 The present invention provides a gas sensor array data fusion method, the method comprising the following steps:

[0079] Step S1: acquiring gas sensor array data of the monitoring area, and performing spatial distribution analysis of the gas sensors in the monitoring area according to the gas sensor array data of the monitoring area, thereby obtaining spatial distribution data of the gas sensors in the monitoring area; dividing the gas sensor monitoring level based on the spatial distribution data of the gas sensors in the monitoring area, thereby obtaining gas sensor monitoring level data;

[0080] In this embodiment, the gas sensor array data in the monitoring area is obtained through the monitoring area sensor management platform, and the spatial distribution analysis of these data is first performed to obtain the spatial distribution data of the gas sensors in the monitoring area. This process usually uses a spatial statistical method based on sensor data, such as the K-means clustering algorithm or the spatial interpolation algorithm (such as the IDW interpolation method), to divide the monitoring area into multiple spatial areas, and analyze the relationship between the position of the sensor and the measurement results. For example, if the monitoring area is a factory workshop, and the gas sensors are installed at different heights and different process areas, the optimized sensor distribution can be obtained by analyzing the spatial layout data of these sensors. On this basis, by establishing a sensor coverage model, the monitoring level of the sensor is divided, such as high altitude, low altitude, underground and other levels, so as to accurately monitor the gas concentration in different areas.

[0081] Step S2: Acquire environmental sensor data of the monitoring area, and divide the environmental sensor data of the monitoring area into sensor data grids, so as to obtain sensor grid data of the monitoring area; construct a multi-dimensional gas sensor network of the monitoring area according to the environmental sensor grid data of the monitoring area and the gas sensor monitoring level data;

[0082] In this embodiment, environmental sensor data of the monitoring area, such as temperature, humidity, air pressure and other environmental parameters, are obtained through the monitoring area sensor management platform. Then, these data are gridded. Commonly used techniques include subdividing the monitoring area into uniform or non-uniform grids, such as using a grid algorithm to allocate grid areas of environmental sensors. Taking an indoor temperature control system as an example, the environmental sensor will record the temperature and humidity data of different areas according to the different subdivided areas of the area division. Then, a multi-dimensional gas sensor network is constructed through the grid data of these environmental sensors and the monitoring level data of the gas sensor, that is, at different spatial levels, the data of the environmental sensor and the gas sensor are collaboratively processed to achieve multi-dimensional information fusion, thereby improving the accuracy of gas concentration prediction.

[0083] Step S3: performing environmental noise denoising on the gas sensor signal based on the multi-dimensional gas sensor network in the monitoring area, thereby obtaining denoised gas sensor signal data; performing sensor response pattern recognition on the denoised gas sensor signal data, thereby obtaining sensor response pattern data;

[0084] In this embodiment, the environmental noise of the gas sensor signal is denoised based on the multi-dimensional gas sensor network data in the monitoring area. The methods used, such as wavelet transform denoising technology, or denoising methods such as Kalman filter, can effectively remove the environmental noise in the gas sensor signal. For example, when the gas sensor signal is affected by factors such as electromagnetic interference or temperature change, the denoising process helps to extract a pure gas concentration signal. Then, a sensor response pattern recognition method based on machine learning (such as support vector machine SVM or deep learning model) is used to classify and identify the denoised signal in order to identify the response patterns of various gases, thereby providing clear signal characteristics for subsequent data processing. For example, when monitoring the exhaust gas emissions of a chemical plant, the sensor will identify its response pattern based on different gas characteristics (such as ammonia, carbon dioxide, etc.).

[0085] Step S4: performing embedded feature optimization of sensor array output signals on the gas sensor array data in the monitoring area based on the sensor response pattern data, thereby obtaining a gas sensor feature vector; performing multi-sensor data fusion processing on the sensor data in the monitoring area based on the gas sensor feature vector, thereby obtaining a gas concentration prediction model;

[0086] In this embodiment, embedded feature optimization is performed on the output signal of the gas sensor array in the monitoring area based on the data of the sensor response pattern. This process includes using machine learning methods to extract and optimize the features of the sensor response pattern data to improve the characterization ability of the sensor data. For example, principal component analysis (PCA) is used to reduce the dimension of the sensor signal data, remove redundant information, and extract key features. Then, based on the optimized gas sensor feature vector, multi-sensor data fusion is performed. This fusion process can use multi-sensor fusion methods such as weighted averaging and Kalman filtering to make full use of the advantages of different sensor data to form a unified gas concentration prediction model. For example, if multiple sensors monitor nitrogen oxides and carbon dioxide respectively, more accurate gas concentration predictions can be obtained through weighted fusion.

[0087] Step S5: Predict the gas concentration in the monitoring area based on the gas concentration prediction model to obtain the original gas concentration prediction data in the monitoring area, and perform dynamic environmental adaptive correction on the original gas concentration prediction data in the monitoring area to obtain the gas concentration prediction data in the monitoring area.

[0088] In this embodiment, the gas concentration in the monitoring area is predicted based on the constructed gas concentration prediction model. Taking environmental protection monitoring as an example, the model can predict the gas concentration in a certain period of time through real-time sensor data, and dynamically adjust and correct it according to known data. For example, a certain monitoring area is affected by changes in temperature and humidity, resulting in deviations in gas concentration prediction. In this process, an adaptive correction method, such as adaptive filtering or Bayesian optimization technology based on a feedback mechanism, can be used to correct the gas concentration prediction. Specifically, when there is a large deviation between the data fed back by the sensor and the predicted value, the correction algorithm can correct the prediction result by adjusting the model parameters to obtain more accurate gas concentration prediction data. For example, when the temperature or humidity changes significantly, these environmental factors will affect the diffusion rate and concentration distribution of the gas, so the prediction result is adjusted through a real-time feedback mechanism. Adaptive filtering (such as Kalman filtering) or Bayesian optimization methods can be used to dynamically update the parameters of the prediction model according to real-time changing data, thereby achieving accurate prediction of gas concentration.

[0089] Optionally, step S1 specifically includes:

[0090] Step S11: acquiring gas sensor array data in the monitoring area, and classifying the gas sensor array data in the monitoring area by sensor type, thereby obtaining MEMS micro-nano gas sensor array data and non-MEMS micro-nano gas sensor array data;

[0091] In this embodiment, the gas sensor array data including the sensor type, installation location and working status data in the monitoring area is obtained through the monitoring area sensor management platform. Sensor type classification can be performed by identifying the model of the sensor or by the working principle of the sensor. For example, in some industrial environments, MEMS (micro-electromechanical system) gas sensors and non-MEMS gas sensors (such as semiconductor gas sensors, electrochemical gas sensors, etc.) have different working mechanisms and response characteristics. Therefore, firstly, the sensor data is classified by type, and the MEMS micro-nano gas sensor array data and the non-MEMS micro-nano gas sensor array data are extracted separately. This can help the subsequent spatial coordinate extraction and distribution analysis, and make targeted processing and optimization of the data of different types of sensors.

[0092] Step S12: extracting gas sensor spatial coordinates from the MEMS micro-nano gas sensor array data and the non-MEMS micro-nano gas sensor array data, respectively, so as to obtain MEMS micro-nano gas sensor spatial coordinate data and non-MEMS micro-nano gas sensor data;

[0093] In this embodiment, the spatial coordinates of the gas sensors are extracted for the MEMS micro-nano gas sensor array data and the non-MEMS micro-nano gas sensor array data. In order to accurately obtain the position data of the sensor, a high-precision positioning system, such as differential GPS (DGPS) or laser radar (LiDAR), is used to perform three-dimensional spatial positioning of the sensor. In a specific application, assuming that the monitoring area is a chemical plant, the sensors are installed on different equipment and pipelines, and the spatial coordinate data needs to be extracted according to the actual installation location of each sensor. By extracting the spatial coordinate data, accurate location data can be provided for the subsequent spatial distribution analysis of gas sensors and the division of monitoring levels, laying the foundation for the optimization of the overall monitoring system.

