A safety management and early warning system for energy operation sites

Abstract

The present application belongs to the technical field of energy operation safety management, and provides a kind of safety management and early warning system of energy operation site, comprising: the historical temperature-distortion rate of frequency is calculated by sensor acquisition frequency and monitoring system receives frequency, temperature range-distortion rate range set is obtained by clustering, correlation is analyzed based on pearson correlation coefficient method, linear or nonlinear influence model is constructed, the distortion rate range corresponding to current temperature is output and stability is judged by fluctuation variance, whether it causes vacuum under stable state is judged and is compensated by increasing acquisition frequency, sensor failure is judged using LSTM time series prediction model under unstable state;The present application does not need to add hardware, adapts to multiple energy operation scenarios, and guarantees monitoring continuity and reliability.

Description

Technical Field

[0001] This invention belongs to the field of energy operation safety management technology, specifically a safety management and early warning system for energy operation sites. Background Technology

[0002] Safety management at energy operation sites (including but not limited to underground coal mines, LNG storage tank areas, and containerized energy storage) relies primarily on , , Real-time acquisition and uploading of data from key parameter sensors, and the timeliness and continuity of data transmission are prerequisites for avoiding safety risks such as leakage of toxic and harmful gases and equipment failure.

[0003] However, extreme temperature environments are common at energy operation sites, which severely interfere with the sensor data transmission link, directly causing the sensor data to fail to be reported in a timely manner, creating a safety monitoring vacuum, and becoming a core technical challenge that restricts the safety management of energy operation sites.

[0004] Specifically, temperature directly affects data transmission in industrial Ethernet cables: in high-temperature environments, the cable insulation layer ages faster and its performance degrades, leading to increased signal transmission loss and even cable short circuits; in low-temperature environments, the cable material hardens and its flexibility decreases, making it prone to loose interfaces and poor contact, significantly increasing the data transmission error rate.

[0005] In existing technologies, traditional industrial Ethernet cables lack designs adapted to extreme temperatures at energy work sites. Their high and low temperature resistance is insufficient to meet the needs of the scenario, and they cannot fundamentally solve the problem of data transmission interruption caused by temperature. This leads to repeated monitoring vacuums, which seriously hinders the reliability and effectiveness of safety management at energy work sites. Therefore, there is an urgent need for a technical solution that can overcome the impact of extreme temperatures on the transmission link and avoid monitoring vacuums, so as to solve this key technical problem in safety management in the energy industry.

[0006] Therefore, the present invention provides a safety management and early warning system for energy operation sites. Summary of the Invention

[0007] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0008] The technical solution adopted by this invention to solve its technical problem is:

[0009] One of the objectives of this invention is to provide a safety management and early warning system for energy work sites, comprising:

[0010] Historical data processing and clustering module: acquires historical temperature-monitoring distortion rate data, performs cluster analysis on historical temperature-monitoring distortion rate, and obtains a set of temperature range-monitoring distortion rate ranges;

[0011] Correlation Analysis and Model Building Module: Based on the temperature range-monitoring distortion rate range set, perform correlation analysis on the temperature range and the monitoring distortion rate range, and build an influence model of temperature range-monitoring distortion rate range based on the correlation analysis results;

[0012] Real-time distortion rate output and stability analysis module: Based on the temperature range-monitoring distortion rate range influence model, it outputs the monitoring distortion rate range corresponding to the current temperature range and performs stability analysis on the current monitoring distortion rate range;

[0013] Vacuum compensation module: If the current monitoring distortion rate range is stable, it determines whether the current monitoring distortion rate range will cause a monitoring vacuum. If it will cause a monitoring vacuum, it increases the sensor acquisition frequency to perform monitoring vacuum compensation. If it will not cause a monitoring vacuum, it continuously monitors the current monitoring distortion rate range. If the current monitoring distortion rate range is unstable, it uses a time-series prediction model to determine whether the sensor is in a faulty state.

