Fuel gas state early warning and monitoring method and system based on cloud-side cooperation

Abstract

The invention discloses a fuel gas state early warning monitoring method and system based on cloud-side cooperation, and relates to the technical field of fuel gas state early warning. Comprising the following steps: a cloud end preprocesses historical gas data and historical environment data in a standard database to obtain historical concentration data, and analyzes the historical concentration data to obtain a flow velocity calibration reference line; the technical key points are as follows: a cloud edge cooperation technology, a judgment technology combining gas flow rate and use state recognition, and a full-link integration technology combining multiple evaluation and scheme correction technology are adopted, so that a safe and efficient gas early warning and monitoring safety management effect in a household scene can be realized, and the monitoring requirement of fragmented use of the household scene is met; and in cooperation with the effect of graded early warning, the use threshold of a user is reduced, and the method has the advantage of wide adaptability, so that a standardized technical framework is provided for industry gas prediction and supervision, a large-scale application foundation is laid, the use effect is good, and the method has a good use prospect.

Description

Technical Field

[0001] This invention relates to the field of gas status early warning technology, specifically to a gas status early warning monitoring method and system based on cloud-edge collaboration. Background Technology

[0002] There are various safety risks associated with the use of gas in the home. Among them, the most significant are explosions, fires, and poisoning accidents caused by gas leaks. When gas leaks and accumulates in a confined space, if its concentration reaches the explosive limit, it will cause a violent explosion when it comes into contact with an open flame, electric spark, or other ignition source, resulting in damage to the house and injuries or fatalities. If the leaked gas spreads indoors and is not vented in time, its toxic components can be inhaled, leading to poisoning. Mild poisoning may cause symptoms such as dizziness and nausea, while severe poisoning can be life-threatening. In addition, aging household gas appliances, corroded and damaged pipes, detached or aged connecting hoses, improper operation such as not closing valves after use, or equipment malfunctions can also increase the probability of such risks.

[0003] To ensure safe use of natural gas, gas status monitoring and early warning systems are typically implemented. These systems utilize specialized testing equipment to monitor key parameters such as gas leaks, combustion status, and delivery pressure in real time. Once an abnormality is detected, a warning signal is promptly sent to the user through specific methods. This serves as a technical means to prevent safety accidents. The core objective is to detect potential safety hazards during gas use and delivery in advance, giving users time to take emergency response measures.

[0004] The existing patent, authorized by CN120259025B, entitled "Invention Patent for a Gas Integrated Management System Based on a Large Model," describes an invention that includes a gas monitoring module and a gas safety management module. This invention automatically calculates and adjusts the monitoring frequency based on the gas usage patterns of target residents and the real-time status of the gas pipeline network, reducing unnecessary data collection and mitigating potential risks to user privacy; ensuring the real-time nature and accuracy of gas data; and achieving deep fusion and analysis of multi-source data by integrating necessary data such as gas concentration, pipeline pressure, and valve status, as well as optional data such as ambient sound spectrum, combined with graph attention networks and digital twin models, thus improving the comprehensiveness and accuracy of gas safety assessments. Furthermore, it employs a multi-layered safety early warning mechanism, effectively enhancing the sensitivity and reliability of gas safety warnings.

[0005] The solutions described in the aforementioned patents all rely on centralized cloud processing, which leads to excessive dependence on hardware parameters and data foundations, increases system complexity, and significantly increases the time required for leak warning. In addition, existing solutions typically compare real-time collected data with standard data without fully considering the influence of the environment, user behavior, and collective factors of the community, resulting in frequent false gas alarms in actual use.

[0006] In summary, existing integrated gas management solutions do not meet market requirements. Therefore, we propose a gas status early warning and monitoring method and system based on cloud-edge collaboration. Summary of the Invention

[0007] To achieve the above objectives, the present invention provides the following technical solution: A gas status early warning and monitoring method based on cloud-edge collaboration includes the following steps: The cloud performs preprocessing on historical gas data and historical environmental data in the standard database to obtain historical concentration data, and analyzes it to obtain the flow rate calibration baseline; Real-time gas data and real-time environmental data are collected at the edge and preprocessed to obtain real-time concentration data. The flow rate calibration baseline is then used for screening and comparison to determine whether there is any suspicious data. The edge device uploads suspicious data to the cloud, and the cloud performs analysis on the suspicious data to generate a concentration calibration range. The risk level is determined by combining the flow rate calibration baseline and concentration calibration range at the edge end, classifying them according to their usage status, and conducting multiple assessments. The edge device executes corresponding early warning and risk control strategies based on the risk level, and regularly acquires real-time concentration data and execution data, performs multiple assessments to obtain the real-time risk level, executes the early warning and risk control strategies corresponding to the real-time risk level, and stores the end-to-end data in the cloud.

[0008] Preferably, both historical and real-time gas data include methane concentration, gas flow rate, and usage status based on valve switch signals and flame sensor signals. Both historical and real-time environmental data include temperature, humidity, kitchen window open / close status, and exhaust fan operation logs.

[0009] Preferably, the cloud-based preprocessing steps for historical gas data and historical environmental data include: Standardize the data format and unify the data units; Perform outlier removal, deleting data containing fault markers for testing instruments, data with a missing data rate greater than a set value for a given period, and data with contradictory usage status. The Kalman filter method is used to process the methane concentration and gas flow rate after outlier removal, so as to obtain the methane concentration and gas flow rate after noise interference is removed. The methane concentration was adjusted by removing noise interference and considering the gas flow rate and operating status using an environmental impact separation method to obtain the corrected methane concentration. The corrected methane concentration, gas flow rate, and usage status are checked for matching degree and rationality using a logical consistency verification method. If both matching degree and rationality verification are qualified, the corrected methane concentration is determined to be historical concentration data. Otherwise, the data is determined to be contradictory, and the entire set of historical data is deleted.

[0010] Preferably, the steps for analyzing and obtaining the flow velocity calibration baseline include: Historical concentration data, gas flow rate, and usage status during corresponding time periods are correlated and summarized to form a three-dimensional correlated dataset. Based on the usage status, state classification is performed on the 3D associated dataset, forming two sub-datasets; The two types of subsets are divided according to the velocity range of the gas flow rate, and the core features of each velocity range are statistically analyzed. The statistical usage status, flow velocity range, and core characteristics are used to form a baseline range, and the integrated and summarized data are used to form a flow velocity calibration baseline. Historical concentration data of the same building, unit type, gas flow rate, and usage status were filtered, abnormal data were removed, and the data were aggregated to obtain a concentration reference range and a similar database was constructed.

[0011] Preferably, the steps for determining whether suspicious data exists include: Acquire real-time concentration data, gas flow rate, usage status, and flow rate calibration baseline; The baseline interval corresponding to the current flow velocity calibration baseline is determined based on the gas flow velocity and the usage status. Determine the usage status. If it is not in use, extract the preset concentration data and preset gas flow rate in the baseline interval and compare them with the real-time concentration data and gas flow rate. If the real-time concentration data is less than the preset concentration data and the gas flow rate is less than the preset gas flow rate, it is determined that there is no suspicious data and it is stored in the standard database. Otherwise, it is determined that there is suspicious data and additional marking information is added. If the data is in use, the fluctuation amplitude comparison method is used to determine whether there is any suspicious data.

