Inspection monitoring and early warning method and system based on odor recognition

Through the combination of sensor arrays and machine learning models, the problem of untimely and inaccurate odor recognition patrol monitoring and early warning in the prior art is solved, real-time, automated monitoring of gas samples and high-accurate odor recognition are achieved, and The predictions of odor changes and abnormal development and source information improve the timeliness and accuracy of early warnings.

CN118393085BActive Publication Date: 2024-11-19SHENZHEN XINWEN INTELLIGENT TECH CO LTD
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202410626478.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-20
Publication Date
2024-11-19
Estimated Expiration
2044-05-20

AI Technical Summary

Technical Problem

The existing odor identification patrol monitoring and early warning technology is difficult to deal with complex gas mixtures, resulting in false alarms or misreports in dynamically changing environments, and is unable to provide predictions of odor changes, abnormal development and source information, resulting in untimely warnings. and inaccurate.

Method used

Using a method based on sensor array and machine learning model, gas samples in the environment are obtained, signal preprocessed, and then input them into the machine learning model for comparison with the pre-stored odor database. If an abnormality is found, multi-dimensional time series analysis is performed. To determine the odor change trend and the source of abnormal development, and trigger corresponding early warnings.

Benefits of technology

Real-time, automated monitoring of gas samples and high-accurate odor recognition are achieved, artificial errors are reduced, and the efficiency and accuracy of patrol monitoring are improved, and potential risks can be discovered in a timely manner and accurate warning information is provided.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118393085B_ABST
    Figure CN118393085B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of gas detection, and provides a patrol monitoring and early warning method and system based on odor recognition, including: obtaining gas samples in the environment, obtaining multiple electrical signals of the gas samples according to a sensor array; preprocessing each electrical signal; inputting the preprocessed multiple electrical signals into a preset machine learning model, outputting a sample odor combination, and judging whether the sample odor combination is abnormal; if the sample odor combination is abnormal, performing a multi-dimensional time series analysis on the sample odor combination to obtain the odor change trend, performing an abnormal analysis on the sample odor combination according to the odor change trend, obtaining the abnormal development trend and the abnormal development source; using the abnormal development trend and the abnormal development source as abnormal electrical signals, and triggering corresponding early warnings according to the abnormal electrical signals. Improve the problem that the traditional method cannot provide odor change trends, predictions of abnormal development, and source information, resulting in untimely and inaccurate early warnings.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of gas detection, and in particular to an inspection, monitoring and early warning method and system based on odor recognition. Background Art

[0002] In various production and life fields, gas leakage, emission of harmful substances or other potential environmental changes pose a serious threat to personnel safety and environmental hygiene. In order to timely detect and deal with these potential hazards, the inspection monitoring and early warning system plays a vital role. With the advancement of science and technology, especially the widespread application of the Internet of Things (IoT) and artificial intelligence (AI), the perception and data processing capabilities of monitoring equipment have been significantly improved. Accordingly, odor recognition technology, as an emerging monitoring method, shows great application potential in many fields such as industrial safety, environmental protection and public health.

[0003] Existing odor recognition inspection and monitoring early warning technologies mainly rely on traditional chemical sensors. Although these sensors can respond to certain specific gas components, they are usually difficult to handle complex gas mixtures. In addition, most of these systems use a simple dichotomy to determine whether the gas composition exceeds the preset threshold. This method is prone to false alarms or missed alarms in a dynamically changing environment, and there is a lack of information about the trend of odor changes, abnormal developments, and source information. Therefore, there are obvious deficiencies in the timeliness and accuracy of early warnings. Summary of the invention

[0004] In order to improve the problem that traditional methods are unable to provide odor change trends, predictions of abnormal developments, and source information, resulting in untimely and inaccurate warnings, the present application provides an inspection monitoring and early warning method and system based on odor recognition.

[0005] The present invention provides an inspection, monitoring and early warning method based on odor recognition, which is applied to inspection and monitoring equipment, wherein the inspection and monitoring equipment is equipped with a sensor array, and each sensor in the sensor array is sensitive to a specific gas; the method comprises: obtaining a gas sample in an environment, and obtaining a plurality of electrical signals of the gas sample according to the sensor array; preprocessing each electrical signal, wherein the preprocessing comprises signal amplification, filtering, denoising, correction and normalization; inputting the preprocessed plurality of electrical signals into a preset machine learning model, outputting a sample odor combination, and comparing the sample odor combination with a pre-stored odor combination in a preset odor database to determine whether the sample odor combination is abnormal; if the sample odor combination is abnormal, performing a multi-dimensional time series analysis on the sample odor combination to obtain an odor change trend, and performing an abnormal process analysis on the sample odor combination according to the odor change trend to obtain an abnormal development trend and an abnormal development source; using the abnormal development trend and the abnormal development source as abnormal electrical signals, and triggering a corresponding early warning according to the abnormal electrical signals.