[0014] As a further improvement of the present invention, the specific process of obtaining historical temperature-monitoring distortion rate data is as follows:

[0015] The sensor acquires data at the corresponding temperature over a historical period, and the monitoring system receives the sensor data at the same frequency. The difference between the sensor's acquisition frequency and the monitoring system's receiving frequency is divided by the sensor's acquisition frequency to calculate the sensor's monitoring distortion rate at the corresponding temperature, thus obtaining historical temperature-monitoring distortion rate data.

[0016] As a further improvement of the present invention, the specific process of obtaining the temperature range-distortion rate range set is as follows:

[0017] Based on historical temperature-monitoring distortion rate data, a temperature-monitoring distortion rate sample dataset was compiled. Each sample contains the temperature value corresponding to the sensor and the monitoring distortion rate at that temperature. Clustering algorithms were then used to perform cluster analysis on the temperature-monitoring distortion rate sample dataset.

[0018] Using temperature value as the first feature dimension and monitoring distortion rate as the second feature dimension, samples with similar temperature values ​​and similar monitoring distortion rates are grouped into the same cluster. The continuous temperature range and monitoring distortion rate range corresponding to the cluster results are determined. The temperature range-monitoring distortion rate range corresponding to all cluster categories are summarized to obtain the temperature range-monitoring distortion rate range set.

[0019] As a further improvement of the present invention, the specific process of performing correlation analysis on the temperature range and the monitoring distortion rate range is as follows:

[0020] For any set of temperature range-monitoring distortion rate range data, extract all temperature value-monitoring distortion rate sample values, with temperature value as independent variable and the corresponding monitoring distortion rate value as dependent variable, and calculate the correlation coefficient using the Pearson correlation coefficient method.

[0021] If the correlation coefficient is greater than or equal to the correlation coefficient threshold, it is determined that the temperature value corresponding to this group has a linear correlation with the range of monitored distortion rate values.

[0022] If the correlation coefficient is less than the correlation coefficient threshold, it is determined that the temperature value and the range of monitored distortion rate values ​​corresponding to this group are not linearly related.

[0023] As a further improvement of the present invention, the specific process of constructing the temperature range-monitoring distortion rate range influence model is as follows:

[0024] If it is determined that there is a linear correlation between the temperature value and the monitoring distortion rate value, then a univariate linear regression algorithm is used to linearly fit the independent variable in the temperature range and the dependent variable in the distortion rate range to construct a temperature range-monitoring distortion rate range influence model.

[0025] If it is determined that the temperature range and the monitoring distortion rate range are not linearly related, then a multinomial regression algorithm is used to perform nonlinear fitting on the independent variable in the temperature range and the dependent variable in the distortion rate range to construct a temperature range-monitoring distortion rate range influence model.

[0026] As a further improvement of the present invention, the specific process of performing stability analysis on the current monitoring distortion rate is as follows:

[0027] Obtain the current real-time temperature data, determine the temperature range to which the current real-time temperature data belongs, call the temperature range-monitoring distortion rate range influence model corresponding to the temperature range to which the current real-time temperature data belongs, and output the monitoring distortion rate range corresponding to the current temperature range based on the temperature range-monitoring distortion rate range influence model.

[0028] As a further improvement of the present invention, the specific process of performing stability analysis on the current monitoring distortion rate range is as follows:

[0029] The monitoring distortion rate range corresponding to the current real-time temperature data is continuously collected. A coverage window is preset, and the average value of the monitoring distortion rate is taken in each coverage window. The average value of the monitoring distortion rate corresponding to 10 coverage windows is collected in total.

[0030] Calculate the variance of the average monitoring distortion rate across 10 coverage windows;

[0031] If the fluctuation variance is less than the fluctuation variance threshold, the current monitored distortion rate range is determined to be in a stable state; if the fluctuation variance is greater than or equal to the fluctuation variance threshold, the current monitored distortion rate range is determined to be in an unstable state, and a fluctuation warning signal is output.