[0012] Preferably, the step of using the fluctuation amplitude comparison method to determine whether there is suspicious data includes: Acquire continuously collected real-time concentration data and gas flow rate; The extreme difference method of continuous data segments is used to calculate the real-time flow velocity fluctuation amplitude. Extract the historical maximum permissible fluctuation range benchmark value from the baseline range; The extreme standard amplitude threshold is calculated based on the historical maximum permissible fluctuation range benchmark value; The real-time flow rate fluctuation amplitude is compared with the limit standard amplitude threshold. If the real-time flow rate fluctuation amplitude is greater than the limit standard amplitude threshold, it is determined that there is suspicious data and a label is added. Otherwise, it is determined that there is no suspicious data and it is stored in the standard database.

[0013] Preferably, when performing analysis on suspicious data in the cloud, real-time concentration data and usage status are extracted from the suspicious data. Based on the real-time concentration data and usage status, data is filtered from similar databases and real-time aggregated data to obtain valid concentration data. Statistical analysis is performed on the valid concentration data to obtain the concentration calibration range and the distribution ratio of usage status.

[0014] Preferably, the steps for performing multiple evaluations include: Based on the usage status, an appropriate pre-judgment scheme is adopted to obtain the pre-judgment result; A core assessment is performed on the preliminary judgment results, and the rationality of the data is verified by comprehensively considering the usage status, gas flow rate, and real-time concentration data to obtain an assessment result. A fallback assessment is performed on the real-time concentration data to obtain a secondary assessment result; False alarm assessments were performed on real-time concentration data, preliminary classification results, concentration calibration ranges, and usage status distribution ratios, resulting in three assessments. Summarize the results of the first, second, and third assessments to output the risk level.

[0015] Preferably, after storing the entire data chain in the cloud, the data is divided into segments according to data characteristics. Continuous trend analysis is performed on continuous data of usage status and high-flow-rate stable data segments, and targeted supplementary analysis is performed on fragmented usage gaps and low-flow-rate high-random data segments. Based on the results of continuous trend analysis and targeted supplementary analysis, the flow rate calibration baseline is corrected and the environmental impact separation method is optimized with real-time concentration data.

[0016] A gas status early warning and monitoring system based on cloud-edge collaboration includes a historical analysis module, a real-time analysis module, an anomaly identification module, a risk identification module, and a risk control module. Historical analysis module: The cloud performs preprocessing on historical gas data and historical environmental data in the standard database to obtain historical concentration data, and analyzes it to obtain the flow rate calibration baseline; Real-time analysis module: Collects real-time gas data and real-time environmental data at the edge, performs preprocessing to obtain real-time concentration data, and calls the flow rate calibration baseline for screening and comparison to determine whether there is any suspicious data; Anomaly detection module: The edge device uploads suspicious data to the cloud, and the cloud performs analysis on the suspicious data to generate a concentration calibration range; Risk identification module: The edge end combines the flow rate calibration baseline and concentration calibration range, classifies according to usage status, performs multiple assessments, and obtains the risk level; Risk management module: The edge device executes corresponding early warning and risk management strategies based on the risk level, and periodically obtains real-time concentration data and execution data, performs multiple assessments to obtain the real-time risk level, executes the early warning and risk management strategies corresponding to the real-time risk level, and stores the end-to-end data in the cloud.

[0017] This invention provides a gas status early warning and monitoring method and system based on cloud-edge collaboration, which has the following beneficial effects: This invention combines filtering technology, environmental impact separation technology, and cloud-edge collaboration technology to address the impact of environmental fluctuations on monitoring data in residential settings. This solution eliminates sensor noise and environmental influences, and uses data from the same apartment type for calibration, resulting in more accurate data. A baseline is constructed using this more accurate data, and the flow velocity calibration baseline is obtained by summarizing the data. This ensures the accuracy of the subsequent risk assessment benchmark, thereby reducing the error rate of risk identification and achieving good results.

[0018] This invention employs a judgment technology that combines gas flow rate and usage status identification. This solves the problem of existing technologies relying solely on concentration thresholds without considering usage status, which can easily confuse fluctuations and leakage risks during normal use. This solution first determines the usage status and matches it with its own historical data and historical and real-time data from the same type of household. For scenarios where gas is in use, it focuses on monitoring the fluctuation range of gas flow rate. For scenarios where gas is not in use, it strictly controls the gas flow rate and concentration thresholds, thereby achieving gas supervision and significantly reducing the false alarm rate.

[0019] This invention employs a multi-assessment technique combined with cross-validation to address the shortcomings of existing solutions that use a single dimension for risk assessment, which are prone to underreporting and false alarms. This allows for accurate risk level assessment. The solution uses core assessment, fallback assessment, and false alarm assessment, and cross-validates data from the same apartment type to eliminate false alarms caused by environmental interference, thus achieving precise risk classification.

[0020] This invention employs hierarchical processing technology and scheme correction technology, which can solve the problems of existing schemes having a single early warning scheme, failing to help relevant personnel quickly understand the situation, and having an increasing false alarm rate over time. Different levels correspond to different early warning strategies and risk control strategies, which can help relevant personnel achieve accurate risk identification and control. Moreover, the scheme correction technology can optimize the entire scheme, enabling it to be used accurately for a long time, making the entire scheme highly applicable and effective.

[0021] This invention employs a full-link integration technology that combines cloud-edge collaboration, gas flow rate and usage status recognition, and multiple evaluation and scheme correction. This technology enables safe and efficient gas early warning monitoring in residential settings, meeting the monitoring needs of fragmented household use. Furthermore, the tiered early warning system lowers the barrier to entry for users, offering broad applicability. It provides a standardized technical framework for gas forecasting and regulation in the industry, laying the foundation for large-scale application. The invention demonstrates excellent performance and promising future prospects. Attached Figure Description

[0022] Figure 1 This is a flowchart of a gas status early warning and monitoring method based on cloud-edge collaboration according to the present invention; Figure 2 This is a flowchart illustrating the analysis of the flow velocity calibration baseline in a cloud-edge collaborative gas status early warning and monitoring method of the present invention. Figure 3 This is a structural block diagram of a gas status early warning and monitoring system based on cloud-edge collaboration according to the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] This invention primarily addresses the current state of household gas usage, directly using the gas concentration detected in the environment as the sole indicator for triggering early warnings. However, this approach is susceptible to environmental influences during use, leading to missed or false alarms.

[0025] Example 1: Please see Figure 1 and Figure 2 This embodiment provides a gas status early warning and monitoring method based on cloud-edge collaboration, including the following steps: S1. The cloud performs preprocessing on historical gas data and historical environmental data in the standard database to obtain historical concentration data, and analyzes it to obtain the flow rate calibration baseline.

[0026] This step mainly involves analyzing historical data and establishing a flow velocity calibration baseline based on that data.

[0027] Both historical and real-time gas data include methane concentration, gas flow rate, and usage status based on valve switch signals and flame sensor signals. Both historical and real-time environmental data include temperature, humidity, kitchen window open / close status, and exhaust fan operation logs.