[0006] As a preferred embodiment, the step of obtaining a gas sample in the environment and obtaining multiple electrical signals of the gas sample according to the sensor array includes: initializing the patrol monitoring equipment and starting the sensor array self-test program to ensure that each sensor is in normal operation; planning corresponding sampling points according to a preset patrol trajectory, and performing a gas collection operation at the corresponding collection point according to the vacuum pump and gas flow control valve preset in the patrol monitoring equipment, and using the ambient air collected by the gas collection operation as a gas sample; guiding the gas sample to the sensor array, reacting the reactant in the sensor array with the gas sample to obtain a reaction result, and generating multiple electrical signals corresponding to the gas sample according to the reaction result.

[0007] As a preferred embodiment, each electrical signal is preprocessed, and the preprocessing includes the steps of signal amplification, filtering, denoising, correction and normalization, including: preliminary amplification of the original electrical signal obtained from the sensor array; using a digital filtering algorithm to remove high-frequency noise in the electrical signal after preliminary amplification; applying signal processing technology to eliminate random noise in the electrical signal; comparing the current electrical signal with historical calibration data, identifying the electrical signal drift, and inputting the pattern of the electrical signal drift into a preset calibration algorithm for correction; and normalizing the amplified, filtered, denoised and corrected electrical signal according to a preset standardization method.

[0008] As a preferred embodiment, the step of inputting the preprocessed multiple electrical signals into a preset machine learning model, outputting a sample odor combination, comparing the sample odor combination with the pre-stored odor combination in a preset odor database, and determining whether the sample odor combination is abnormal includes: initializing the machine learning model and loading the pre-trained parameter configuration so that the machine learning model is in a state of waiting to receive data; taking the corrected preprocessed electrical signals as a set, and inputting the set as a feature into the machine learning model, wherein the machine learning model obtains the sample odor combination in the environment by calculation based on the input features; wherein each sample odor combination is represented by a set of feature weights in the machine learning model, representing a specific gas mixture; comparing the sample odor combination output by the machine learning model with the known normal gas sample combination in the preset odor database, and obtaining the similarity between the sample gas combination and the pre-stored odor combination in the odor database; and evaluating the similarity based on a pre-defined abnormality determination threshold, and if the similarity is lower than the threshold, determining that the sample odor combination is abnormal.

[0009] As a preferred solution, if the sample odor combination is abnormal, a multi-dimensional time series analysis is performed on the sample odor combination to obtain an odor change trend, and an abnormal process analysis is performed on the sample odor combination according to the odor change trend to obtain an abnormal development trend and an abnormal development source, including: when it is confirmed that the sample odor combination is abnormal, a multi-dimensional time series analysis is performed on the odor combination to obtain odor combination data at continuous time points, and the dynamic changes of the odor combination data at continuous time points are analyzed, and the odor change trend is analyzed according to the dynamic changes; the odor change trend is compared with a preset normal odor change trend, abnormal data points deviating from the normal odor change trend are determined, and an abnormal trajectory of the sample odor combination is constructed according to the abnormal data points; the odor change trend is input into a preset time series database, the abnormal trend of the odor change trend over time is obtained, and the abnormal development trend is analyzed according to the abnormal trend; the starting point of the abnormal development of the odor is analyzed according to the abnormal development trend and the odor change trend; the position of the development source of the abnormal odor combination is located according to the starting point of the abnormal development of the odor and the starting point of the abnormal trajectory, and the abnormal development source is obtained according to the position of the development source of the abnormal odor combination.

[0010] As a preferred scheme, the step of taking the abnormal development trend and the abnormal development source as abnormal electrical signals and triggering corresponding early warnings according to the abnormal electrical signals includes: encoding the located abnormal development trend and abnormal development source information into a specific abnormal electrical signal, and the abnormal electrical signal will be used to activate the early warning program; automatically selecting a preset early warning level and early warning type according to the characteristics of the abnormal electrical signal; controlling corresponding response equipment according to the early warning level and the early warning type; transmitting the abnormal electrical signal to an alarm system, and the alarm system controls a display to show detailed information on the abnormal development trend and the abnormal development source.