[0032] As a further improvement of the present invention, the specific process of determining whether the current monitoring distortion rate range will cause a monitoring vacuum is as follows:

[0033] Extract the current stable monitoring distortion rate range and analyze its numerical relationship with the monitoring vacuum threshold:

[0034] If the lower limit of the current stable monitoring distortion rate range is greater than or equal to the monitoring vacuum threshold, it is determined that a monitoring vacuum will be caused, and a monitoring vacuum warning signal is output.

[0035] If the upper limit of the current stable monitoring distortion rate range is less than the monitoring vacuum threshold, it is determined that no monitoring vacuum will be caused.

[0036] If the current stable monitoring distortion rate range exceeds the monitoring vacuum threshold, it is determined that a monitoring vacuum will be caused, and a monitoring vacuum warning signal will be output.

[0037] As a further improvement of the present invention, the specific process of monitoring vacuum compensation is as follows:

[0038] Extract the average value of the monitoring distortion rate that is currently in a stable state, the current raw acquisition frequency of the sensor, and the current actual receiving frequency of the sensor data by the monitoring system;

[0039] With the goal of restoring the actual receiving frequency of the monitoring system to the original acquisition frequency level after increasing the acquisition frequency, the target acquisition frequency to be adjusted is calculated.

[0040] Determine in advance the maximum sampling frequency supported by the sensor hardware;

[0041] If the calculated target acquisition frequency does not exceed the maximum acquisition frequency of the hardware, then the sensor acquisition frequency is adjusted to the target frequency.

[0042] As a further improvement of the present invention, the specific process of determining whether the sensor is in a fault state through the time-series prediction model is as follows:

[0043] Extract the historical time-series data of the sensor, including: monitoring distortion rate time-series data, temperature time-series data, and sensor operating status time-series data under the same scene and temperature range in the past 30 days;

[0044] Using the current moment as the endpoint, capture the monitoring distortion rate time series data and temperature fluctuation time series data within the last 10 minutes, and integrate them with the historical time series data after unifying the time granularity to form the input dataset for the time series prediction model.

[0045] A time series prediction model was constructed using a Long Short-Term Memory (LSTM) network. Temperature time series data and sensor operating status time series data were used as input features, and monitoring distortion rate time series data were used as output labels. The time series prediction model was trained to obtain the model.

[0046] Input the current time series data into the time series prediction model and output the monitoring distortion rate prediction trend curve for the next 5 minutes. If the monitoring distortion rate prediction trend curve satisfies the condition that the fluctuation amplitude is greater than or equal to the distortion rate fluctuation amplitude threshold and the predicted value is greater than or equal to the deviation from the normal range threshold, then the sensor is determined to be in a fault state and a sensor fault warning signal is output.

[0047] A second objective of this invention is to provide a method for safety management and early warning at energy work sites, comprising the following steps:

[0048] S10: Obtain historical temperature-monitoring distortion rate data, perform cluster analysis on historical temperature-monitoring distortion rate, and obtain a set of temperature range-monitoring distortion rate ranges;

[0049] S20: Based on the temperature range-monitoring distortion rate range set, conduct a correlation analysis on the temperature range and the monitoring distortion rate range, and construct an influence model of temperature range-monitoring distortion rate range based on the correlation analysis results;

[0050] S30: Based on the temperature range-monitoring distortion rate range influence model, output the monitoring distortion rate range corresponding to the current temperature range, and perform stability analysis on the current monitoring distortion rate range;

[0051] S40: If the current monitoring distortion rate range is stable, determine whether the current monitoring distortion rate range will cause a monitoring vacuum. If it will cause a monitoring vacuum, increase the sensor acquisition frequency and perform monitoring vacuum compensation. If it will not cause a monitoring vacuum, continuously monitor the current monitoring distortion rate range. If the current monitoring distortion rate range is unstable, determine whether the sensor is in a fault state through a time-series prediction model.

[0052] The beneficial effects of this invention are as follows:

[0053] 1. To address the issue of sensor data transmission distortion and monitoring vacuum caused by extreme temperatures at energy operation sites, a closed-loop process of historical data clustering, correlation modeling, real-time analysis, and dynamic compensation is used to accurately predict the monitoring distortion rate at different temperatures. In stable conditions, the vacuum is compensated by increasing the acquisition frequency, while in unstable conditions, an LSTM model is used to identify sensor faults. This approach ensures the continuity of data transmission from the root cause, eliminates blind spots in safety monitoring, and the constructed model library can output distortion rates specifically, avoiding the accuracy loss of general solutions and having a wide range of applications.