[0094] Step S13: performing a gas sensor spatial distribution analysis based on the MEMS micro-nano gas sensor spatial coordinate data and the non-MEMS micro-nano gas sensor data, thereby obtaining the gas sensor spatial distribution data of the monitoring area;

[0095] In this embodiment, the spatial distribution analysis of gas sensors is performed based on the spatial coordinate data and sensor type data extracted from the MEMS and non-MEMS gas sensor arrays. The analysis method can adopt spatial statistical methods, such as Voronoi diagrams, K-means clustering, etc. By performing cluster analysis on the distribution of sensors, it is possible to identify which areas have dense sensors and which areas have scarce sensors. For example, in an underground parking lot in a city, there may be an uneven layout of sensors. Through this step, the sensor positions can be rationalized, the layout of sensors can be optimized, and effective gas monitoring can be ensured in each area. The results of the spatial distribution analysis will provide data support for subsequent sensor capability evaluation and monitoring level division, ensuring that the sensors can cover the entire monitoring area and reduce blind spots.

[0096] Step S14: evaluating the monitoring capability of the sensor based on the spatial distribution data of the gas sensor in the monitoring area, thereby obtaining the monitoring capability data of the gas sensor;

[0097] In this embodiment, the sensor monitoring capability is evaluated based on the spatial distribution data of the gas sensor. The evaluation method may adopt techniques such as sensitivity analysis and coverage analysis. For example, by evaluating the sensitivity and response time of each sensor, its reliability and accuracy under different environmental conditions can be determined. Assuming that in a high temperature and high humidity environment, some non-MEMS gas sensors may be affected by performance degradation, the evaluation process can help identify these potential problems and ensure the accuracy of the monitoring data. In addition, the coverage of the sensor is evaluated, and the effective coverage area of ​​the sensor can be obtained by calculating the monitoring radius of each sensor and its overlap with other sensors. Ultimately, these evaluation data will help optimize the configuration of the gas sensor to ensure that the gas concentration in the entire monitoring area can be fully and accurately monitored.

[0098] Step S15: Divide the gas sensor monitoring level according to the gas sensor monitoring capability data and the spatial distribution data of the gas sensors in the monitoring area, so as to obtain the gas sensor monitoring level data.

[0099] In this embodiment, the gas sensor monitoring level is divided based on the monitoring capability data of the gas sensor and the spatial distribution data of the gas sensor. Monitoring level division refers to dividing the monitoring area into multiple different monitoring levels based on factors such as the spatial distribution of the gas sensor, monitoring capabilities, and environmental conditions. Common division methods include a layered method based on sensor coverage, or an area division method based on air flow patterns. For example, in a monitoring system for a multi-story building, each floor can be used as a monitoring level, and the sensors on each floor can be divided into different small areas for more detailed monitoring according to the spatial distribution and capability evaluation results. The key to this step is to maximize the monitoring capability of each sensor by reasonably dividing the monitoring levels, ensuring that accurate monitoring data can be obtained under different gas concentration conditions, while avoiding overlapping or insufficient coverage of monitoring in certain areas, thereby improving the overall efficiency and accuracy of the system.

[0100] Optionally, step S14 is specifically:

[0101] Step S141: classifying the spatial distribution data of gas sensors in the monitoring area by sensor type, thereby obtaining the spatial distribution data of MEMS micro-nano gas sensors and the spatial distribution data of non-MEMS micro-nano gas sensors;

[0102] In this embodiment, the spatial distribution data of gas sensors in the monitoring area are divided into sensor types. The specific method is to identify the spatial distribution of MEMS micro-nano gas sensors and non-MEMS micro-nano gas sensors based on the working principle, model or other identification characteristics of the gas sensor. For example, MEMS sensors are usually small and installed in a compact location, while non-MEMS sensors are distributed in a wider area. The specific method of sensor type division can be identified by sensor model or communication protocol, and then the corresponding spatial distribution data is classified to obtain the spatial distribution data of MEMS sensors and the spatial distribution data of non-MEMS sensors. This step can provide data support for the subsequent gas monitoring capability assessment and optimize the sensor layout in the system design stage.

[0103] Step S142: constructing a MEMS micro-nano gas sensor coverage model based on the MEMS micro-nano gas sensor spatial distribution data; constructing a non-MEMS micro-nano gas sensor coverage model based on the non-MEMS micro-nano gas sensor spatial distribution data;

[0104] In this embodiment, based on the spatial distribution data of MEMS micro-nano gas sensors and non-MEMS micro-nano gas sensors, corresponding gas sensor coverage models are constructed respectively. Specifically, MEMS sensors usually have high accuracy and small response range, so the coverage model of MEMS sensors can be constructed based on the data of the spatial distribution of sensors and combined with their detection radius, and the coverage range and overlapping area of ​​the sensors in the monitoring area can be calculated. At the same time, non-MEMS sensors usually have a wide response range. The construction of the coverage model takes into account the characteristics of the sensor such as sensitivity and response time. The coverage model of non-MEMS sensors is constructed by analyzing the spatial distribution data. For MEMS sensors, considering that they have a fast response speed and are suitable for monitoring local areas, their effective coverage range in the monitoring area can be calculated based on their spatial distribution data. For example, assuming that the detection radius of the MEMS sensor is 2 meters, the specific coverage area of ​​each sensor in the monitoring area will be obtained by calculation. For non-MEMS sensors, due to their large detection range, their coverage model can be constructed by a similar method. This model can help determine the monitoring capability and coverage effect of the sensor under different environmental conditions, thereby optimizing the sensor layout of the monitoring system.

[0105] Step S143: extracting gas sensing features from the gas sensor array data in the monitoring area, thereby obtaining gas sensing data in the monitoring area, and constructing a gas diffusion model in the monitoring area according to the gas sensing data in the monitoring area and the spatial distribution data of the gas sensors in the monitoring area;

[0106] In this embodiment, gas sensing feature extraction is performed on the gas sensor array data of the monitoring area. By analyzing the data collected by the sensor, including the response time, signal strength, working status of the sensor, etc., the gas sensing data of the monitoring area is extracted. These data can reflect the changing trend and distribution of gas concentration. Then, combined with the spatial distribution data of the gas sensor, the diffusion behavior of the gas in the monitoring area is simulated by constructing a gas diffusion model. For example, in an indoor factory, the gas will diffuse with the flow of air and the operation of the equipment. Key features such as gas concentration, signal strength, and sensor response time are extracted from the real-time data obtained by the sensor. Subsequently, a gas diffusion model is constructed in combination with the spatial distribution data of the monitoring area. The model takes into account the diffusion characteristics of the gas, such as gas fluidity, ambient temperature, humidity, and other factors, and simulates the concentration change of the gas in the space. For example, in a closed room, the gas diffuses from a certain leakage point. The changing trend of the gas concentration is predicted according to the gas diffusion model, and data support is provided for the subsequent sensor response capability evaluation.

[0107] Step S144: evaluating the gas monitoring capability of the MEMS micro-nano gas sensor according to the gas diffusion model of the monitoring area and the coverage model of the MEMS micro-nano gas sensor, thereby obtaining the gas monitoring capability data of the MEMS micro-nano gas sensor; evaluating the gas monitoring capability of the non-MEMS micro-nano gas sensor according to the gas diffusion model of the monitoring area and the coverage model of the non-MEMS micro-nano gas sensor, thereby obtaining the gas monitoring capability data of the non-MEMS micro-nano gas sensor;

[0108] In this embodiment, the gas monitoring capabilities of MEMS micro-nano gas sensors and non-MEMS micro-nano gas sensors are evaluated by combining the gas diffusion model and the sensor coverage model. First, based on the gas diffusion model, the diffusion path and concentration change of the gas are analyzed, and the monitoring capability of the MEMS sensor in different gas concentration areas is calculated in combination with the MEMS micro-nano gas sensor coverage model. Secondly, according to the coverage model and gas diffusion model of the non-MEMS sensor, a similar evaluation is performed on the monitoring capability of the non-MEMS sensor. For example, in an industrial plant, the gas diffusion model shows that the gas concentration in some areas is high, and the MEMS sensor can effectively monitor the concentration changes in these areas due to its high accuracy and local coverage ability, while the non-MEMS sensor can cover a wider area but with lower accuracy. By evaluating the monitoring capability of each type of sensor, the comprehensiveness and accuracy of the monitoring system can be ensured. According to its coverage model and gas diffusion model, its monitoring capability is analyzed, and the gas concentration range that the sensor can effectively monitor is calculated. For example, it is assumed that the MEMS sensor can respond quickly in the high concentration gas area, and its response in the low concentration area is slow. Comprehensively evaluate the effective monitoring area of ​​the MEMS sensor. As for non-MEMS sensors, although they have a larger coverage area, they may have slower response time and lower accuracy. Their monitoring capabilities will be evaluated based on their coverage model and gas diffusion model. This evaluation result can help design a more reasonable gas monitoring network so that each sensor can play the best role in the appropriate environment.