[0011] The present application also provides a patrol monitoring and early warning system based on odor recognition, which is applied to patrol monitoring equipment, wherein the patrol monitoring equipment is equipped with a sensor array, and each sensor in the sensor array is sensitive to a specific gas; the patrol monitoring and early warning system based on odor recognition includes: an acquisition module, which is used to acquire gas samples in the environment, and acquire multiple electrical signals of the gas samples according to the sensor array; a processing module, which is used to preprocess each electrical signal, and the preprocessing includes signal amplification, filtering, denoising, correction and normalization; a comparison module, which is used to input the preprocessed multiple electrical signals into a preset machine learning model, output a sample odor combination, compare the sample odor combination with the pre-stored odor combination in a preset odor database, and determine whether the sample odor combination is abnormal; an analysis module, which is used to perform multi-dimensional time series analysis on the sample odor combination if there is an abnormality in the sample odor combination, obtain the odor change trend, and perform abnormal process analysis on the sample odor combination according to the odor change trend to obtain the abnormal development trend and the abnormal development source; an early warning module, which is used to use the abnormal development trend and the abnormal development source as abnormal electrical signals, and trigger corresponding early warnings according to the abnormal electrical signals.

[0012] The present application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, any of the above-mentioned inspection, monitoring and early warning methods based on odor recognition is implemented.

[0013] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the inspection, monitoring and early warning method based on odor recognition as described in any one of the above.

[0014] Compared with the prior art, the present application has the following beneficial effects: timely warning and high accuracy. By integrating sensor arrays and machine learning technology, real-time, automated monitoring of gas samples and high-accuracy odor identification are achieved, which not only reduces human errors, but also greatly improves the efficiency and accuracy of patrol monitoring. Through automated abnormal process analysis and early warning triggering, it helps to discover potential risk conditions earlier, and then provide timely early warning information, improving the traditional method that cannot provide odor change trends, abnormal development predictions and source information, resulting in untimely and inaccurate early warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0016] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0017] Figure 1 It is a flow chart of a patrol monitoring and early warning method and system based on odor recognition provided by an embodiment of the present invention;

[0018] Figure 2 is a schematic block diagram of the structure of the inspection, monitoring and early warning system provided by an embodiment of the present invention;

[0019] Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention.

[0020] Description of reference numerals:

[0021] 10. Inspection, monitoring and early warning system; 11. Acquisition module; 12. Processing module; 13. Comparison module; 14. Analysis module; 15. Early warning module; 20. Electronic equipment; 21. Memory; 22. Processor. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0024] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0025] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0026] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.

[0027] Embodiment 1:

[0028] like Figure 1 As shown, an inspection, monitoring and early warning method based on odor recognition provided in an embodiment of the present application is applied to an inspection and monitoring device, the inspection and monitoring device is configured with a sensor array, each sensor in the sensor array is sensitive to a specific gas; the method includes steps S100 to S500.

[0029] Step S100, obtaining a gas sample in an environment, and obtaining a plurality of electrical signals of the gas sample according to a sensor array.

[0030] In this step, the patrol monitoring equipment performs self-inspection through an internal startup program and automatically navigates to each gas sampling point according to the set patrol path; specifically, a vacuum pump and a gas flow control valve are used to collect gas samples in the ambient air at the sampling point, and the gas samples are introduced into the sensor array, which generates corresponding electrical signals for subsequent processing after reaction.

[0031] For example, if the monitoring target is the air quality near a petrochemical plant, the patrol monitoring equipment will be programmed to collect gas samples at specific locations, such as near warehouses where chemicals are stored, and the sensor array is configured with sensors for detecting volatile organic compounds (VOCs) such as benzene and formaldehyde.

[0032] Step S200, preprocessing each electrical signal, the preprocessing including signal amplification, filtering, noise removal, correction and normalization.

[0033] In this step, in order to ensure the clarity and accuracy of the electrical signal, a series of processing is performed on the electrical signal obtained from the sensor array; specifically, the strength of the electrical signal is enhanced by signal amplification, high-frequency noise is removed by using a digital filtering algorithm, random noise is eliminated by using signal processing technology, and the signal is corrected according to the calibration data, and finally normalized according to the established method.

[0034] For example, if the signal generated by a methane sensor is too weak after collecting a gas sample, the signal will first be amplified, then passed through a low-pass filter to remove ambient noise, then thresholded to correct for any unexpected deviations, and finally normalized to meet the input specifications of the machine learning model.

[0035] Step S300, input the preprocessed multiple electrical signals into a preset machine learning model, output a sample odor combination, compare the sample odor combination with the pre-stored odor combination in a preset odor database, and determine whether there is any abnormality in the sample odor combination.

[0036] In this step, the optimized electrical signal data is input into a pre-trained machine learning model; specifically, the model will extract and analyze the features in these signals, identify the corresponding sample odor combinations, and then match them with the normal odor combinations in the odor database to detect whether any abnormal gas samples appear.