[0054] 2. Optimization is achieved through software-level data analysis (clustering, regression), parameter adjustment (acquisition frequency, coverage window), and auxiliary measures (data compression, priority transmission). No modification to sensor hardware or addition of transmission equipment is required, significantly reducing the cost and complexity of on-site modifications in energy operations. It aligns with practical needs in industrial scenarios, balancing real-time performance and reliability, and improving safety management efficiency: Shortening the coverage window enhances monitoring timeliness, using fluctuation variance to quantify stability, and combining multi-level early warnings to trigger timely responses; continuous monitoring of the receiving frequency after compensation ensures the compensation effect, significantly improving the response speed and reliability of safety management at energy operation sites. Attached Figure Description

[0055] The invention will now be further described with reference to the accompanying drawings.

[0056] Figure 1 This is a system module diagram of a safety management and early warning system for energy operation sites according to the present invention;

[0057] Figure 2 This is a flowchart illustrating the steps of a safety management and early warning method for energy operation sites according to the present invention. Detailed Implementation

[0058] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0059] Example 1

[0060] like Figure 1 As shown in the embodiment of the present invention, a safety management and early warning system for energy operation sites includes:

[0061] Historical data processing and clustering module: acquires historical temperature-monitoring distortion rate data, performs cluster analysis on historical temperature-monitoring distortion rate, and obtains a set of temperature range-monitoring distortion rate ranges;

[0062] In the historical data processing and clustering module, the specific process for obtaining historical temperature-monitoring distortion rate data is as follows:

[0063] From historical energy operation site safety management scenarios (such scenarios rely on) , , Sensors are used for safety monitoring, specifically in scenarios including but not limited to underground coal mines, LNG storage tank areas, and containerized energy storage. Data is acquired from these sensors (specifically including but not limited to those used in these scenarios). , , The sensor is used to obtain the acquisition frequency of the sensor under the corresponding temperature environment in the historical period, and the receiving frequency of the monitoring system for the sensor data; the monitoring distortion rate of the sensor under the corresponding temperature environment is calculated by dividing the difference between the acquisition frequency of the sensor and the receiving frequency of the monitoring system by the acquisition frequency of the sensor, and the historical temperature-monitoring distortion rate data is obtained.

[0064] For example, the calculation process of the monitoring distortion rate is as follows: (the sensor's acquisition frequency - the monitoring system's receiving frequency) ÷ the sensor's acquisition frequency = the sensor's monitoring distortion rate under the corresponding temperature environment;

[0065] In the historical data processing and clustering module, the specific process of performing cluster analysis on historical temperature-monitoring distortion rate to obtain the temperature range-distortion rate range set is as follows:

[0066] Based on the aforementioned historical temperature-monitoring distortion rate data, a temperature-monitoring distortion rate sample dataset is formed, where each sample contains the temperature value corresponding to the sensor and the monitoring distortion rate at that temperature; clustering algorithms (such as K-means clustering algorithm) are used to perform cluster analysis on the temperature-monitoring distortion rate sample dataset:

[0067] Using the temperature value of the samples as the first feature dimension and the corresponding monitoring distortion rate as the second feature dimension, samples with similar temperature values ​​and similar monitoring distortion rates are grouped into the same category. For each clustering result, the continuous temperature range corresponding to that clustering result is determined, and the range of monitoring distortion rates corresponding to all samples within that clustering result is statistically analyzed, thus giving each clustering result a set of temperature range-monitoring distortion rate ranges. The temperature range-monitoring distortion rate ranges corresponding to all clustering categories are summarized to obtain the set of temperature range-monitoring distortion rate ranges.