[0109] Step S145: performing sensor cross-interference evaluation on the gas monitoring capability data of the MEMS micro-nano gas sensor and the gas monitoring capability data of the non-MEMS micro-nano gas sensor, thereby obtaining a regional gas sensor cross-interference factor;

[0110] In this embodiment, the cross-interference factor of the regional gas sensor is calculated by performing a cross-interference evaluation on the gas monitoring capability data of the MEMS micro-nano gas sensor and the non-MEMS micro-nano gas sensor. The key to the cross-interference evaluation is to consider the mutual influence between different types of sensors. For example, two adjacent sensors may interfere with each other due to the difference in gas diffusion or reaction speed, thereby affecting the accuracy of the monitoring data. By establishing a cross-interference evaluation model, the degree of interference between sensors can be identified and quantified. For example, if two MEMS sensors are installed too close, the sensor data will be inaccurate due to the mutual influence of the signals. For example, a MEMS sensor and a non-MEMS sensor may be installed too close, causing signal interference when detecting a certain gas. In order to quantitatively evaluate this interference, the cross-interference factor will be calculated based on the spatial position, detection range and gas diffusion model of the sensor. These factors will describe the interference intensity of the two types of sensors in the actual monitoring process, help identify potential interference problems, and provide data support for the next step of correction.

[0111] Step S146: calibrate the actual monitoring capability of the gas sensor to the gas monitoring capability data of the MEMS micro-nano gas sensor and the gas monitoring capability data of the non-MEMS micro-nano gas sensor according to the regional gas sensor cross-interference factor, so as to obtain the gas sensor monitoring capability data.

[0112] In this embodiment, the gas monitoring capability data of MEMS and non-MEMS micro-nano gas sensors are corrected for actual monitoring capability according to the cross-interference factor of the regional gas sensor. Through correction, errors caused by factors such as interference between sensors and environmental changes can be reduced. For example, in some cases, the actual response capability of the sensor may change due to environmental factors (such as humidity, temperature changes, etc.), or the monitoring data may be affected by cross-interference between sensors. By analyzing the cross-interference factor, the sensor data is adjusted to be closer to the actual gas concentration. This correction can be performed through a machine learning model, for example, a regression model is trained through historical data to predict and correct the sensor data. The monitoring data of the sensor can be corrected by weighted averaging, regression model, etc. to reduce errors caused by interference. For example, assuming that there is a strong gas interference source in a certain area, the monitoring data of the MEMS sensor is affected, and the data will be adjusted by known interference factors to make it closer to the actual gas concentration. This process will effectively improve the accuracy and reliability of the gas monitoring system and ensure that the data still accurately reflects the changes in gas concentration in a changing environment.

[0113] Optionally, step S15 is specifically:

[0114] Step S151: performing sensor redundancy evaluation according to the gas sensor monitoring capability data and the spatial distribution data of the gas sensors in the monitoring area, thereby obtaining regional sensor deployment redundancy data;

[0115] In this embodiment, the redundancy of the sensor is evaluated based on the monitoring capability data of the gas sensor and the spatial distribution data of the gas sensors in the monitoring area. The redundancy evaluation aims to ensure that there are enough sensors in each monitoring area to cope with environmental changes or equipment failures. In specific implementation, the coverage of sensors in different areas is calculated taking into account the monitoring capability range (such as detection accuracy, response speed, etc.) and spatial distribution of each sensor. For example, if the sensor density in a certain area is too high, the redundancy value will be higher; conversely, if the sensors in a certain area are relatively sparse, the redundancy value will be lower. The redundancy evaluation results can help determine which areas need to add sensors or rearrange existing sensors to ensure the stability and accuracy of the entire monitoring system.

[0116] Step S152: performing sensor measurement accuracy and sensitivity level classification based on the gas sensor monitoring capability data, thereby obtaining sensor monitoring capability level data;

[0117] In this embodiment, the level division of measurement accuracy and sensitivity is performed based on the monitoring capability data of the sensor (such as the response time, sensitivity, accuracy, etc. of the sensor). The performance analysis is performed on each gas sensor, and the sensor is divided into multiple levels according to its performance parameters. For example, the measurement accuracy of the sensor can be divided into high accuracy, medium accuracy and low accuracy; the sensitivity can be divided into high sensitivity, medium sensitivity and low sensitivity levels according to the sensor's response sensitivity to changes in gas concentration. After the division is completed, the sensor's monitoring capability level data will be generated as the basis for subsequent monitoring level division and optimization. This level division helps to identify which areas require high-sensitivity sensors and which areas can use low-sensitivity sensors, thereby reasonably arranging the use of sensors.

[0118] Step S153: constructing a gas sensor monitoring level division model according to the gas sensor spatial distribution data of the monitoring area and the sensor monitoring capability level data, and performing monitoring level division based on the gas sensor monitoring level division model, thereby obtaining initial gas sensor monitoring level data;

[0119] In this embodiment, a monitoring level division model for gas sensors is constructed based on the spatial distribution data of gas sensors and the sensor monitoring capability level data. By analyzing the spatial layout of the monitoring area and combining the performance level of the sensor, the entire monitoring area is divided into multiple monitoring levels. Specifically, some areas require more high-sensitivity sensors to monitor gas changes in real time, while other areas can use lower-sensitivity sensors for data collection. For example, in high-risk areas of industrial plants, it is divided into a high-sensitivity monitoring level to ensure timely detection of gas leaks; in non-high-risk areas, it is divided into a low-sensitivity monitoring level.

[0120] Step S154: classifying the initial gas sensor monitoring layer data and the regional sensor deployment redundancy data into abnormal monitoring layer areas through the gas sensor monitoring layer division model, thereby obtaining layer overlapping area data and layer blank area data;

[0121] In this embodiment, the gas sensor monitoring level division model is used to classify abnormal areas of the initially divided gas sensor monitoring level data, mainly into level overlapping areas and level blank areas. Overlapping areas refer to places where sensor redundancy is too high and sensors are deployed too densely; blank areas refer to places where sensor coverage is insufficient, which may lead to gas monitoring blind spots. These areas are identified by analyzing the spatial distribution of gas sensors and the data at the monitoring level. For example, in some monitoring levels, it is found that too many high-sensitivity sensors are concentrated in a small area, causing redundancy; in other areas, due to uneven sensor layout, monitoring blank areas are caused. Based on these evaluation results, suggestions for adjusting the sensor layout are provided.

[0122] Step S155: Perform dynamic weighted optimization adjustment on the initial gas sensor monitoring layer data according to the layer overlapping area data and the layer blank area data, so as to obtain the gas sensor monitoring layer data.

[0123] In this embodiment, the initial gas sensor monitoring layer data is dynamically weighted and optimized, and the optimization process is weighted based on the data of the overlapping layer area and the blank layer area. First, the sensor weights of these areas are reduced according to the overlapping area data to avoid excessive redundancy; for the blank area, the weights of the sensors in the corresponding area are increased to ensure the coverage of the sensors in these areas. The weight of the sensor refers to the relative importance or priority of each sensor or sensor area in the system decision-making process in the gas sensor monitoring system. Specifically, the weight is a numerical value that represents the proportion of a certain sensor or area in the monitoring layer adjustment and optimization process. Specifically, if the sensor redundancy in a certain area is too high, the priority of the sensor in the area will be automatically reduced, and its weight will be allocated to other monitoring blank areas to ensure the reasonable allocation of resources. For example, in an industrial plant, if there are too many sensors in a certain area, resulting in waste of resources, the sensor deployment in the area will be optimized to ensure that the coverage and monitoring capabilities of the monitoring system are balanced. The optimized monitoring layer data will be used to adjust the layout of the sensors to ensure that the sensor layout in the entire monitoring area is more reasonable and the monitoring effect is more accurate.