[0028] In this step, the data processed in the cloud is the user's historical data. This step is carried out on a kitchen-by-kitchen basis, that is, data is collected and processed from the target kitchen.

[0029] The cloud-based preprocessing steps for historical gas data and historical environmental data include: Standardize the data format and unify the data units.

[0030] When using standardized data units, methane concentration is expressed as a percentage lower explosive limit, gas flow rate is expressed as cubic meters per hour, temperature is expressed as degrees Celsius, and humidity is expressed as a percentage relative humidity.

[0031] The percentage lower explosive limit is, for example, if the lower explosive limit of natural gas is known to be 5%, and the concentration of natural gas is detected to be 1%, then the concentration unit after standardizing the data units is 1% / 5%×100%=20%. The concentration unit conversion is mainly for the convenience of observing the degree of distance from explosion. In actual use, the volume percentage of natural gas in the air can also be used directly as the standardized data unit.

[0032] Perform outlier removal, deleting data containing instrument malfunction markers, data with a missing data rate greater than a set value, and data with contradictory usage status.

[0033] A data missing rate greater than the set value is generally defined as a data missing rate greater than 3%. Conflicting usage status data, such as the coexistence of usage and non-usage states (e.g., a valve is closed but gas flow is present), is also considered. This step primarily analyzes the normal state; therefore, conflicting usage status data will not be accepted.

[0034] The Kalman filter method is used to process the methane concentration and gas flow rate after outlier removal, so as to obtain the methane concentration and gas flow rate after noise interference is removed.

[0035] The Kalman filter method is used to remove high-frequency transient interference. The specific steps are as follows: First, the theoretical concentration and flow rate values ​​at the current moment are predicted using the state equation; second, the difference between the actual observation value and the predicted value is calculated; finally, the filter gain is dynamically adjusted based on the difference, and the predicted value is corrected using the adjusted gain, so as to finally obtain the historical concentration and flow rate data after removing transient interference.

[0036] An environmental impact separation method was used to process the methane concentration, gas flow rate, and usage status after removing noise interference, resulting in a corrected methane concentration.

[0037] The environmental impact separation method is used to eliminate systematic errors caused by temperature, humidity, and ventilation. First, a preset temperature and humidity influence coefficient table is called to calculate the methane concentration deviation caused by temperature and humidity. Specifically, the concentration deviation value = deviation coefficient corresponding to the current temperature × temperature deviation + deviation coefficient corresponding to the current humidity × humidity deviation. For example, for every 5°C increase in temperature, the concentration detection deviation increases by 0.1% of the lower explosive limit; for every 10% increase in humidity, the concentration detection deviation increases by 0.08% of the lower explosive limit. The historical concentration obtained after Kalman filtering is subtracted from this concentration deviation value to complete temperature and humidity compensation. Subsequently, based on the kitchen window opening / closing status records and exhaust fan operation logs, if no direct records are available, the methane concentration is determined through combustion... The ventilation status is derived by inversely from the airflow velocity fluctuation characteristics. When ventilation is enhanced, the gas is easily diluted, and the gas flow velocity will momentarily decrease. When ventilation is weakened, the gas is easily accumulated, and the gas flow velocity will momentarily increase. The methane concentration data is adjusted according to the ventilation status: When ventilation is enhanced, the methane concentration needs to be corrected to decrease, with the correction factor = 1 - ventilation enhancement factor. The ventilation enhancement factor is set according to the decrease in flow velocity. For example, if the gas flow velocity decreases by ≥20%, the ventilation enhancement factor is taken as 0.1. When ventilation is weakened, the methane concentration needs to be corrected to increase, with the correction factor = 1 + ventilation weakening factor. For example, if the gas flow velocity increases by ≥20%, the ventilation weakening factor is taken as 0.1. Finally, the preliminarily corrected methane concentration is obtained.

[0038] The corrected methane concentration, gas flow rate, and usage status are checked for matching degree and rationality using a logical consistency verification method. If both matching degree and rationality verification are qualified, the corrected methane concentration is determined to be historical concentration data. Otherwise, the data is determined to be contradictory, and the entire set of historical data is deleted.

[0039] When performing a matching degree verification, it is necessary to check whether the concentration during non-use periods after correction is close to the middle value of the preset standard range. At the same time, it is necessary to determine whether the concentration during use periods shows reasonable fluctuations with changes in gas flow rate. When performing a reasonableness verification, it is necessary to determine whether the methane concentration shows synchronous and corresponding fluctuations as the gas flow rate increases. After matching degree verification and reasonableness verification, logically contradictory data can be eliminated to obtain valid and true concentration data.

[0040] The steps for obtaining the flow velocity calibration baseline include: By associating historical concentration data, gas flow rate, and usage status during the corresponding time period, a three-dimensional associated dataset is formed. The three-dimensional data specifically consists of historical concentration data, gas flow rate, and usage status.

[0041] Based on the usage status, the state classification of the 3D associated dataset is performed, forming two sub-datasets.

[0042] The two types of subsets store the usage state subset and the non-use state subset respectively, thereby avoiding cross-interference between data from different scenarios.

[0043] The two types of subsets are divided according to the velocity range of the gas flow rate, and the core features of each velocity range are statistically analyzed.

[0044] For the non-use status subset, it is divided into a whole because its gas flow rate is less than 0.01 cubic meters per hour; for the use status subset, it is divided into 0.01-0.3 cubic meters per hour, 0.3-0.5 cubic meters per hour, 0.5-0.7 cubic meters per hour, 0.7-0.9 cubic meters per hour, etc.

[0045] For the used state subset, the core features of each flow velocity interval are the maximum, minimum, average, and standard deviation of the historical concentration data. For the unused state subset, the maximum and average values ​​are calculated.

[0046] The data on usage status, flow rate range, and core characteristics are used to form a baseline range, and the integrated data forms the flow rate calibration baseline.

[0047] Each flow rate interval corresponds to a baseline interval, and the unused state subset also corresponds to a baseline interval.

[0048] Historical concentration data of the same building, unit type, gas flow rate, and usage status were filtered, abnormal data were removed, and the data were aggregated to obtain a concentration reference range and a similar database was constructed.

[0049] Generally, 10-15 historical concentration data points from the same building, with the same apartment type and adjacent floors are selected. When the floor difference exceeds 8 floors, the environmental differences are significant, and the accuracy of the analysis will decrease.

[0050] For example, if there are 13 sets of data and 10 baseline intervals, then the concentration reference range includes 10 ranges, each of which is composed of 13 sets of baseline interval data.

[0051] This invention combines filtering technology, environmental impact separation technology, and cloud-edge collaboration technology to address the impact of environmental fluctuations on monitoring data in residential settings. This solution eliminates sensor noise and environmental influences, and uses data from the same apartment type for calibration, resulting in more accurate data. A baseline is constructed using this more accurate data, and the flow velocity calibration baseline is obtained by summarizing the data. This ensures the accuracy of the subsequent risk assessment benchmark, thereby reducing the error rate of risk identification and achieving good results.

[0052] S2. Real-time gas data and real-time environmental data are collected at the edge and preprocessed to obtain real-time concentration data. The flow rate calibration baseline is then called for screening and comparison to determine whether there is any suspicious data.