[0037] For example, if there is a risk of ammonia leakage in an industrial area, the machine learning model will analyze the obtained electrical signal set and compare it with the ammonia characteristic pattern in the odor database. If the detected signal intensity or pattern is significantly different from the normal background sample, it is considered that there is an ammonia leak.

[0038] Step S400, if there is an abnormality in the sample odor combination, a multi-dimensional time series analysis is performed on the sample odor combination to obtain the odor change trend, and an abnormal process analysis is performed on the sample odor combination according to the odor change trend to obtain the abnormal development trend and the abnormal development source.

[0039] In this step, an in-depth analysis is conducted on the abnormal odor combination to reveal its changing pattern in the time series; specifically, the time series analysis technology is used to track the dynamic changes of the sample odor combination over time, identify abnormal data points compared with the normal pattern, and reconstruct the abnormal process through these data points to determine the trend and source location of the abnormal development.

[0040] For example, if a persistently intensifying abnormal gas sample is detected in an area where volatile chemicals are stored, time series analysis can reveal a consistent gradual increase in odor intensity, helping to identify the starting point and potential source of the leak.

[0041] Step S500, taking the abnormal development trend and the abnormal development source as abnormal electrical signals, and triggering corresponding warnings according to the abnormal electrical signals.

[0042] In this step, the collected abnormal information is converted into a specific electrical signal through encoding to trigger an early warning response; specifically, this abnormal electrical signal determines the early warning level and type according to its nature, triggers the alarm system to issue a warning, and clearly displays the abnormal development trend and detailed information on the source of the abnormal development on the display, so that operators can respond quickly.

[0043] For example, once a leak of harmful gas exceeding the standard is detected, an abnormal electrical signal will be sent to the central control system, automatically activating a high-level sound and light alarm, and at the same time sending detailed leak source information to operators, instructing them to close the leaking area and start emergency evacuation procedures.

[0044] In this embodiment, the patrol monitoring equipment uses a sensor array to automatically obtain gas samples in the environment, and converts the detected gas samples into multiple electrical signals. Subsequently, these electrical signals are preprocessed one by one, including signal amplification, filtering, denoising, correction and normalization to improve the accuracy of subsequent analysis. The preprocessed electrical signals are input into a preset machine learning model, and the model analyzes and outputs the odor combination of the gas sample, and compares these odor combinations with the pre-stored odor combinations in the odor database to determine whether there is an abnormality. If the odor combination of the gas sample is identified as abnormal, the odor combination of the gas sample will be subjected to a multi-dimensional time series analysis, the odor change trend will be calculated, and the abnormal process will be further analyzed to identify the abnormal development trend and the source of the abnormal development. Finally, the patrol monitoring equipment triggers an early warning electrical signal based on the obtained abnormal development trend and the source of the abnormal development to initiate the corresponding early warning response.

[0045] By integrating sensor arrays and machine learning technology, real-time, automated monitoring of gas samples and highly accurate odor identification are achieved, which not only reduces human errors, but also greatly improves the efficiency and accuracy of patrol monitoring. Through automated abnormal process analysis and early warning triggering, this method helps to detect potential risk conditions earlier, thereby providing timely early warning information, improving the problem that traditional methods are unable to provide odor change trends, predictions of abnormal developments, and source information, resulting in untimely and inaccurate early warnings.

[0046] Embodiment 2:

[0047] In step S100, the patrol monitoring equipment needs to be initialized and the sensor array self-test program needs to be started to ensure that each sensor is in normal operation; the corresponding sampling points are planned according to the preset patrol trajectory, and the gas collection operation is performed at the corresponding collection points according to the vacuum pump and gas flow control valve preset in the patrol monitoring equipment, and the ambient air collected by the gas collection operation is used as a gas sample; the gas sample is guided to the sensor array, and the reaction substance in the sensor array reacts with the gas sample to obtain a reaction result, and a plurality of electrical signals corresponding to the gas sample are generated according to the reaction result.

[0048] The equipment is initialized through an automated control program, and a self-test and calibration process is performed to ensure the accuracy and stability of the sensor array. Specifically, the reaction between the gas sample and the specific reactants in the sensor array is used to convert the detected chemical components into corresponding electrical signals, which serve as the basis for subsequent identification and analysis.

[0049] For example, consider a monitoring scenario located in a chemical plant area. Here, patrol monitoring equipment is programmed to regularly test the air quality in the storage area to catch any harmful gas leaks. The equipment arrives at the predetermined leak point, such as around a chlorine storage tank, and the air pump collects the surrounding air and detects it through a chlorine sensor. The sensor's reaction to chlorine gas produces an electrical signal indicating the presence and concentration of the gas.