[0124] Optionally, step S2 specifically includes:

[0125] Step S21: Acquire the monitoring area environmental sensor data, and perform environmental sensor spatial distribution statistics on the monitoring area environmental sensor data, thereby obtaining environmental sensor spatial distribution data;

[0126] In this embodiment, the sensing data of the environmental sensors are collected from each environmental monitoring point in the sensor management platform of the monitoring area, including environmental parameters such as temperature, humidity, and atmospheric pressure. Then, the monitoring area is spatially analyzed using geographic information system (GIS) technology, and the geographical location of each sensor and its monitoring data are recorded in the database. By statistically analyzing the spatial distribution of the sensor locations, the distribution density, coverage, and other data of the sensors in each area can be obtained, providing a basis for subsequent spatial optimization. For example, in a large factory area, environmental sensors are distributed in multiple workshops, offices, and external areas. By statistically analyzing their spatial distribution data, it is possible to identify which areas have densely distributed sensors and which areas may have insufficient sensor coverage.

[0127] Step S22: performing sensor data grid division based on the spatial distribution data of the environmental sensors, thereby obtaining sensor grid data of the monitoring area;

[0128] In this embodiment, after obtaining the spatial distribution data of the environmental sensors, the monitoring area can be divided into multiple grid units according to the density and distribution characteristics of the sensors. The size of each grid unit can be dynamically adjusted according to the distribution of the environmental sensors. For example, in urban environmental monitoring, the size of the grid can be adjusted according to factors such as the building density and traffic flow in the urban area. For example, the center of the city is divided into smaller grids to ensure accurate monitoring of high-density areas; while suburbs or open areas can be divided into larger grids. In this process, spatial analysis software can be used for automated gridding to ensure that each grid has a sufficient number of sensor data support.

[0129] Step S23: mapping the sensor grid sensor data of the monitoring area to the sensor grid data of the monitoring area according to the environmental sensor data of the monitoring area, thereby obtaining the environmental sensor data of the grid of the monitoring area;

[0130] In this embodiment, the collected environmental sensor data is mapped to the grid unit, and the environmental data in each grid unit will be interpolated based on the sensor data. For example, if there are multiple environmental sensor data points in a grid unit, the measurement results of these sensors can be fused by weighted average or Kriging interpolation method to generate comprehensive environmental data for the grid unit. In this way, environmental data such as temperature, humidity, and air pressure can be obtained for each grid unit. For example, when monitoring an area of ​​atmospheric pollution, these environmental sensor data are needed to evaluate the diffusion trend of gas pollution, and the gridded data can provide fine-grained environmental background data for subsequent gas concentration models.

[0131] Step S24: Calculate the gas concentration influencing factors of the monitoring area grid environmental sensor data and the monitoring area gas sensor data, thereby obtaining a gas concentration influencing factor set, and screen the monitoring area grid environmental sensor data according to the gas concentration influencing factor set, thereby obtaining the monitoring area grid main component environmental sensor data;

[0132] In this embodiment, a mathematical model (such as multivariate linear regression or machine learning algorithm) is established to calculate the influencing factors of environmental sensor data on gas concentration. These factors include environmental factors such as temperature, humidity, air pressure, wind speed, etc., which will affect the diffusion, adsorption or chemical reaction of gas concentration. Then, the environmental factors that have a greater impact on gas concentration are screened out by statistical methods to obtain a set of main environmental factors. These main factors are extracted through methods such as principal component analysis (PCA) and used in subsequent gas concentration prediction models. For example, if humidity and temperature are the main factors affecting gas concentration in certain areas, then the environmental sensor data in these areas will be given a higher weight to optimize the calculation results of gas concentration.

[0133] Step S25: Perform spatial multi-dimensional data fusion on the main component environmental sensing data of the monitoring area grid and the gas sensor monitoring level data, so as to obtain a multi-dimensional gas sensor network in the monitoring area.

[0134] In this embodiment, based on the previously obtained grid principal component environmental sensor data and the monitoring level data of the gas sensor, the environmental data and the gas sensor data are combined through multidimensional data fusion technology (such as Kalman filtering, Bayesian estimation, etc.) to establish a multidimensional gas sensor network. For example, by performing spatial interpolation processing on the sensor data, different monitoring levels (such as high altitude, ground, etc.) and different environmental factors can be fused to optimize the estimation of gas concentration. For example, in the exhaust emission monitoring of a thermal power plant, gas sensors at different heights may have different gas concentration measurement values. After combining with environmental data, the gas concentration change trend in a specific area can be accurately evaluated. Finally, the multidimensional gas sensor network obtained by data fusion can provide more comprehensive and accurate gas concentration monitoring data, providing support for subsequent environmental monitoring and pollution control.

[0135] Optionally, step S3 specifically includes:

[0136] Step S31: collecting real-time sensing data of the monitoring area based on the multi-dimensional gas sensing network of the monitoring area, thereby obtaining real-time gas sensing data of the monitoring area and real-time environmental sensing data of the monitoring area;

[0137] In this embodiment, real-time gas sensor data and environmental sensor data are collected through a multi-dimensional gas sensor network in the monitoring area. Gas sensors are connected through a wireless sensor network (WSN) or a wired sensor network to detect the concentration of gases such as carbon dioxide, carbon monoxide, nitrogen oxides, and volatile organic compounds (VOCs) in real time. Environmental sensors include data such as temperature, humidity, air pressure, and wind speed. These sensors are distributed in different areas, and the real-time collected data is gathered to the data processing center through a transmission system for further analysis. For example, in a factory area, gas sensors are distributed in production workshops, storage areas, etc., while environmental sensors are responsible for recording parameters such as temperature and humidity to ensure the comprehensiveness and accuracy of the data.

[0138] Step S32: performing gas sensor signal noise point identification on the real-time gas sensor data of the monitoring area, thereby obtaining gas sensor signal noise point data;

[0139] In this embodiment, a signal processing algorithm is used to identify noise points in real-time gas sensor data. Common noise points include erroneous data caused by environmental interference, electromagnetic waves, sensor failures, etc. Filtering algorithms, such as median filtering and low-pass filtering, are used to detect abnormal fluctuations in gas sensor data. Real-time data is monitored by setting a threshold. Once the data exceeds the preset standard deviation range, it is considered as noise data. For example, when monitoring nitrogen oxides in real time, a data point jumps due to interference from the equipment itself or the outside world. At this time, the data point needs to be identified and marked as a noise point to avoid its impact on subsequent analysis.

[0140] Step S33: performing sensor signal noise time series correlation according to the gas sensor signal noise data and the real-time environmental sensor data of the monitoring area, thereby obtaining environmental noise source data;

[0141] In this embodiment, the noise data of the gas sensor signal and the environmental sensor data are combined, and the noise in the signal and the environmental factors are correlated and analyzed through a timing analysis algorithm (such as dynamic time warping (DTW) and correlation analysis). There is a certain timing relationship between noise and specific environmental factors (such as wind speed changes and temperature fluctuations). For example, when the wind speed changes drastically, the gas concentration measurement value will fluctuate, resulting in noise in the sensor data. Through timing correlation analysis, it is possible to confirm which environmental factors have the greatest impact on the gas sensor data in certain time periods, thereby marking these environmental factors as noise sources and eliminating their interference with the gas concentration analysis.