[0053] If no suspicious data is found, the data is directly packaged and uploaded to a standard database in the cloud for subsequent analysis.

[0054] The preprocessing steps are the same as those in S1.

[0055] The edge device connects to existing kitchen monitoring instruments to directly collect concentration, gas flow rate, and environmental data. It also receives data from three dimensions: valve switch signals, flame sensor signals, and gas flow rate. Through comprehensive analysis, it determines whether the device is in use or not.

[0056] The steps to determine whether suspicious data exists include: Acquire real-time concentration data, gas flow rate, usage status, and flow rate calibration baseline.

[0057] Real-time concentration data, gas flow rate, and usage status are acquired in real time, while the flow rate calibration baseline is determined based on the model of the acquisition instrument, and the corresponding flow rate calibration baseline can be directly retrieved.

[0058] The baseline interval corresponding to the current flow rate calibration baseline is determined based on the gas flow rate and the usage status.

[0059] Determine the usage status. If it is not in use, extract the preset concentration data and preset gas flow rate from the baseline range and compare them with the real-time concentration data and gas flow rate. If the real-time concentration data is less than the preset concentration data and the gas flow rate is less than the preset gas flow rate, it is determined that there is no suspicious data and it is stored in the standard database. Otherwise, it is determined that there is suspicious data and additional marking information is added.

[0060] The preset concentration data is generally set to 10% of the lower explosive limit, while the preset gas flow rate is 0.01 cubic meters per hour, because the gas flow rate is lower than 0.01 cubic meters per hour under normal non-use conditions.

[0061] If the data is in use, the fluctuation amplitude comparison method is used to determine whether there is any suspicious data.

[0062] The steps for determining whether there is suspicious data using the fluctuation amplitude comparison method include: Acquire continuously collected real-time concentration data and gas flow rate; calculate the real-time flow rate fluctuation amplitude using the continuous data segment extreme difference method.

[0063] The calculation method for real-time flow velocity fluctuation amplitude is as follows: extract the effective data segment, calculate the extreme value difference of the data segment, and convert it into a percentage. Specifically, when calculating the real-time flow velocity fluctuation amplitude, the extreme value difference method of continuous data segment is used to calculate the current fluctuation amplitude. First, extract the real-time gas flow velocity data within 30 seconds and filter out the effective data points; then, calculate the maximum and minimum values ​​of gas flow velocity within the effective data segment; finally, calculate the fluctuation amplitude using the formula: real-time flow velocity fluctuation amplitude = (maximum flow velocity - minimum flow velocity) / historical average flow velocity of the flow velocity range × 100%.

[0064] Extract the historical maximum permissible fluctuation range benchmark value from the baseline range; calculate the extreme standard amplitude threshold based on the historical maximum permissible fluctuation range benchmark value; the extreme standard amplitude threshold is preset to the historical maximum permissible fluctuation range benchmark value with an increase of 30%-50%.

[0065] The system compares the real-time flow velocity fluctuation amplitude with the limit standard amplitude threshold. If the real-time flow velocity fluctuation amplitude is greater than the limit standard amplitude threshold, it is determined that there is suspicious data and additional marking information is added. The marking information includes the current flow velocity range, baseline range, current fluctuation amplitude, and excess ratio. Otherwise, it is determined that there is no suspicious data and it is stored in the standard database for subsequent analysis.

[0066] This invention employs a judgment technology that combines gas flow rate and usage status identification. This solves the problem of existing technologies relying solely on concentration thresholds without considering usage status, which can easily confuse fluctuations and leakage risks during normal use. This solution first determines the usage status and matches it with its own historical data and historical and real-time data from the same type of household. For scenarios where gas is in use, it focuses on monitoring the fluctuation range of gas flow rate. For scenarios where gas is not in use, it strictly controls the gas flow rate and concentration thresholds, thereby achieving gas supervision and significantly reducing the false alarm rate.

[0067] S3. The edge device uploads suspicious data to the cloud, and the cloud performs analysis on the suspicious data to generate a concentration calibration range.

[0068] During daily use, the edge device periodically uploads suspicious data and additional tagging information to the cloud. When the cloud receives suspicious data, it performs analysis. During the analysis of suspicious data in the cloud, real-time concentration data and usage status are extracted from the suspicious data. Based on the real-time concentration data and usage status, data is filtered from similar databases and real-time aggregated data to obtain valid concentration data. Valid concentration data is data collected within 30 days under normal sensor conditions without data loss or other issues. Statistical analysis is performed on the valid concentration data to obtain the concentration calibration range and the distribution ratio of usage status.

[0069] The more effective concentration data available, the higher the hardware requirements. Generally, 20-30 days of effective concentration data is sufficient.

[0070] Statistical analysis involves calculating the mean, standard deviation, and confidence interval. The confidence interval is typically 95%. For example, if the concentration calibration range is a gas flow rate of 0.5 cubic meters per hour and the operating conditions are met, the 95% confidence interval for the concentration is 0.7%-2.1% of the lower explosive limit.

[0071] The distribution of the number of households in use is divided into 10-15 households.

[0072] S4. The edge end is combined with the flow rate calibration baseline and concentration calibration range, classified according to the usage status, and multiple assessments are carried out to obtain the risk level.

[0073] This step sets warning thresholds, specifically including concentration thresholds, concentration deviation thresholds, and flow rate fluctuation thresholds. The emergency risk concentration threshold is set at 60% of the lower explosion limit, the concentration deviation warning threshold for the same apartment type is ±30%, the flow rate fluctuation warning threshold during use is 50%, the gas flow rate threshold when not in use is 0.01 cubic meters per hour, and the concentration threshold when not in use is 10% of the lower explosion limit.

[0074] The steps for conducting multiple assessments include: Based on the usage status, an appropriate pre-judgment scheme is adopted to obtain the pre-judgment result.

[0075] When the system is determined to be in use, the baseline interval from the previous analysis is invoked. The baseline interval is merged with the concentration reference range, and the union is taken to obtain the concentration set interval. Then, the real-time concentration data is compared with the concentration set interval to determine whether the real-time concentration data is within the concentration set interval. If it is within the concentration set interval, it is determined to be normal. If it exceeds the interval but does not exceed the 30% lower explosive limit, it is determined to be suspicious for use. If it exceeds the 30% lower explosive limit, it is determined to be a high-risk suspicious leak.

[0076] In abnormal use conditions, the gas flow rate is not equal to zero and the concentration is greater than 10% of the lower explosive limit, which is judged as a suspected leak; when the gas flow rate is equal to zero and the concentration is less than or equal to 10% of the lower explosive limit, it is judged as normal idle.

[0077] The preliminary assessment results include normal use, normal idleness, suspicious use, suspected leakage, and high-risk suspected leakage.

[0078] A core assessment is performed on the preliminary judgment results, and the rationality of the data is verified by comprehensively considering the usage status, gas flow rate, and real-time concentration data to obtain an assessment result.

[0079] When verifying the reasonableness of the data, if it is pre-determined to be in normal use or normal idle state, then its own baseline is retrieved, and the current real-time concentration data and gas flow rate are compared to see if they are within the corresponding range. At the same time, the logical consistency between the use state and the gas flow rate is verified. If both of these conditions are met, it is determined to be risk-free; if one condition is not met, it is temporarily listed as a case requiring cross-validation.