[0068] Correlation Analysis and Model Building Module: Based on the temperature range-monitoring distortion rate range set, perform correlation analysis on the temperature range and the monitoring distortion rate range, and build an influence model of temperature range-monitoring distortion rate range based on the correlation analysis results;

[0069] In the correlation analysis and model building module, the specific process of performing correlation analysis on the temperature range and the monitoring distortion rate range based on the temperature range-monitoring distortion rate range set is as follows:

[0070] For any set of temperature range-monitoring distortion rate range data in the temperature range-monitoring distortion rate range set, extract all temperature value-monitoring distortion rate sample values ​​(i.e., each temperature value in the temperature range and the corresponding monitoring distortion rate value at that temperature value).

[0071] Using the temperature value within the group as the independent variable and the monitoring distortion rate corresponding to the independent variable as the dependent variable, the Pearson correlation coefficient method is used to calculate the linear correlation coefficient between the independent and dependent variables.

[0072] Preset the correlation coefficient threshold (e.g., 0.7, which can be adjusted according to the actual scenario);

[0073] If the calculated correlation coefficient is greater than or equal to the correlation coefficient threshold, it is determined that the temperature value corresponding to this group has a linear correlation with the range of monitored distortion rate values.

[0074] If the calculated correlation coefficient is less than the correlation coefficient threshold, it is determined that the temperature value and the monitoring distortion rate range corresponding to this group are not linearly related.

[0075] In the correlation analysis and model building module, based on the correlation analysis results, the specific process of building the temperature range-monitoring distortion rate range influence model is as follows:

[0076] If it is determined that there is a linear correlation between the temperature value and the monitoring distortion rate value, then a univariate linear regression algorithm is used to linearly fit the independent variable in the temperature range and the dependent variable in the distortion rate range to construct a temperature range-monitoring distortion rate range influence model.

[0077] If it is determined that the temperature range and the monitoring distortion rate range are not linearly related, then a multinomial regression algorithm is used to perform nonlinear fitting on the independent variable in the temperature range and the dependent variable in the distortion rate range to construct a temperature range-monitoring distortion rate range influence model.

[0078] Real-time distortion rate output and stability analysis module: Based on the temperature range-monitoring distortion rate range influence model, it outputs the monitoring distortion rate range corresponding to the current temperature range and performs stability analysis on the current monitoring distortion rate range;

[0079] In the real-time distortion rate output and stability analysis module, the monitored distortion rate corresponding to the current temperature range is output. The specific process of performing stability analysis on the current monitored distortion rate is as follows:

[0080] Obtain real-time temperature data in the current energy operation site safety management scenario, determine the temperature range to which the current real-time temperature data belongs, and call the temperature range-monitoring distortion rate range impact model corresponding to the temperature range to which the current real-time temperature data belongs.

[0081] The model outputs the monitoring distortion rate range corresponding to the current temperature range, taking into account the influence of the temperature range and the monitoring distortion rate range.

[0082] In the real-time distortion rate output and stability analysis module, the specific process of performing stability analysis on the current monitored distortion rate range is as follows:

[0083] The monitoring distortion rate range corresponding to real-time temperature data in the current energy operation site safety management scenario is continuously collected. A coverage window (e.g., 20 seconds) is preset, and the average monitoring distortion rate is taken within each coverage window. A total of 10 coverage windows are collected to collect the average monitoring distortion rate.

[0084] Calculate the variance of the average monitoring distortion rate across 10 coverage windows;

[0085] If the fluctuation variance is less than the fluctuation variance threshold, the current monitored distortion rate range is determined to be in a stable state; if the fluctuation variance is greater than or equal to the fluctuation variance threshold, the current monitored distortion rate range is determined to be in an unstable state, and a fluctuation warning signal is output.

[0086] Vacuum compensation module: If the current monitoring distortion rate range is stable, it determines whether the current monitoring distortion rate range will cause a monitoring vacuum. If it will cause a monitoring vacuum, it increases the sensor acquisition frequency to perform monitoring vacuum compensation. If it will not cause a monitoring vacuum, it continuously monitors the current monitoring distortion rate range. If the current monitoring distortion rate range is unstable, it uses a time-series prediction model to determine whether the sensor is in a faulty state.