[0142] Step S34: selecting a gas concentration influencing factor from a gas concentration influencing factor set according to the environmental noise source data, thereby obtaining a noise source gas concentration influencing factor, and performing signal environmental noise denoising on the real-time gas sensor data in the monitoring area according to the noise source gas concentration influencing factor, thereby obtaining gas sensor denoised signal data;

[0143] In this embodiment, the gas concentration influencing factors related to the noise source are selected based on the obtained environmental noise source data. For example, when it is found that the wind speed change is the main interference factor of the gas concentration change, the wind speed can be used as one of the influencing factors of the gas concentration. Then, the main environmental factors related to the gas concentration change are screened out through multiple regression analysis or principal component analysis (PCA). When denoising, these influencing factors are used to adjust the original gas sensor data to eliminate noise. For example, the wind speed data is used to correct the fluctuation of the gas concentration. When the wind speed is high, the gas sensor data is corrected through the model to reflect the actual gas concentration level.

[0144] Step S35: performing sensor response pattern recognition on the gas sensor denoised signal data to obtain sensor response pattern data.

[0145] In this embodiment, the response pattern recognition of the denoised gas sensor data is performed by machine learning or pattern recognition technology. Use trained classifiers (such as support vector machines (SVM), random forests, neural networks, etc.) to identify the response patterns of gas sensors under different environmental conditions. These patterns are usually manifested as the response characteristics of gas sensors to changes in different gas concentrations. For example, some gas sensors are sensitive to changes in nitrogen oxide concentrations, while other sensors may be more sensitive to sulfur dioxide or carbon monoxide. By identifying the response pattern of the gas sensor, the system can further optimize the measurement and analysis of gas concentrations to ensure the accuracy and reliability of the monitoring results. In implementation, the data of the gas sensor may be compared with the preset response pattern to determine the accuracy and relevance of the current data to ensure that the final output data is reliable.

[0146] Optionally, step S35 is specifically:

[0147] Step S351: extracting the signal time domain features of the gas sensor denoised signal data to obtain the gas sensor signal time domain data, and performing frequency domain conversion on the gas sensor signal time domain data to obtain the gas sensor signal spectrum;

[0148] In this embodiment, time domain features are extracted from the denoised gas sensor signal. Common time domain features include the mean, variance, maximum, minimum, peak, waveform factor, etc. of the signal. These features help capture basic information about changes in gas concentration. For example, for a carbon monoxide sensor, the fluctuation of gas concentration can be described by calculating the mean and variance of the signal. Next, the time domain data is converted to the frequency domain, for example, using a fast Fourier transform (FFT) to convert the time domain signal into a frequency domain signal to obtain the spectrum of the gas sensor signal. Frequency domain features help analyze the periodicity and frequency characteristics of changes in gas concentration. For example, if the spectrum of the sensor data shows a significant peak in a certain frequency band, it may mean that there is a periodic source of interference in the change in gas concentration.

[0149] Step S352: obtaining a gas property expert experimental database, and extracting gas property from the gas property expert experimental database, thereby obtaining gas property data;

[0150] In this embodiment, gas-related experimental data is obtained from a gas property expert experimental database. The database records information such as sensor response characteristics, concentration range, reaction time, sensitivity, selectivity, etc. of different gases. For example, a database contains response data of gases such as methane, nitrogen oxides, and carbon dioxide at different concentrations. By extracting gas properties from the database, the main characteristics of each gas are extracted, such as the concentration-response curve, reaction rate, etc. of the gas. For example, methane responds slowly to the sensor at low concentrations, but responds quickly at high concentrations. These gas property data can provide basic data for subsequent simulation responses.

[0151] Step S353: performing gas sensor response simulation according to the gas characteristic data and the gas sensor denoised signal data, thereby obtaining gas sensor simulated response data;

[0152] In this embodiment, a mathematical model or a machine learning algorithm is used to simulate the response of the gas sensor based on the gas characteristic data and the denoised signal data of the gas sensor. Common simulation methods include regression models based on the relationship between gas concentration and sensor response, neural network models, etc. For example, for a carbon monoxide sensor, the output signal of the sensor at different concentrations can be simulated by using a regression model of gas concentration and sensor response. The gas characteristic data provides an accurate gas concentration range and sensor characteristics for the simulation, while the denoised signal ensures that the input data of the simulation is more accurate and reduces noise interference.

[0153] Step S354: performing structural integration of gas sensor response feature mapping based on the gas sensor simulated response data, thereby obtaining gas sensor simulated response pattern data;

[0154] In this embodiment, the simulated response data of the gas sensor is feature mapped by structured integration. By mapping the sensor response data to a certain response pattern, the simulated response pattern of the gas sensor is obtained. Common methods include cluster analysis, dimensionality reduction technology (such as principal component analysis PCA), etc. For example, the simulated response data of the gas sensor is divided into several categories, each category representing the response pattern of the sensor to different gases. Through the clustering algorithm, the sensor can be divided into different patterns according to its response characteristics to different gas concentrations, thereby obtaining more intuitive and structured response pattern data.

[0155] Step S355: performing response pattern similarity calculation on the gas sensor simulated response pattern data and the gas sensor signal spectrum, thereby obtaining simulated response pattern similarity data;

[0156] In this embodiment, the similarity between the simulated response pattern data of the gas sensor and the signal spectrum of the gas sensor is calculated to evaluate the response similarity of different sensors in similar gas environments. Commonly used similarity calculation methods include cosine similarity, Euclidean distance, correlation coefficient, etc. For example, when analyzing the gas sensor responses of carbon monoxide and nitrogen oxides, the similarity between their simulated response patterns and spectra can be calculated. Through similarity calculation, it can be determined which sensors have similar responses to different gases and which show significant differences. This provides a basis for subsequent response pattern recognition and sensor selection.

[0157] Step S356: Perform sensor response pattern recognition based on the simulated response pattern similarity data to obtain sensor response pattern data.

[0158] In this embodiment, the sensor response pattern is identified using pattern recognition technology based on the obtained simulated response pattern similarity data. The simulated response pattern is classified by a classification algorithm (such as support vector machine SVM, K-nearest neighbor KNN, random forest, etc.). For example, if two sensors show a high degree of similarity in the simulated response pattern, they are classified into the same category, indicating that the response patterns of the two sensors to a certain type of gas are similar. Finally, through pattern recognition, the sensor response pattern data is obtained, which can be used for gas identification and concentration measurement in real-time monitoring to accurately detect the type and concentration of the gas.

[0159] Optionally, step S353 is specifically:

[0160] The gas sensor denoising signal data is classified by sensor type, thereby obtaining MEMS micro-nano gas sensor signal data and non-MEMS micro-nano gas sensor signal data;

[0161] In this embodiment, the obtained gas sensor denoising signal data includes output signals of different types of gas sensors. These signal data are classified by using a classification algorithm (such as a support vector machine SVM or a decision tree algorithm) based on pre-calibrated sensor type characteristics (such as the response characteristics of MEMS micro-nano sensors and non-MEMS micro-nano sensors). For example, MEMS micro-nano sensors usually have higher sensitivity and shorter response time, while non-MEMS micro-nano sensors may have different response curves. Through classification processing, the signals are classified into two categories, MEMS and non-MEMS, to obtain their respective signal data. For example, by analyzing the frequency response characteristics of the sensor, the output characteristics of MEMS and non-MEMS sensors can be effectively distinguished.

[0162] Initialize the response simulation parameters according to the gas characteristic data and the MEMS micro-nano gas sensor signal data, so as to obtain the MEMS micro-nano type sensor initialization response simulation parameters;

[0163] In this embodiment, the sensor response simulation parameters are initialized according to the gas characteristics (such as the chemical composition, reaction temperature and concentration range of the gas) and the signal data of the MEMS micro-nano gas sensor. 2 Based on the known gas characteristic data, the sensor can deduce the sensitivity, response time, maximum concentration response and other parameters of the gas by looking up experimental data or models. This initialization process includes inferring the effect of gas concentration on the sensor from the relationship between gas concentration and sensor signal. By establishing a mathematical model, parameters such as linear or nonlinear relationships are determined, which are used as the basic data for subsequent response simulation. After the simulation parameters are initialized, an accurate starting point can be provided for subsequent sensor response simulation.