[0080] If a suspected fluctuation is identified in advance, the correlation between gas flow rate fluctuation and concentration change is verified. Under normal circumstances, the fluctuation should show that the concentration increases in a reasonable manner when the gas flow rate increases, and the real-time concentration data is stable when the gas flow rate is stable. At the same time, the fluctuation is compared with the upper limit of the fluctuation range of its own baseline range. If the fluctuation pattern is consistent and does not exceed the upper limit of the fluctuation range of the baseline range, it is determined to be pending cross-verification; if the fluctuation pattern is abnormal, it is directly determined to be a preliminary abnormality.

[0081] If a leak is pre-determined to be a suspected leak or a high-risk suspected leak, priority should be given to verifying the match between the usage status and the leak characteristics. In non-usage scenarios, if the gas flow velocity is greater than 0.01m... 3 If the concentration continues to rise, it directly meets the characteristics of a leak. In the application scenario, if the real-time concentration data exceeds the baseline by 30% and there is no corresponding fluctuation in the gas flow rate, it meets the characteristics of an abnormal leak. At the same time, it is confirmed that the real-time concentration data has not reached the emergency threshold of 60% of the lower explosion limit. If the above conditions are met, it is judged as a preliminary anomaly. If the preset emergency threshold has been reached, the priority upgrade of the fallback assessment will be triggered directly.

[0082] The core assessment yielded three results: no risk, pending cross-validation, and preliminary anomaly. If there is no risk, no fallback assessment or false alarm assessment is required. If there is a preliminary anomaly, fallback assessment and false alarm assessment are performed sequentially. If cross-validation is pending, false alarm assessment is performed.

[0083] A fallback assessment is performed on the real-time concentration data to obtain a secondary assessment result.

[0084] The fallback assessment is mandatory, meaning that all assessments must be conducted. It analyzes real-time concentration data. If the real-time concentration data reaches the emergency threshold of 60% of the lower explosive limit, it is directly determined as an emergency risk. If the real-time concentration data does not reach the emergency threshold of 60% of the lower explosive limit, it is determined as not reaching the emergency threshold, and the result of the core assessment is adopted and marked as not reaching the emergency threshold.

[0085] The fallback assessment results in either an emergency risk or failure to reach the emergency threshold. If the emergency threshold is not reached, a false alarm assessment is performed.

[0086] False alarm assessments were performed on real-time concentration data, preliminary classification results, concentration calibration ranges, and usage status distribution ratios, resulting in three assessments.

[0087] False alarm assessment mainly relies on common data from users of the same apartment type to eliminate false alarms caused by regional environmental interference and fluctuations in common usage, thereby ensuring the accuracy of the final risk assessment.

[0088] During false alarm assessment, if the core assessment result is pending cross-validation, then the status of at least 60% of users with the same apartment type under the same conditions is statistically analyzed. If the status of most users is normal, it is determined to be a normal usage fluctuation and the suspicious label is removed.

[0089] For example: If a current user is marked as having suspicious fluctuations, and statistics show that at least 60% of users in the same apartment type are in normal condition under the same conditions, then the current user's fluctuations are determined to be normal usage fluctuations, and the suspicious flag is removed.

[0090] If the majority of users are in an abnormal state, it is determined to be a regional common fluctuation, marked as pending observation, without triggering an alert, and will be continuously monitored.

[0091] If statistics show that at least 60% of users with the same apartment type are in an abnormal state under the same conditions at the current time, it is determined to be a regional common fluctuation, such as the spread of oil fumes around the building or fluctuations in pipeline pressure, rather than a gas leak in a single household, and only continuous monitoring will be conducted without triggering an early warning.

[0092] If the core assessment result is a preliminary anomaly, the status of at least 60% of users with the same apartment type will be statistically analyzed. If the status of most users is normal, the determination of a real leak will be strengthened. If the status of most users is abnormal, it will be determined as a regional pipeline or environmental anomaly, upgraded to a regional risk candidate, and the data will be simultaneously fed back to the cloud for regional data aggregation and analysis.

[0093] The results of the false alarm assessment include normal use, regional common fluctuations, actual leakage, and regional risk candidates.

[0094] Summarize the results of the first, second, and third assessments to output the risk level.

[0095] In the fallback assessment, the emergency risk assessment has the highest priority. Once the assessment condition is triggered, it will be directly identified as an emergency risk level. At the same time, the integration of other assessment results will be skipped. The authenticity of the risk will be determined by the combination of the core assessment and the false alarm assessment results. This will avoid misjudgment caused by a single pre-judgment or a single assessment. In addition, the risk level assessment should be differentiated based on the usage status.

[0096] The risk level is divided into six levels, and the specific determination is as follows: Normal level: It needs to meet the following conditions: the pre-judgment result is normal use or normal idleness, the core assessment result is no risk, the fallback assessment result is that the emergency threshold has not been reached, and the false alarm assessment result is normal use fluctuation or regional common fluctuation.

[0097] For example: the preliminary assessment results show that the gas flow rate fluctuation is within the range of ≤±10%, and the real-time concentration data is within its own baseline range of 0.6%-1.7% lower explosion limit; the fallback assessment results show that the emergency threshold has not been reached; the false alarm elimination assessment shows that 75% of users in the same type of house are in normal condition. Based on the above assessment results, the final judgment is that it is at the normal level.

[0098] Suspicious fluctuation level: It needs to meet the following conditions: the pre-judgment result is suspicious usage fluctuation, the core assessment result is pending cross-validation, the fallback assessment result is not reached in the emergency threshold, and the false alarm assessment result is normal usage fluctuation or regional common fluctuation.

[0099] For example: the initial judgment showed suspicious usage fluctuations. Through core assessment, it was found that the gas flow rate fluctuations were related to the changes in real-time concentration data, but slightly exceeded its own baseline fluctuation limit. After eliminating false alarms, statistics showed that no less than 60% of users in the same type of house also had similar fluctuations at the same time and under the same conditions. Finally, it was determined to be a suspicious fluctuation, and no audible and visual alarms were triggered. Only a notification message was pushed.

[0100] Mild Leakage Level: Requires the following conditions to be met: the preliminary judgment result is suspected leakage, the core assessment result is preliminary anomaly, the fallback assessment result is that the emergency threshold has not been reached, the false alarm assessment result is a real leakage, and the real-time concentration data is ≤30% of the lower explosive limit. The 30% lower explosive limit is the set threshold for determining that the risk has exceeded the limit.

[0101] For example: When not in use, a preliminary assessment indicates a suspected leak. A core assessment reveals a gas flow rate of 0.03 m / s. 3 / h, the real-time concentration data is 12% lower explosive limit, the real-time concentration data exceeds the preset non-use state 10% lower explosive limit, the false alarm assessment shows that most users in the same type of house are normal, the bottom line is not reached 60% lower explosive limit, and it is finally determined to be a minor leak.