[0050] In step S200, the original electrical signal obtained from the sensor array is preliminarily amplified; a digital filtering algorithm is used to remove high-frequency noise in the electrical signal after the preliminary amplification; signal processing technology is applied to eliminate random noise in the electrical signal; the current electrical signal is compared with historical calibration data to identify the electrical signal drift, and the pattern of the electrical signal drift is input into a preset calibration algorithm for correction; the amplified, filtered, denoised and corrected electrical signal is normalized according to a preset standardization method.

[0051] The signal quality is improved through a series of electrical signal purification and enhancement steps, laying the foundation for accurate identification; specifically, the original electrical signal is processed through the comprehensive use of amplification, digital filtering, signal correction and other technologies to prepare clear and consistent input data for the subsequent machine learning model analysis.

[0052] For example, when monitoring an ammonia leak, the gas sensor is sensitive to changes in temperature and humidity, which causes the signal to drift. In this case, the system can compare the current sensor output with historical calibration data and use the calibration algorithm to adjust the signal to ensure the accuracy of subsequent analysis.

[0053] In step S300, the machine learning model is initialized and the pre-trained parameter configuration is loaded to put the machine learning model in a state of waiting to receive data; the corrected pre-processed electrical signal is taken as a set, and the set is input into the machine learning model as a feature, and the machine learning model obtains the sample odor combination in the environment through calculation based on the input features; wherein each sample odor combination is represented by a set of feature weights in the machine learning model, representing a specific gas mixture; the sample odor combination output by the machine learning model is compared with the known normal gas sample combination in the preset odor database to obtain the similarity between the sample gas combination and the pre-stored odor combination in the odor database; the similarity is evaluated according to a pre-defined abnormality judgment threshold, and if the similarity is lower than the threshold, the sample odor combination is judged to be abnormal.

[0054] Through the efficient calculation of the machine learning model, pattern recognition and feature extraction are performed on the preprocessed electrical signals; specifically, the pre-trained model is used to accelerate the recognition process of sample odor combinations, and compared with the pre-stored odor combinations in the odor database, abnormal situations are automatically evaluated.

[0055] For example, this monitoring equipment is used in an electronics manufacturing factory to look for solvent vapor leaks. After the machine learning model calculates the odor combination corresponding to the received electrical signal, if the combination matches the solvent vapor characteristics recorded in the database, but has a low similarity with the odor combination of the factory air under normal circumstances, the system will issue an early warning of a solvent vapor leak.

[0056] In step S400, when it is confirmed that there is an abnormality in the sample odor combination, a multidimensional time series analysis is performed on the odor combination to obtain odor combination data at continuous time points, and the dynamic changes of the odor combination data at continuous time points are analyzed, and the odor change trend is analyzed based on the dynamic changes.

[0057] Specifically, a baseline time series is formed based on historical data, and the sample odor combinations under normal conditions are recorded. Once the machine learning model finds that the odor combination of the current gas sample does not match the standard odor pattern in the odor database, it indicates an abnormality. At this time, the system will start a multidimensional time series analysis program to collect odor data at continuous time points starting from the time the abnormality occurred. These data are usually represented by the electrical signal intensity of the sensor and are associated with a timestamp, recording the odor intensity corresponding to each time point. Next, time series analysis techniques, such as autoregressive models (AR) and moving average models (MA), are used to analyze the changes in these data over time to derive the trend of odor changes.

[0058] The odor change trend is compared with the preset normal odor change trend, the abnormal data points that deviate from the normal odor change trend are determined, and the abnormal trajectory of the sample odor combination is constructed based on the abnormal data points.

[0059] Specifically, the system compares the newly derived odor change trend with the preset normal odor change trend archive. The normal odor change trend is determined by the standard behavior in the historical data, such as seasonal fluctuations, daily changes and other natural fluctuations. Through comparison, the system can identify abnormal data points that deviate from the normal odor change trend. Based on these abnormal points, the system constructs a trajectory of the change of odor combinations from normal to abnormal, which describes how the abnormal state develops and changes over time.

[0060] The odor change trend is input into a preset time series database, the abnormal trend of the odor change trend over time is obtained, and the abnormal development trend is analyzed based on the abnormal trend.

[0061] Specifically, the odor change trend data collected at a previously collected time point is input into a preset time series database. The database contains the changes of multiple different odor combinations over time, and the normal change parameters under various circumstances have been pre-set. The system uses this database to identify and distinguish between normal change patterns and abnormal change patterns. The abnormal trend is obtained by comparing the currently collected data with the normal change pattern stored in the database. When it is found that the data shows a trend that does not match the normal pattern, the system will consider this to be an abnormal change trend. Statistical analysis and data mining techniques, such as trend analysis, anomaly detection algorithms, etc., are used to further analyze the abnormal trend in order to reveal and determine the specific pattern and speed of abnormal development, which provides a basis for subsequent determination of the specific source of development and taking action.