[0164] Initializing response simulation parameters according to non-MEMS micro-nano gas sensor signal data and gas characteristic data, thereby obtaining initialization response simulation parameters of non-MEMS micro-nano type sensors;

[0165] In this embodiment, for non-MEMS micro-nano gas sensors. By acquiring the signal data of the non-MEMS micro-nano sensor and the characteristic data of the corresponding gas, the response simulation parameters are initialized. Taking the nitrogen oxide (NOx) sensor as an example, the non-MEMS sensor usually has different characteristics from the MEMS sensor in long-term stability and selectivity. By analyzing the experimental data of different gases in different concentration ranges, the response time, sensitivity and maximum response value of the non-MEMS sensor are determined. This process can be based on the experimental database, using a mathematical model between gas concentration and sensor output (such as linear regression or neural network model) to initialize the response simulation parameters.

[0166] Perform sensor physical response modeling on the MEMS micro-nano gas sensor signal data and the non-MEMS micro-nano gas sensor signal data, so as to obtain the MEMS micro-nano sensor response model and the non-MEMS micro-nano sensor response model;

[0167] In this embodiment, physical response models of the sensors are constructed based on the signal data of MEMS and non-MEMS gas sensors. MEMS sensors usually have a smaller size and higher sensitivity, and the response model can be established using physical principles such as the thermoelectric effect of micro sensors and gas molecule adsorption reactions. For example, the relationship between the temperature sensor and the gas reaction is used to describe the response characteristics of the MEMS sensor. Non-MEMS sensors respond to gas concentrations through changes in the conductivity of metal oxide semiconductors, and the nonlinear relationship between conductivity and gas concentration can be used to construct a response model. These models can accurately describe the response behavior of different types of sensors under different gas concentrations by fitting experimental data.

[0168] Through the MEMS micro-nano sensor response model, and using the MEMS micro-nano type sensor initialization response simulation parameters, the gas sensor response simulation of the multi-dimensional gas sensor network in the monitoring area is performed to obtain the MEMS micro-nano type sensor response simulation data;

[0169] In this embodiment, the obtained MEMS micro-nano sensor response model is used in combination with the initialized response simulation parameters to perform gas sensor response simulation on the multi-dimensional gas sensor network in the monitoring area. The simulation can be performed by simulating the concentration distribution of gas at different sensor locations and the response effects of different gases on MEMS micro-nano sensors. For example, finite element analysis (FEA) or computational fluid dynamics (CFD) simulation technology is used to simulate the diffusion process of gas in the monitoring area and how the sensor senses the changes in these gases. Through this simulation, the response data of the MEMS sensor in the actual environment can be obtained, the impact of gas concentration changes on the sensor can be predicted, and the sensor layout can be optimized.

[0170] Through the non-MEMS micro-nano sensor response model and using the non-MEMS micro-nano sensor initialization response simulation parameters, the gas sensor response simulation of the multi-dimensional gas sensor network in the monitoring area is performed to obtain the non-MEMS micro-nano sensor response simulation data;

[0171] In this embodiment, the response model and initialization simulation parameters of the non-MEMS micro-nano gas sensor are used. Through these models, the response of the non-MEMS sensor to the change of gas concentration in the monitoring area is simulated. For example, a similar resistance change model or gas adsorption reaction model is used to simulate the response of the non-MEMS sensor under different gas concentrations. The simulation results can help understand how the non-MEMS sensor responds to gas changes in actual monitoring, and improve its detection capability by adjusting the sensor's layout position or sensitivity and other parameters.

[0172] According to the gas sensor monitoring level data, the sensor response simulation coupling is performed on the MEMS micro-nano type sensor response simulation data and the non-MEMS micro-nano type sensor response simulation data, so as to obtain the gas sensor simulation response mode data.

[0173] In this embodiment, the obtained response simulation data of MEMS and non-MEMS sensors are coupled and combined with the gas sensor monitoring level data (such as the spatial layout of the sensor, sensing accuracy, etc.) to generate gas sensor simulation response pattern data. For example, the simulation data of MEMS and non-MEMS sensors are combined using a weighted average method or a multivariate regression model to evaluate how the response patterns of different sensors interact with each other. In practical applications, it is necessary to conduct a comprehensive evaluation of the data of different types of sensors to achieve more accurate gas concentration detection and gas type identification. For example, the sensor response data and the physical location of the sensor layout can be combined to optimize the sensor layout, improve monitoring accuracy, and reduce the impact of environmental factors. Ultimately, the obtained gas sensor simulation response pattern data will provide a reference for actual deployment.

[0174] Optionally, step S4 is specifically:

[0175] Step S41: extracting sensor output signal data from the gas sensor array data in the monitoring area to obtain sensor output signal data, and performing multi-dimensional signal matrix conversion on the sensor output signal data to obtain a sensor output signal matrix;

[0176] In this embodiment, raw sensor output signal data is obtained from multiple gas sensor arrays in the monitoring area. Assume that various types of gas sensors are deployed in the monitoring area, such as MEMS gas sensors and metal oxide gas sensors. The signal data of each sensor is obtained through the data acquisition module of the sensor (such as an ADC conversion module). Then, these signal data are converted in time and space dimensions. For example, if there are multiple sensors in the monitoring area and the data collected by these sensors have different timestamps and spatial coordinates, the step can be to convert the signal data into a multidimensional matrix through a matrix method, in which each column represents the output data of a certain sensor, and each row represents the output signal of the sensor at a certain moment. This multidimensional signal matrix can provide the spatiotemporal distribution characteristics of the gas concentration in the area, which serves as the basis for subsequent signal analysis and prediction.

[0177] Step S42: performing response pattern feature matching on the sensor output signal matrix based on the sensor response pattern data, thereby obtaining output signal matching response pattern data;

[0178] In this embodiment, feature matching is performed using the obtained sensor output signal matrix and the obtained sensor response pattern data. The sensor response pattern data is obtained through the aforementioned gas sensor response simulation or experimental data, and includes the sensor's response characteristics to different gas types and concentration changes. Using machine learning methods such as dynamic time warping (DTW) or nearest neighbor matching (KNN), the sensor output signal matrix is ​​compared with the response pattern data to identify the response pattern that best matches the current monitoring signal. For example, in monitoring CO 2 By matching the output signal of the gas sensor with the predefined response pattern, it can be determined that the current sensor output signal corresponds to CO 2 response mode, thereby providing accurate characteristic data for subsequent gas concentration prediction.

[0179] Step S43: performing signal feature extraction on the output signal matching response pattern data to obtain output signal feature data, and performing response pattern embedded feature selection on the output signal feature data to obtain a gas sensor feature vector;

[0180] In this embodiment, signal feature extraction is performed on the matched response pattern data. For example, frequency domain, time domain or time-frequency domain features are extracted from the matched signal using techniques such as Fourier transform, principal component analysis (PCA) or wavelet transform. For a gas sensor, the extracted feature data may include the amplitude, waveform, periodicity, phase, etc. of the signal. In addition, for each response pattern, the most representative features may be screened out by an embedded feature selection method (such as feature selection based on L1 regularization, Lasso regression, etc.), thereby generating a feature vector of the gas sensor. For example, by extracting features such as the frequency domain energy of the signal, the sharpness of the signal, and the volatility, a feature vector containing time domain and frequency domain features is formed, representing the gas response capability of the sensor.

[0181] Step S44: uniformly encoding the gas sensor feature vectors to obtain the gas sensor feature coding vectors, and performing feature vector fusion on the gas sensor feature coding vectors to obtain a vector fusion feature matrix;

[0182] In this embodiment, the obtained feature vectors of each sensor are uniformly encoded. For example, one-hot encoding, hash encoding or vectorization technology is used to convert the feature vectors of each sensor into a unified encoding form, so that the feature vectors of different sensors can be processed in the same data space. Then, the data from different sensors are feature fused. For example, deep learning methods such as weighted averaging, principal component analysis (PCA) or convolutional neural network (CNN) can be used to fuse the feature vectors of multiple gas sensors. Through fusion, redundant information can be effectively reduced, and the complementarity between different sensor data can be enhanced to obtain a new vector fusion feature matrix. This fused feature matrix can better represent the gas concentration distribution information in the monitoring area, and can be used as input for subsequent gas concentration prediction models.

[0183] Step S45: construct a gas concentration prediction model based on the vector fusion feature matrix and the monitoring area sensor data.