[0087] In the monitoring vacuum compensation module, if the current monitoring distortion rate range is stable, the specific process for determining whether the current monitoring distortion rate range will cause a monitoring vacuum is as follows:

[0088] Pre-set the monitoring vacuum threshold (this threshold is the critical failure standard for safety monitoring at the energy operation site, for example, the monitoring distortion rate is greater than or equal to 30%, that is, more than 30% of the data collected by the sensor cannot be received by the monitoring system).

[0089] By comparing the current distortion rate range with the monitoring vacuum threshold, the current stable distortion rate range is extracted, and its numerical relationship with the monitoring vacuum threshold is analyzed.

[0090] If the lower limit of the current stable monitoring distortion rate range is greater than or equal to the monitoring vacuum threshold, it is determined that the current monitoring distortion rate range will cause a monitoring vacuum, and a monitoring vacuum warning signal is output.

[0091] If the upper limit of the current stable monitoring distortion rate range is less than the monitoring vacuum threshold, it is determined that the current monitoring distortion rate range will not cause a monitoring vacuum.

[0092] If the current stable monitoring distortion rate range crosses the monitoring vacuum threshold (i.e., the lower limit of the current stable monitoring distortion rate range is less than the monitoring vacuum threshold and the upper limit of the current stable monitoring distortion rate range is greater than or equal to the threshold), then it is determined that the current monitoring distortion rate range will cause a monitoring vacuum, and a monitoring vacuum warning signal is output.

[0093] In the vacuum compensation module, if a vacuum is to be monitored, the sensor acquisition frequency is increased. The specific process for vacuum compensation is as follows:

[0094] Extract the average value of the current stable monitoring distortion rate (i.e., the average value of the distortion ratio that causes the monitoring vacuum), and at the same time obtain the current raw acquisition frequency of the sensor and the current actual receiving frequency of the monitoring system for the sensor data.

[0095] With the goal of restoring the actual receiving frequency of the monitoring system to the original acquisition frequency level after increasing the acquisition frequency, the target acquisition frequency to be adjusted is calculated.

[0096] For example, if the current average distortion rate is 30% and the original sensor acquisition frequency is 10 times per minute, since a distortion rate of 30% means that only 70% of the acquired data can be successfully received, the acquisition frequency needs to be increased to the original acquisition frequency divided by the undistorted proportion (1 minus the distortion rate), that is, 10 times divided by 70%, the result is approximately 14.3 times per minute, which is rounded up to 15 times per minute;

[0097] Determine in advance the maximum sampling frequency supported by the sensor hardware;

[0098] If the calculated target acquisition frequency does not exceed the maximum acquisition frequency of the hardware, then the sensor acquisition frequency is adjusted to the target frequency.

[0099] If the target acquisition frequency exceeds the maximum acquisition frequency of the hardware, the acquisition frequency will be adjusted to the maximum acquisition frequency supported by the hardware, and auxiliary measures such as data compression and data priority transmission will be triggered simultaneously to maximize the amount of sensor data received by the monitoring system.

[0100] After the sensor acquisition frequency adjustment is completed, the actual receiving frequency of the sensor data by the monitoring system is continuously monitored. If the actual receiving frequency reaches or exceeds the original acquisition frequency of the sensor, the monitoring vacuum compensation is determined to be successful.

[0101] If the actual receiving frequency still does not reach the original acquisition frequency, then combine the current monitoring distortion rate and repeat the above steps of calculating the target acquisition frequency and adjusting the acquisition frequency until the actual receiving frequency of the monitoring system meets the requirements of security monitoring.

[0102] In the monitoring vacuum compensation module, if a monitoring vacuum is not caused, the current monitoring distortion rate range is continuously monitored. The specific process is as follows:

[0103] Based on the preset coverage window duration, the window duration is shortened to half of the original duration (for example, if the original coverage window duration is 20 seconds, it is adjusted to 10 seconds) to improve the timeliness of monitoring.