[0062] According to the abnormal development trend and odor change trend, the starting point of odor abnormality development can be analyzed.

[0063] Specifically, the above trajectory is further analyzed using data analysis technology, including trend decomposition and cluster analysis of abnormal points. From this, the trend of abnormal development can be identified, such as a gradual increase, periodic mutation, or a continuous high-value platform. Based on the trend analysis results, the specific moment when the odor abnormality began to deteriorate can be traced back to the starting point of the abnormality.

[0064] According to the starting point of the abnormal development of the odor and the starting point of the abnormal trajectory, the position of the source of the development of the abnormal odor combination is located, and the source of the abnormal development is obtained according to the position of the source of the development of the abnormal odor combination.

[0065] Specifically, the system combines the results of time series analysis with spatial distribution models, such as the Gaussian plume diffusion model, to estimate the development path of abnormal odor combinations and locate their source locations. In this way, the monitoring system can provide the exact location of the source of abnormal development, providing important information for taking action or further investigation.

[0066] Through in-depth analysis of time series, the dynamic changes and development trends of sample odor combinations are revealed, and statistical analysis and time series methods are applied to track and predict the development process of odor anomalies, thereby determining the origin and development direction of the anomalies for more effective supervision and immediate response.

[0067] For example, the patrol monitoring equipment is deployed in an industrial area to monitor the leakage of specific harmful gases. When a sensor detects that the concentration of a harmful gas begins to exceed the normal level, the system records the data at this time point and continues to monitor subsequent changes. Through time series analysis, the system observes that the concentration of the gas has a trend of gradually increasing, which is significantly different from the normal situation, indicating that a leak may have occurred. Then, by analyzing the trend, the location and time of the abnormal data points, the system infers that the initial location of the leak is likely to be the equipment or storage container near the initial detection of the abnormal concentration. Subsequently, the system notifies the staff to go to the location for inspection and emergency treatment.

[0068] In step S500, the located abnormal development trend and abnormal development source information are encoded into a specific abnormal electrical signal, and the abnormal electrical signal will be used to activate the early warning program; according to the characteristics of the abnormal electrical signal, the preset early warning level and early warning type are automatically selected; the corresponding response equipment is controlled according to the early warning level and early warning type; the abnormal electrical signal is transmitted to the alarm system, and the alarm system controls the display to show detailed information on the abnormal development trend and the source of the abnormal development.

[0069] Through data coding and automated early warning processes, timely and effective safety management and emergency response are achieved; specifically, abnormal information is converted into an easily recognizable electrical signal format and provided to the alarm system to trigger customized early warnings and formulate corresponding emergency measures to ensure the efficiency and timeliness of abnormal handling.

[0070] For example, when monitoring a phosphine leak in a pesticide production plant, once the abnormal development trend and source are determined, the patrol monitoring equipment will send an abnormal alarm through a wired or wireless network, allowing the staff in the central monitoring room to quickly dispatch an emergency response team to the abnormal source location to deal with the leak, and at the same time activate the emergency broadcast system in the factory to guide the staff to evacuate as needed or take safety precautions.

[0071] Embodiment 3:

[0072] like Figure 2 As shown, the present application also provides a patrol monitoring and early warning system 10 based on odor recognition, which is applied to patrol monitoring equipment. The patrol monitoring equipment is equipped with a sensor array, and each sensor in the sensor array is sensitive to a specific gas; the patrol monitoring and early warning system 10 based on odor recognition includes an acquisition module 11, a processing module 12, a comparison module 13, an analysis module 14 and an early warning module 15.

[0073] The acquisition module 11 is mainly used to acquire gas samples in the environment and acquire multiple electrical signals of the gas samples according to the sensor array.

[0074] The processing module 12 is mainly used to pre-process each electrical signal, and the pre-processing includes signal amplification, filtering, noise removal, correction and normalization.

[0075] The comparison module 13 is mainly used to input the preprocessed multiple electrical signals into a preset machine learning model, output a sample odor combination, compare the sample odor combination with the pre-stored odor combination in a preset odor database, and determine whether there is any abnormality in the sample odor combination.

[0076] The analysis module 14 is mainly used to perform multi-dimensional time series analysis on the sample odor combination if there is an abnormality in the sample odor combination to obtain the odor change trend, and perform abnormal process analysis on the sample odor combination based on the odor change trend to obtain the abnormal development trend and the abnormal development source.

[0077] The early warning module 15 is mainly used to take the abnormal development trend and the abnormal development source as abnormal electrical signals, and trigger corresponding early warnings according to the abnormal electrical signals.