[0184] In this embodiment, a gas concentration prediction model is constructed based on the obtained vector fusion feature matrix and other sensor data of the monitoring area (such as environmental parameters, temperature and humidity, wind speed, etc.). Specifically, traditional regression models (such as linear regression, support vector machine regression) or deep learning models (such as long short-term memory network LSTM, convolutional neural network CNN, etc.) can be used to predict gas concentration. Taking LSTM as an example, first, by inputting the fusion feature matrix, an LSTM network model is trained, and the timing characteristics of the network are used to capture the dynamic trend of gas concentration over time. Then, the real-time sensor data of the monitoring area is used as input, and the prediction result of gas concentration is output through the model. For example, the model predicts the CO2 concentration in a specific area in the next hour. 2 The concentration change trend can provide support for gas monitoring, control and early warning.

[0185] Optionally, the present specification also provides a gas sensor array data fusion system for executing the gas sensor array data fusion method as described above, the gas sensor array data fusion system comprising:

[0186] The gas sensor monitoring level division module is used to obtain the gas sensor array data of the monitoring area, and perform spatial distribution analysis of the gas sensors in the monitoring area based on the gas sensor array data of the monitoring area, so as to obtain the spatial distribution data of the gas sensors in the monitoring area; and divide the gas sensor monitoring level based on the spatial distribution data of the gas sensors in the monitoring area, so as to obtain the gas sensor monitoring level data;

[0187] A multi-dimensional gas sensor network construction module is used to obtain the environmental sensor data of the monitoring area and divide the environmental sensor data of the monitoring area into sensor data grids, thereby obtaining the sensor grid data of the monitoring area; a multi-dimensional gas sensor network of the monitoring area is constructed according to the environmental sensor grid data of the monitoring area and the gas sensor monitoring level data;

[0188] The sensor response pattern recognition module is used to perform environmental noise denoising of the gas sensor signal based on the multi-dimensional gas sensor network in the monitoring area, thereby obtaining gas sensor denoised signal data; perform sensor response pattern recognition on the gas sensor denoised signal data, thereby obtaining sensor response pattern data;

[0189] The sensor data fusion module is used to optimize the embedded features of the sensor array output signal of the gas sensor array data in the monitoring area based on the sensor response mode data, so as to obtain the gas sensor feature vector; based on the gas sensor feature vector, multi-sensor data fusion processing is performed on the sensor data in the monitoring area to obtain the gas concentration prediction model;

[0190] The environmental adaptive correction module is used to predict the gas concentration in the monitoring area based on the gas concentration prediction model, so as to obtain the original gas concentration prediction data in the monitoring area, and to perform dynamic environmental adaptive correction on the original gas concentration prediction data in the monitoring area, so as to obtain the gas concentration prediction data in the monitoring area.

Claims

1. A gas sensor array data fusion method, characterized in that: The following steps are involved: Step S1: Acquire gas sensor array data in the monitoring area, and perform spatial distribution analysis of gas sensors in the monitoring area based on the gas sensor array data in the monitoring area, thereby obtaining spatial distribution data of gas sensors in the monitoring area; divide gas sensor monitoring levels based on the spatial distribution data of gas sensors in the monitoring area, thereby obtaining gas sensor monitoring level data; Step S2: Acquire environmental sensor data of the monitoring area, and divide the environmental sensor data of the monitoring area into sensor data grids, so as to obtain sensor grid data of the monitoring area; construct a multi-dimensional gas sensor network of the monitoring area according to the environmental sensor grid data of the monitoring area and the gas sensor monitoring level data; Step S3: performing gas sensor signal environmental noise denoising based on the multi-dimensional gas sensor network in the monitoring area, thereby obtaining gas sensor denoised signal data; Performing sensor response pattern recognition on the gas sensor denoised signal data to obtain sensor response pattern data; Step S4: performing embedded feature optimization of sensor array output signals on the gas sensor array data in the monitoring area based on the sensor response pattern data, thereby obtaining a gas sensor feature vector; Based on the gas sensor feature vector, multi-sensor data fusion processing is performed on the monitoring area sensor data to obtain a gas concentration prediction model; Step S5: Predict the gas concentration in the monitoring area based on the gas concentration prediction model to obtain the original gas concentration prediction data in the monitoring area, and perform dynamic environmental adaptive correction on the original gas concentration prediction data in the monitoring area to obtain the gas concentration prediction data in the monitoring area.

2. The gas sensor array data fusion method according to claim 1, characterized in that: Step S1 is specifically as follows: Step S11: acquiring gas sensor array data in the monitoring area, and classifying the gas sensor array data in the monitoring area by sensor type, thereby obtaining MEMS micro-nano gas sensor array data and non-MEMS micro-nano gas sensor array data; Step S12: extracting gas sensor spatial coordinates from the MEMS micro-nano gas sensor array data and the non-MEMS micro-nano gas sensor array data, respectively, so as to obtain MEMS micro-nano gas sensor spatial coordinate data and non-MEMS micro-nano gas sensor data; Step S13: performing a gas sensor spatial distribution analysis based on the MEMS micro-nano gas sensor spatial coordinate data and the non-MEMS micro-nano gas sensor data, thereby obtaining the gas sensor spatial distribution data of the monitoring area; Step S14: evaluating the monitoring capability of the sensor based on the spatial distribution data of the gas sensors in the monitoring area, thereby obtaining the monitoring capability data of the gas sensors; Step S15: Divide the gas sensor monitoring level according to the gas sensor monitoring capability data and the spatial distribution data of the gas sensors in the monitoring area, so as to obtain the gas sensor monitoring level data.

3. The gas sensor array data fusion method according to claim 2, characterized in that: Step S14 is specifically as follows: Step S141: classifying the spatial distribution data of gas sensors in the monitoring area by sensor type, thereby obtaining the spatial distribution data of MEMS micro-nano gas sensors and the spatial distribution data of non-MEMS micro-nano gas sensors; Step S142: constructing a MEMS micro-nano gas sensor coverage model based on the MEMS micro-nano gas sensor spatial distribution data; constructing a non-MEMS micro-nano gas sensor coverage model based on the non-MEMS micro-nano gas sensor spatial distribution data; Step S143: extracting gas sensing features from the gas sensor array data in the monitoring area, thereby obtaining gas sensing data in the monitoring area, and constructing a gas diffusion model in the monitoring area according to the gas sensing data in the monitoring area and the spatial distribution data of the gas sensors in the monitoring area; Step S144: evaluating the gas monitoring capability of the MEMS micro-nano gas sensor according to the gas diffusion model of the monitoring area and the coverage model of the MEMS micro-nano gas sensor, thereby obtaining the gas monitoring capability data of the MEMS micro-nano gas sensor; evaluating the gas monitoring capability of the non-MEMS micro-nano gas sensor according to the gas diffusion model of the monitoring area and the coverage model of the non-MEMS micro-nano gas sensor, thereby obtaining the gas monitoring capability data of the non-MEMS micro-nano gas sensor; Step S145: performing sensor cross-interference evaluation on the gas monitoring capability data of the MEMS micro-nano gas sensor and the gas monitoring capability data of the non-MEMS micro-nano gas sensor, thereby obtaining a regional gas sensor cross-interference factor; Step S146: calibrate the actual monitoring capability of the gas sensor to the gas monitoring capability data of the MEMS micro-nano gas sensor and the gas monitoring capability data of the non-MEMS micro-nano gas sensor according to the regional gas sensor cross-interference factor, so as to obtain the gas sensor monitoring capability data.

4. The gas sensor array data fusion method according to claim 2, characterized in that: Step S15 is specifically as follows: Step S151: performing sensor redundancy evaluation according to the gas sensor monitoring capability data and the spatial distribution data of the gas sensors in the monitoring area, thereby obtaining regional sensor deployment redundancy data; Step S152: performing sensor measurement accuracy and sensitivity level classification based on the gas sensor monitoring capability data, thereby obtaining sensor monitoring capability level data; Step S153: constructing a gas sensor monitoring level division model according to the gas sensor spatial distribution data of the monitoring area and the sensor monitoring capability level data, and performing monitoring level division based on the gas sensor monitoring level division model, thereby obtaining initial gas sensor monitoring level data; Step S154: classifying the initial gas sensor monitoring layer data and the regional sensor deployment redundancy data into abnormal monitoring layer areas through the gas sensor monitoring layer division model, thereby obtaining layer overlapping area data and layer blank area data; Step S155: Perform dynamic weighted optimization adjustment on the initial gas sensor monitoring layer data according to the layer overlapping area data and the layer blank area data, so as to obtain the gas sensor monitoring layer data.