[0102] Moderate Leakage Level: It requires that the preliminary judgment result is a suspected leak or a high-risk suspected leak, the core assessment result is a preliminary anomaly, the fallback assessment result is that the emergency threshold has not been reached, the false alarm assessment result is a real leak, and the lower explosive limit is 30% < real-time concentration data ≤ 60% lower explosive limit, where 60% lower explosive limit is the set emergency threshold.

[0103] For example: In use, the initial assessment indicates a high-risk suspected leak. Through core assessment, the real-time concentration data is 45% of the lower explosive limit. It is found that the gas flow rate is not fluctuating. The false alarm assessment shows that most users in the same type of house are normal. The lower explosive limit is not reached. Finally, it is determined to be a moderate leak, triggering a strong audible and visual alarm and shutting off the solenoid valve.

[0104] Severe Leakage Level: The following conditions must be met: the preliminary judgment result is a suspected leak or a high-risk suspected leak; the core assessment result is a preliminary anomaly; the fallback assessment result is that the emergency threshold has not been reached; the false alarm assessment result is a real leak; and the lower explosive limit is 50% < real-time concentration data ≤ 60% of the lower explosive limit, or the real-time concentration data rises rapidly, for example, within 10 minutes, the real-time concentration data rises by more than 15% of the lower explosive limit.

[0105] Alternatively, the following conditions must be met: the preliminary judgment result is a suspected leak or a high-risk suspected leak; the core assessment result is a preliminary anomaly; the fallback assessment result is that the emergency threshold has not been reached; the false alarm assessment result is that the regional risk candidate is identified; and the user's real-time concentration data is close to 60% of the lower explosive limit.

[0106] For example: When not in use, if the real-time concentration data reaches 58% of the lower explosive limit, and rises from 30% of the lower explosive limit to 58% within 10 minutes, it is identified as a preliminary anomaly by the core assessment. After the false alarm elimination assessment, it is determined to be a real leak. Finally, it is determined to be a serious leak, which immediately triggers the double valve shut-off operation, issues a voice evacuation prompt, and coordinates with the gas company.

[0107] Emergency Risk Level: The following conditions must be met: the preliminary judgment result is a high-risk suspected leak, the core assessment result is a preliminary anomaly, the false alarm assessment result is a real leak, and the real-time concentration data rises rapidly, exceeding 60% of the lower explosive limit within 10 minutes.

[0108] Alternatively, a fallback assessment can be conducted to directly determine an emergency risk, i.e., if the real-time concentration data is greater than 60% of the lower explosive limit.

[0109] For example, regardless of the preliminary judgment and core assessment results, as long as the corrected concentration reaches 62% of the lower explosive limit, the fallback assessment will directly lock it as an emergency risk, skipping the false alarm elimination assessment stage, and immediately triggering the highest level of response measures, including double valve closure, shutting down surrounding electrical appliances, and coordinating with multiple parties to conduct emergency notifications.

[0110] The regional risk level is a special level because it does not belong to an individual user. Therefore, it is not within the scope of this assessment and needs to be judged again by the cloud in combination with data from multiple households: when the false alarm assessment result is a candidate for regional risk, and the cloud statistics show that ≥5 households of the same type in the same building are all initially abnormal under the same time and conditions, the cloud will trigger a regional risk warning and push a notification to all users in the region. The edge terminal of a single household will be judged as a serious leakage level.

[0111] When determining the risk level, the first step is to check whether the fallback assessment has triggered an emergency risk. If it has, the result is output directly. If it has not, the preliminary judgment result is cross-matched with the core assessment result to identify the categories of no risk, suspicious, or abnormal. Based on this, the verification results of the same apartment type in the false alarm elimination assessment are further combined to accurately match the specific level six risk. At the same time, the level determination is optimized by referring to details such as concentration thresholds and fluctuation trends, so as to ensure that the determination result is accurate and meets the safety requirements of a home environment.

[0112] This invention employs a multi-assessment technique combined with cross-validation to address the shortcomings of existing solutions that use a single dimension for risk assessment, which are prone to underreporting and false alarms. This allows for accurate risk level assessment. The solution uses core assessment, fallback assessment, and false alarm assessment, and cross-validates data from the same apartment type to eliminate false alarms caused by environmental interference, thus achieving precise risk classification.

[0113] S5, at the edge, executes corresponding early warning and risk control strategies based on the risk level, and periodically acquires real-time concentration data and execution data, performs multiple assessments to obtain the real-time risk level, executes the early warning and risk control strategies corresponding to the real-time risk level, and stores the end-to-end data in the cloud.

[0114] The warning strategies are as follows: Normal level: No action taken; Suspicious fluctuation level: A notification is sent via SMS or relevant software indicating a possible minor leak, and ventilation is advised; Minor leak level: Intermittent audible and visual alarms are activated, and a notification is sent via SMS or relevant software indicating a possible minor leak, and ventilation is advised; Moderate leak level: Intermittent audible and visual alarms are activated, with increased brightness and volume, and a notification is sent via SMS or relevant software indicating a non-use fluctuation, and the gas valve should be shut off; Severe leak level: Continuous audible and visual alarms are activated, with increased brightness and volume, and a notification is sent via SMS or relevant software indicating a gas leak, requiring immediate evacuation; Emergency risk level: Continuous audible and visual alarms are activated, with increased brightness and volume, and a faster alarm cycle, and a notification is sent via SMS or relevant software indicating a gas leak, requiring immediate evacuation, and simultaneous notification to property management personnel and the gas company.

[0115] The risk management strategy is divided into normal level and suspicious fluctuation level: Continuous high-frequency monitoring at the edge, and monitoring of the same apartment type in the cloud; if no sustained anomalies are detected within 20 minutes, the suspicious marker is automatically cleared, and the regular monitoring frequency is restored; Slight leakage level: Ventilation equipment such as exhaust fans is activated, and the dynamics of the same apartment type are monitored in the cloud; if the real-time concentration drops to less than or equal to 10% of the lower explosive limit within fifteen minutes, the audible and visual alarms are automatically deactivated, and regular monitoring is restored; if the real-time concentration data continues to rise, it is upgraded to a moderate leakage level, and the corresponding risk management strategy is implemented; Moderate leakage level: The gas solenoid valve is immediately triggered at the edge, the valve is closed, and ventilation equipment such as exhaust fans is activated; cloud recording... Handling log; continuously track real-time concentration data changes; if the real-time concentration data does not decrease, it is upgraded to a serious leak; serious leak level: the edge terminal activates the solenoid valve and gas meter to remotely close the valve, and continuously provides voice prompts for evacuation; turn off the power to electrical appliances around the kitchen; generate an emergency repair work order in the cloud and simultaneously dispatch the work order to the gas company's repair system; continuously track the real-time concentration data uploaded by the edge terminal and synchronize it with the repair personnel in real time; when there is regional risk, the cloud notifies the gas company to suspend gas supply to the pipeline network in that area; push emergency notifications to all users in the area; edge terminal: execute valve closure, sound and light alarms, and voice prompts according to the serious leak level; After storing the end-to-end data in the cloud, the data is divided into segments according to data characteristics. Continuous trend analysis is performed on continuous data of usage status and high-flow-rate stable data segments, and targeted supplementary analysis is performed on fragmented usage gaps and low-flow-rate high-random data segments. Based on the results of continuous trend analysis and targeted supplementary analysis, the flow rate calibration baseline is corrected and the environmental impact separation method is optimized with real-time concentration data.