[0078] By integrating sensor technology, signal processing algorithms, machine learning models and database comparative analysis, a complete workflow from data collection to abnormal warning is realized; specifically, the acquisition module 11 is responsible for real-time collection of gas samples and converting them into electrical signals, the processing module 12 optimizes the quality of electrical signals through advanced signal processing technology, the comparison module 13 analyzes the optimized electrical signals through machine learning models and matches them with the odor database to identify possible abnormalities, the analysis module 14 uses time series analysis technology to further track and identify the development process of abnormalities, and finally the warning module 15 interprets the analysis results, converts them into specific warning signals, and performs real-time abnormal warnings.

[0079] It should be noted that those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system and each module described above can refer to the corresponding process in the above-mentioned patrol monitoring and early warning method embodiment based on odor recognition, and will not be repeated here.

[0080] Embodiment 4:

[0081] like Figure 3 As shown, the present application also provides an electronic device 20: including a memory 21 and a processor 22, the memory 21 stores a computer program that can be run on the processor 22, and when the processor 22 executes the computer program, it implements the inspection, monitoring and early warning method based on odor recognition of any one of Example 1.

[0082] Through the cooperation of hardware components and supporting computer programs, the electronic device 20 can perform a series of complex signal processing and data analysis tasks; specifically, the memory 21 stores the computer program, which contains preset machine learning models and algorithms, and the processor 22 is responsible for running the program to perform functions such as data acquisition, signal preprocessing, odor analysis and comparison, and abnormal trend analysis, thereby realizing automated gas monitoring and early warning.

[0083] Embodiment 5:

[0084] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by the processor 22, the processor 22 executes the inspection, monitoring and early warning method based on odor recognition as described in any one of the embodiments 1.

[0085] By loading a computer program with complex data processing capabilities into a computer-readable storage medium, the processor 22 is endowed with the ability to complete the entire odor monitoring and early warning task; specifically, the computer program contains codes for controlling the processor 22 to execute a complete set of operating procedures including the acquisition, preprocessing, anomaly identification, trend analysis and early warning triggering of gas sensor signals, thereby ensuring the timeliness and accuracy of the system early warning.

[0086] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A patrol monitoring and early warning method based on odor recognition, characterized in that: Applied to patrol monitoring equipment, the patrol monitoring equipment is equipped with a sensor array, each sensor in the sensor array is sensitive to a specific gas; the method comprises: Acquire a gas sample in an environment, and acquire a plurality of electrical signals of the gas sample according to the sensor array; Preprocessing each electrical signal, wherein the preprocessing includes signal amplification, filtering, noise removal, correction and normalization; Initialize the machine learning model and load the pre-trained parameter configuration so that the machine learning model is in a state of waiting to receive data; take the pre-processed electrical signal as a set, and input the set as a feature into the machine learning model, and the machine learning model obtains the sample odor combination in the environment by calculation according to the input features; wherein each sample odor combination is represented by a set of feature weights in the machine learning model, representing a specific gas mixture; compare the sample odor combination output by the machine learning model with the known normal gas sample combination in the preset odor database, and obtain the similarity between the sample gas combination and the pre-stored odor combination in the odor database; evaluate the similarity according to a pre-defined abnormality judgment threshold, and if the similarity is lower than the threshold, determine that the sample odor combination is abnormal; When it is confirmed that the sample odor combination is abnormal, a multi-dimensional time series analysis is performed on the odor combination to obtain odor combination data at continuous time points, and the dynamic changes of the odor combination data at continuous time points are analyzed, and the odor change trend is analyzed according to the dynamic changes; the odor change trend is compared with a preset normal odor change trend, abnormal data points that deviate from the normal odor change trend are determined, and an abnormal trajectory of the sample odor combination is constructed according to the abnormal data points; the odor change trend is input into a preset time series database, and the abnormal trend of the odor change trend over time is obtained, and an abnormal development trend is analyzed according to the abnormal trend; the starting point of the abnormal development of the odor is analyzed according to the abnormal development trend and the odor change trend; the position of the development source of the abnormal odor combination is located according to the starting point of the abnormal development of the odor and the starting point of the abnormal trajectory, and the abnormal development source is obtained according to the position of the development source of the abnormal odor combination; The abnormal development trend and the abnormal development source are used as abnormal electrical signals, and corresponding warnings are triggered according to the abnormal electrical signals.

2. The inspection, monitoring and early warning method based on odor recognition according to claim 1 is characterized in that: The step of acquiring a gas sample in an environment and acquiring a plurality of electrical signals of the gas sample according to the sensor array comprises: Initialize patrol monitoring equipment and start the sensor array self-test program to ensure that each sensor is in normal operation; According to the preset patrol track, the corresponding sampling points are planned, and the gas collection operation is performed at the corresponding collection points according to the vacuum pump and gas flow control valve preset in the patrol monitoring equipment, and the ambient air collected by the gas collection operation is used as the gas sample; The gas sample is guided to the sensor array, and the reaction material in the sensor array reacts with the gas sample to obtain a reaction result, and a plurality of electrical signals corresponding to the gas sample are generated according to the reaction result.