5. The gas sensor array data fusion method according to claim 1, characterized in that: Step S2 is specifically as follows: Step S21: Acquire the environmental sensor data of the monitoring area, and perform environmental sensor spatial distribution statistics on the environmental sensor data of the monitoring area, so as to obtain environmental sensor spatial distribution data; Step S22: performing sensor data grid division based on the spatial distribution data of the environmental sensors, thereby obtaining sensor grid data of the monitoring area; Step S23: mapping the sensor grid sensor data of the monitoring area to the sensor grid data of the monitoring area according to the environmental sensor data of the monitoring area, thereby obtaining the environmental sensor data of the grid of the monitoring area; Step S24: Calculate the gas concentration influencing factors of the monitoring area grid environmental sensor data and the monitoring area gas sensor data, thereby obtaining a gas concentration influencing factor set, and screen the monitoring area grid environmental sensor data according to the gas concentration influencing factor set, thereby obtaining the monitoring area grid main component environmental sensor data; Step S25: Perform spatial multi-dimensional data fusion on the main component environmental sensing data of the monitoring area grid and the gas sensor monitoring level data, so as to obtain a multi-dimensional gas sensor network in the monitoring area.

6. The gas sensor array data fusion method according to claim 1, characterized in that: Step S3 is specifically as follows: Step S31: collecting real-time sensing data of the monitoring area based on the multi-dimensional gas sensing network of the monitoring area, thereby obtaining real-time gas sensing data of the monitoring area and real-time environmental sensing data of the monitoring area; Step S32: performing gas sensor signal noise point identification on the real-time gas sensor data of the monitoring area, thereby obtaining gas sensor signal noise point data; Step S33: performing sensor signal noise time series correlation according to the gas sensor signal noise data and the real-time environmental sensor data of the monitoring area, thereby obtaining environmental noise source data; Step S34: selecting a gas concentration influencing factor from a gas concentration influencing factor set according to the environmental noise source data, thereby obtaining a noise source gas concentration influencing factor, and performing signal environmental noise denoising on the real-time gas sensor data in the monitoring area according to the noise source gas concentration influencing factor, thereby obtaining gas sensor denoised signal data; Step S35: performing sensor response pattern recognition on the gas sensor denoised signal data to obtain sensor response pattern data.

7. The gas sensor array data fusion method according to claim 6, characterized in that: Step S35 is specifically as follows: Step S351: extracting the signal time domain features of the gas sensor denoised signal data to obtain the gas sensor signal time domain data, and performing frequency domain conversion on the gas sensor signal time domain data to obtain the gas sensor signal spectrum; Step S352: obtaining a gas property expert experimental database, and extracting gas property from the gas property expert experimental database, thereby obtaining gas property data; Step S353: performing gas sensor response simulation according to the gas characteristic data and the gas sensor denoised signal data, thereby obtaining gas sensor simulated response data; Step S354: performing structural integration of gas sensor response feature mapping based on the gas sensor simulated response data, thereby obtaining gas sensor simulated response pattern data; Step S355: performing response pattern similarity calculation on the gas sensor simulated response pattern data and the gas sensor signal spectrum, thereby obtaining simulated response pattern similarity data; Step S356: Perform sensor response pattern recognition based on the simulated response pattern similarity data to obtain sensor response pattern data.

8. The gas sensor array data fusion method according to claim 7, characterized in that: Step S353 is specifically as follows: The gas sensor denoising signal data is classified by sensor type, thereby obtaining MEMS micro-nano gas sensor signal data and non-MEMS micro-nano gas sensor signal data; Initialize the response simulation parameters according to the gas characteristic data and the MEMS micro-nano gas sensor signal data, so as to obtain the MEMS micro-nano type sensor initialization response simulation parameters; Initializing response simulation parameters according to non-MEMS micro-nano gas sensor signal data and gas characteristic data, thereby obtaining initialization response simulation parameters of non-MEMS micro-nano type sensors; Perform sensor physical response modeling on the MEMS micro-nano gas sensor signal data and the non-MEMS micro-nano gas sensor signal data, so as to obtain the MEMS micro-nano sensor response model and the non-MEMS micro-nano sensor response model; Through the MEMS micro-nano sensor response model, and using the MEMS micro-nano type sensor initialization response simulation parameters, the gas sensor response simulation of the multi-dimensional gas sensor network in the monitoring area is performed to obtain the MEMS micro-nano type sensor response simulation data; Through the non-MEMS micro-nano sensor response model and using the non-MEMS micro-nano sensor initialization response simulation parameters, the gas sensor response simulation of the multi-dimensional gas sensor network in the monitoring area is performed to obtain the non-MEMS micro-nano sensor response simulation data; According to the gas sensor monitoring level data, the sensor response simulation coupling is performed on the MEMS micro-nano type sensor response simulation data and the non-MEMS micro-nano type sensor response simulation data, so as to obtain the gas sensor simulation response mode data.

9. The gas sensor array data fusion method according to claim 1, characterized in that: Step S4 is specifically as follows: Step S41: extracting sensor output signal data from the gas sensor array data in the monitoring area to obtain sensor output signal data, and performing multi-dimensional signal matrix conversion on the sensor output signal data to obtain a sensor output signal matrix; Step S42: performing response pattern feature matching on the sensor output signal matrix based on the sensor response pattern data, thereby obtaining output signal matching response pattern data; Step S43: performing signal feature extraction on the output signal matching response pattern data to obtain output signal feature data, and performing response pattern embedded feature selection on the output signal feature data to obtain a gas sensor feature vector; Step S44: uniformly encoding the gas sensor feature vectors to obtain the gas sensor feature coding vectors, and performing feature vector fusion on the gas sensor feature coding vectors to obtain a vector fusion feature matrix; Step S45: construct a gas concentration prediction model based on the vector fusion feature matrix and the monitoring area sensor data.

10. A gas sensor array data fusion system, characterized in that: Used to execute the gas sensor array data fusion method as claimed in claim 1, the gas sensor array data fusion system comprises: The gas sensor monitoring level division module is used to obtain the gas sensor array data of the monitoring area, and perform spatial distribution analysis of the gas sensors in the monitoring area based on the gas sensor array data of the monitoring area, so as to obtain the spatial distribution data of the gas sensors in the monitoring area; and divide the gas sensor monitoring level based on the spatial distribution data of the gas sensors in the monitoring area, so as to obtain the gas sensor monitoring level data; A multi-dimensional gas sensor network construction module is used to obtain the environmental sensor data of the monitoring area and divide the environmental sensor data of the monitoring area into sensor data grids, thereby obtaining the sensor grid data of the monitoring area; a multi-dimensional gas sensor network of the monitoring area is constructed according to the environmental sensor grid data of the monitoring area and the gas sensor monitoring level data; The sensor response pattern recognition module is used to perform environmental noise denoising of the gas sensor signal based on the multi-dimensional gas sensor network in the monitoring area, thereby obtaining gas sensor denoised signal data; perform sensor response pattern recognition on the gas sensor denoised signal data, thereby obtaining sensor response pattern data; The sensor data fusion module is used to optimize the embedded features of the sensor array output signal of the gas sensor array data in the monitoring area based on the sensor response mode data, so as to obtain the gas sensor feature vector; based on the gas sensor feature vector, multi-sensor data fusion processing is performed on the sensor data in the monitoring area to obtain the gas concentration prediction model; The environmental adaptive correction module is used to predict the gas concentration in the monitoring area based on the gas concentration prediction model, so as to obtain the original gas concentration prediction data in the monitoring area, and to perform dynamic environmental adaptive correction on the original gas concentration prediction data in the monitoring area, so as to obtain the gas concentration prediction data in the monitoring area.

Images