[0116] High flow velocity refers to a gas flow velocity between 0.3 and 1.0 m / s. 3 In the high flow rate range of / h, the continuous data duration is ≥30 seconds and the flow rate fluctuation is ≤±10%; there is no significant environmental interference, and the continuous data and high flow rate stable data segments during use are included, such as stable combustion data during continuous cooking during dinner cooking and continuous gas supply data during long-term soup simmering.

[0117] Low flow rate refers to a gas flow rate between 0 and 0.10 m / s. 3 In the low flow rate range of / h, the high random data segments are characterized by fragmented data distribution, with a single data segment duration of less than 10 seconds and gas flow rate fluctuation greater than or equal to ±15%, indicating significant environmental interference. Fragmented usage intervals and low flow rate high random data segments include instantaneous residual gas flow rate data after turning off the stove during cooking intervals, small fluctuation data from the sensor during non-use periods, and instantaneous concentration dilution data caused by opening windows for ventilation.

[0118] The specific steps for continuous trend analysis of continuous data and high-flow-rate stable data segments under usage conditions are as follows: First, data cleaning is carried out, that is, instantaneous extreme values ​​are removed from the data segments, and only continuous and stable data sequences are retained; Second, pattern extraction is performed, using gas flow rate as the abscissa and concentration as the ordinate to draw a correlation curve, and the mean concentration, standard deviation, and 95% confidence interval are calculated for different flow rate intervals. A stable trend equation for a synchronous and reasonable increase in concentration when the gas flow rate increases is extracted. This equation is a linear regression equation; Finally, deviation analysis is performed, comparing the extracted trend pattern with the currently used flow rate calibration baseline, and calculating the deviation value between the two in each flow rate interval. The deviation value is calculated by subtracting the mean value of the corresponding interval of the baseline from the mean concentration obtained from the analysis. Flow rate intervals with deviations exceeding 5% are marked as key areas for subsequent baseline correction.

[0119] The specific steps for targeted supplementary analysis of fragmented data segments with gaps in usage and low flow rates and high randomness are as follows: First, data classification is carried out, further subdividing the fragmented data according to the type of environmental interference to form subsets; second, interference impact quantification is implemented, calculating the concentration deviation under different interference conditions for each subset, comparing the concentration fluctuations in different temperature and humidity ranges, and optimizing the temperature and humidity impact coefficient table; finally, data completion and pattern fitting are performed, using the adjacent segment trend extension method to complete the missing parts in the fragmented data, constructing a complete four-dimensional sequence of time, flow rate, concentration, and environmental interference. At the same time, the concentration fluctuation pattern in the low flow rate range is fitted to clarify the normal fluctuation range of this range, providing a reference for the correction of the low flow rate range of the baseline.

[0120] The correction of the flow velocity calibration baseline is based on the correlation between flow velocity and concentration obtained through continuous trend analysis. This is combined with the normal fluctuation range of the low flow velocity range obtained from targeted supplementary analysis, and reference to the concentration distribution data of the same flow velocity range for the same apartment type. For the high flow velocity range, the average concentration and 95% confidence interval of each flow velocity range obtained from continuous trend analysis are used to replace the values ​​of the corresponding intervals in the original baseline. For the low flow velocity range, the low flow velocity threshold of the original baseline is adjusted according to the normal fluctuation range obtained from supplementary analysis. Then, the corrected baseline is compared with real-time concentration data to ensure that the deviation between the predicted concentration value and the real-time concentration data for each flow velocity range does not exceed 3%. After verification, the final corrected flow velocity calibration baseline is determined.

[0121] The optimization of the environmental impact separation method involves optimizing the parameters of the Kalman filter and the environmental impact separation method based on the environmental interference quantification results obtained from the supplementary analysis, thereby improving the accuracy of environmental error correction. Specifically: Kalman filter parameter optimization: Given the high randomness of fragmented low-velocity data, the filter parameters are adjusted; Environmental impact coefficient table update: The temperature, humidity, and ventilation interference quantification coefficients obtained from the supplementary analysis are updated to the original environmental impact coefficient table; Ventilation correction logic optimization: Based on the analysis of the impact of ventilation switching on low-velocity data, the ventilation status judgment threshold is optimized.

[0122] After optimization, the corrected flow velocity calibration baseline and the updated environmental impact separation method parameters are integrated to generate an optimized parameter list. The cloud will push this optimized parameter list to the corresponding edge device. After receiving it, the edge device will perform integrity verification, replace the original local data with the optimized content, and then update the local operating system. The edge device will use the optimized baseline and method to process the reset real-time data, verify the accuracy of the corrected data, and ensure that the optimization effect is implemented, thereby completing the iterative closed loop.

[0123] This invention employs hierarchical processing technology and scheme correction technology, which can solve the problems of existing schemes having a single early warning scheme, failing to help relevant personnel quickly understand the situation, and having an increasing rate of false alarms and missed alarms over time. Different levels correspond to different early warning strategies and risk control strategies, which can help relevant personnel achieve accurate risk identification and control. Moreover, the scheme correction technology can optimize the entire scheme, making the entire scheme more accurate and effective.

[0124] This invention employs a full-link integration technology that combines cloud-edge collaboration, gas flow rate and usage status recognition, and multiple evaluation and scheme correction. This technology enables safe and efficient gas early warning monitoring in residential settings, meeting the monitoring needs of fragmented household use. Furthermore, the tiered early warning system lowers the barrier to entry for users, offering broad applicability. It provides a standardized technical framework for gas forecasting and regulation in the industry, laying the foundation for large-scale application. The invention demonstrates excellent performance and promising future prospects.

[0125] Example 2: Based on Example 1, such as Figure 3 As shown, this embodiment also provides a gas status early warning and monitoring system based on cloud-edge collaboration, including a historical analysis module, a real-time analysis module, an anomaly identification module, a risk identification module, and a risk control module: Historical analysis module: The cloud performs preprocessing on historical gas data and historical environmental data in the standard database to obtain historical concentration data, and analyzes it to obtain the flow rate calibration baseline.

[0126] Real-time analysis module: Collects real-time gas data and real-time environmental data at the edge, performs preprocessing to obtain real-time concentration data, and calls the flow rate calibration baseline for screening and comparison to determine whether there is any suspicious data.

[0127] Anomaly detection module: The edge device uploads suspicious data to the cloud, and the cloud performs analysis on the suspicious data to generate a concentration calibration range.

[0128] Risk identification module: The edge end combines the flow rate calibration baseline and concentration calibration range, classifies according to usage status, performs multiple assessments, and obtains the risk level.

[0129] Risk management module: The edge device executes corresponding early warning and risk management strategies based on the risk level, and periodically obtains real-time concentration data and execution data, performs multiple assessments to obtain the real-time risk level, executes the early warning and risk management strategies corresponding to the real-time risk level, and stores the end-to-end data in the cloud.