3. The inspection, monitoring and early warning method based on odor recognition according to claim 1 is characterized in that: The preprocessing of each electrical signal includes the steps of signal amplification, filtering, noise removal, correction and normalization, including: Preliminary amplification of the raw electrical signal obtained from the sensor array; Use digital filtering algorithms to remove high-frequency noise from the electrical signal after preliminary amplification; Apply signal processing techniques to eliminate random noise in electrical signals; Comparing the current electrical signal with the historical calibration data, identifying the electrical signal drift, and inputting the pattern of the electrical signal drift into a preset calibration algorithm for correction; The amplified, filtered, denoised and corrected electrical signals are normalized according to a preset standardization method.

4. The inspection, monitoring and early warning method based on odor recognition according to claim 1 is characterized in that: The step of using the abnormal development trend and the abnormal development source as abnormal electrical signals and triggering corresponding warnings according to the abnormal electrical signals includes: Encoding the located abnormal development trend and abnormal development source information into a specific abnormal electrical signal, and the abnormal electrical signal will be used to activate the early warning program; Automatically select the preset warning level and warning type according to the characteristics of abnormal electrical signals; Control corresponding response equipment according to the warning level and the warning type; The abnormal electrical signal is transmitted to an alarm system, and the alarm system controls a display to display detailed information on the abnormal development trend and the source of the abnormal development.

5. A patrol monitoring and early warning system based on odor recognition, characterized in that: Applied to patrol monitoring equipment, the patrol monitoring equipment is equipped with a sensor array, each sensor in the sensor array is sensitive to a specific gas; the patrol monitoring and early warning system based on odor recognition includes: An acquisition module, used for acquiring a gas sample in an environment, and acquiring a plurality of electrical signals of the gas sample according to the sensor array; A processing module, used for preprocessing each electrical signal, wherein the preprocessing includes signal amplification, filtering, noise removal, correction and normalization; A comparison module is used to initialize the machine learning model and load the pre-trained parameter configuration so that the machine learning model is in a state of waiting to receive data; the pre-processed electrical signal is taken as a set, and the set is input into the machine learning model as a feature, and the machine learning model obtains the sample odor combination in the environment by calculation according to the input features; wherein each sample odor combination is represented by a set of feature weights in the machine learning model, representing a specific gas mixture; the sample odor combination output by the machine learning model is compared with the known normal gas sample combination in the preset odor database to obtain the similarity between the sample gas combination and the pre-stored odor combination in the odor database; the similarity is evaluated according to a pre-defined abnormality judgment threshold, and if the similarity is lower than the threshold, it is determined that the sample odor combination is abnormal; An analysis module is used for, when confirming that the sample odor combination is abnormal, performing a multi-dimensional time series analysis on the odor combination to obtain odor combination data at continuous time points, analyzing the dynamic changes of the odor combination data at continuous time points, and analyzing the odor change trend according to the dynamic changes; comparing the odor change trend with a preset normal odor change trend, determining abnormal data points that deviate from the normal odor change trend, and constructing an abnormal trajectory of the sample odor combination according to the abnormal data points; inputting the odor change trend into a preset time series database, obtaining the abnormal trend of the odor change trend over time, and analyzing the abnormal development trend according to the abnormal trend; analyzing the starting point of the abnormal development of the odor according to the abnormal development trend and the odor change trend; locating the location of the development source of the abnormal odor combination according to the starting point of the abnormal development of the odor and the starting point of the abnormal trajectory, and obtaining the abnormal development source according to the location of the development source of the abnormal odor combination; The early warning module is used to take the abnormal development trend and the abnormal development source as abnormal electrical signals, and trigger corresponding early warnings according to the abnormal electrical signals.

6. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, the inspection, monitoring and early warning method based on odor recognition according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by the processor, the processor executes the inspection, monitoring and early warning method based on odor recognition as described in any one of claims 1 to 4.

Citation Information

Patent Citations

  • Cable burning smell continuous detection method based on inspection robot

    CN114047303A

  • Dangerous article intelligent detection and early warning method based on computer machine vision

    CN115099760A

  • Intelligent detection and early warning system for chemical plant environment and harmful gas

    CN117409546A

  • Alarm system and method based on AI algorithm

    CN118038640A

  • Computer-implemented system and method for analyzing mixtures of gases

    US20060155486A1