[0130] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0131] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0132] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A gas status early warning and monitoring method based on cloud-edge collaboration, characterized in that, Includes the following steps: The cloud performs preprocessing on historical gas data and historical environmental data in the standard database to obtain historical concentration data, and analyzes it to obtain the flow rate calibration baseline; Real-time gas data and real-time environmental data are collected at the edge and preprocessed to obtain real-time concentration data. The flow rate calibration baseline is then used for screening and comparison to determine whether there is any suspicious data. The edge device uploads suspicious data to the cloud, and the cloud performs analysis on the suspicious data to generate a concentration calibration range. The risk level is determined by combining the flow rate calibration baseline and concentration calibration range at the edge end, classifying them according to their usage status, and conducting multiple assessments. The edge device executes corresponding early warning and risk control strategies based on the risk level, and regularly acquires real-time concentration data and execution data, performs multiple assessments to obtain the real-time risk level, executes the early warning and risk control strategies corresponding to the real-time risk level, and stores the end-to-end data in the cloud.

2. The cloud-edge collaboration based gas state early warning monitoring method according to claim 1, characterized in that: Both historical and real-time gas data include methane concentration, gas flow rate, and usage status based on valve switch signals and flame sensor signals. Both historical and real-time environmental data include temperature, humidity, kitchen window open / close status, and exhaust fan operation logs.

3. The cloud-edge collaborative gas state early warning monitoring method according to claim 2, characterized in that: The cloud-based preprocessing steps for historical gas data and historical environmental data include: Standardize data formats and unify data units; Perform outlier removal, deleting data containing fault markers for testing instruments, data with a missing data rate greater than a set value for a given period, and data with contradictory usage status. The Kalman filter method is used to process the methane concentration and gas flow rate after outlier removal, so as to obtain the methane concentration and gas flow rate after noise interference is removed. The methane concentration was adjusted by removing noise interference and considering the gas flow rate and operating status using an environmental impact separation method to obtain the corrected methane concentration. The corrected methane concentration, gas flow rate, and usage status are checked for matching degree and rationality using a logical consistency verification method. If both matching degree and rationality verification are qualified, the corrected methane concentration is determined to be historical concentration data. Otherwise, the data is determined to be contradictory, and the entire set of historical data is deleted.

4. The cloud-edge collaboration based gas state early warning monitoring method according to claim 1, characterized in that: The steps for obtaining the flow velocity calibration baseline include: Historical concentration data, gas flow rate, and usage status during corresponding time periods are correlated and summarized to form a three-dimensional correlated dataset. Based on the usage status, state classification is performed on the 3D associated dataset, forming two sub-datasets; The two types of subsets are divided according to the velocity range of the gas flow rate, and the core features of each velocity range are statistically analyzed. The statistical usage status, flow velocity range, and core characteristics are used to form a baseline range, and the integrated and summarized data are used to form a flow velocity calibration baseline. Historical concentration data of the same building, unit type, gas flow rate, and usage status were filtered, abnormal data were removed, and the data were aggregated to obtain a concentration reference range and a similar database was constructed.

5. The cloud-edge collaboration based gas state early warning monitoring method according to claim 4, characterized in that: The steps to determine whether suspicious data exists include: Acquire real-time concentration data, gas flow rate, usage status, and flow rate calibration baseline; The baseline interval corresponding to the current flow velocity calibration baseline is determined based on the gas flow velocity and the usage status. Determine the usage status. If it is not in use, extract the preset concentration data and preset gas flow rate in the baseline interval and compare them with the real-time concentration data and gas flow rate. If the real-time concentration data is less than the preset concentration data and the gas flow rate is less than the preset gas flow rate, it is determined that there is no suspicious data and it is stored in the standard database. Otherwise, it is determined that there is suspicious data and additional marking information is added. If the data is in use, the fluctuation amplitude comparison method is used to determine whether there is any suspicious data.

6. The gas status early warning and monitoring method based on cloud-edge collaboration according to claim 5, characterized in that: The steps for determining whether there is suspicious data using the fluctuation amplitude comparison method include: Acquire continuously collected real-time concentration data and gas flow rate; The extreme difference method of continuous data segments is used to calculate the real-time flow velocity fluctuation amplitude. Extract the historical maximum permissible fluctuation range benchmark value from the baseline range; The extreme standard amplitude threshold is calculated based on the historical maximum permissible fluctuation range benchmark value; The real-time flow rate fluctuation amplitude is compared with the limit standard amplitude threshold. If the real-time flow rate fluctuation amplitude is greater than the limit standard amplitude threshold, it is determined that there is suspicious data and a label is added. Otherwise, it is determined that there is no suspicious data and it is stored in the standard database.

7. The gas status early warning and monitoring method based on cloud-edge collaboration according to claim 4, characterized in that: When performing analysis on suspicious data in the cloud, real-time concentration data and usage status are extracted from the suspicious data. Based on the real-time concentration data and usage status, data is filtered from similar databases and real-time aggregated data to obtain valid concentration data. Statistical analysis is performed on the valid concentration data to obtain the concentration calibration range and the distribution ratio of usage status.

8. The gas status early warning and monitoring method based on cloud-edge collaboration according to claim 7, characterized in that: The steps for conducting multiple assessments include: Based on the usage status, an appropriate pre-judgment scheme is adopted to obtain the pre-judgment result; A core assessment is performed on the preliminary judgment results, and the rationality of the data is verified by comprehensively considering the usage status, gas flow rate, and real-time concentration data to obtain an assessment result. A fallback assessment is performed on the real-time concentration data to obtain a secondary assessment result; False alarm assessments were performed on real-time concentration data, preliminary classification results, concentration calibration ranges, and usage status distribution ratios, resulting in three assessments. Summarize the results of the first, second, and third assessments to output the risk level.

9. A gas status early warning and monitoring method based on cloud-edge collaboration according to claim 8, characterized in that: After storing the end-to-end data in the cloud, the data is divided into segments according to data characteristics. Continuous trend analysis is performed on continuous data of usage status and high-flow-rate stable data segments, and targeted supplementary analysis is performed on fragmented usage gaps and low-flow-rate high-random data segments. Based on the results of continuous trend analysis and targeted supplementary analysis, the flow rate calibration baseline is corrected and the environmental impact separation method is optimized with real-time concentration data.

10. A cloud-edge collaboration based gas state early warning monitoring system, characterized in that, include: Historical analysis module: The cloud performs preprocessing on historical gas data and historical environmental data in the standard database to obtain historical concentration data, and analyzes it to obtain the flow rate calibration baseline; Real-time analysis module: Collects real-time gas data and real-time environmental data at the edge, performs preprocessing to obtain real-time concentration data, and calls the flow rate calibration baseline for screening and comparison to determine whether there is any suspicious data; Anomaly detection module: The edge device uploads suspicious data to the cloud, and the cloud performs analysis on the suspicious data to generate a concentration calibration range; Risk identification module: The edge end combines the flow rate calibration baseline and concentration calibration range, classifies according to usage status, performs multiple assessments, and obtains the risk level; Risk management module: The edge device executes corresponding early warning and risk management strategies based on the risk level, and periodically obtains real-time concentration data and execution data, performs multiple assessments to obtain the real-time risk level, executes the early warning and risk management strategies corresponding to the real-time risk level, and stores the end-to-end data in the cloud.

Images