[0104] Under the shortened and adjusted coverage window, the average value of 10 sets of current monitoring distortion rate is continuously collected. Based on the average value of these 10 sets of current monitoring distortion rate data, the fluctuation variance is calculated. The fluctuation variance reflects the stability of the change in the current monitoring distortion rate.

[0105] The calculated variance is compared with the preset variance threshold:

[0106] If the fluctuation variance is less than the fluctuation variance threshold, it means that the current monitoring distortion rate range is still in a stable state. Continue to collect and monitor the monitoring distortion rate range according to the adjusted coverage window.

[0107] If the fluctuation variance is greater than or equal to the fluctuation variance threshold, it indicates that there is abnormal fluctuation in the current monitoring distortion rate range and it is necessary to determine whether this will cause a monitoring vacuum.

[0108] In the vacuum compensation monitoring module, if the current monitored distortion rate range is unstable, the specific process of determining whether the sensor is in a faulty state through the time-series prediction model is as follows:

[0109] Extract the historical time-series data of the sensor, including: monitoring distortion rate time-series data under the same scene and temperature range in the past 30 days (stored at 1 minute / data point), temperature time-series data, and sensor operating status time-series data (such as power supply stability and hardware self-test results).

[0110] Using the current moment as the endpoint, capture the monitoring distortion rate time series data (at 10 seconds / data point) and temperature fluctuation time series data within the last 10 minutes. After unifying the time granularity with the historical time series data, integrate them to form the input dataset for the time series prediction model.

[0111] A time series prediction model is constructed using a Long Short-Term Memory (LSTM) network. The model is trained by using temperature time series data (specific temperature values) and sensor operating status time series data from historical time series data as input features and monitoring distortion rate time series data as output labels.

[0112] Input the current time series data into the trained time series prediction model, set the prediction duration to the next 5 minutes, and output the monitoring distortion rate prediction trend curve within the next 5 minutes.

[0113] Preset fault judgment thresholds, including distortion rate fluctuation thresholds (such as the distortion rate prediction trend curve fluctuating by more than 20% within the next 5 minutes) and deviation from the normal range thresholds (such as the distortion rate prediction trend curve exceeding the distribution range of 95% of historical data in the same scenario).

[0114] If the predicted trend curve of the monitored distortion rate satisfies the condition that the fluctuation amplitude is greater than or equal to the distortion rate fluctuation amplitude threshold and the predicted value is greater than or equal to the deviation from the normal range threshold, then the sensor is determined to be in a fault state (such as abnormal fluctuation of distortion rate caused by probe aging, damage to internal components, etc.), and a sensor fault warning signal is output.

[0115] If the predicted trend curve of the monitored distortion rate meets only one of the conditions or does not meet any of them, it is determined that the sensor is not in a faulty state, and the distortion rate instability is caused by temporary environmental interference (such as instantaneous temperature change or temporary interference in the transmission link), and an early warning signal is output.

[0116] Example 2

[0117] like Figure 2 As shown in Example 1, the present invention provides a safety management and early warning method for energy operation sites, comprising the following steps:

[0118] S10: Obtain historical temperature-monitoring distortion rate data, perform cluster analysis on historical temperature-monitoring distortion rate, and obtain a set of temperature range-monitoring distortion rate ranges;

[0119] S20: Based on the temperature range-monitoring distortion rate range set, conduct a correlation analysis on the temperature range and the monitoring distortion rate range, and construct an influence model of temperature range-monitoring distortion rate range based on the correlation analysis results;

[0120] S30: Based on the temperature range-monitoring distortion rate range influence model, output the monitoring distortion rate range corresponding to the current temperature range, and perform stability analysis on the current monitoring distortion rate range;

[0121] S40: If the current monitoring distortion rate range is stable, determine whether the current monitoring distortion rate range will cause a monitoring vacuum. If it will cause a monitoring vacuum, increase the sensor acquisition frequency and perform monitoring vacuum compensation. If it will not cause a monitoring vacuum, continuously monitor the current monitoring distortion rate range. If the current monitoring distortion rate range is unstable, determine whether the sensor is in a fault state through a time-series prediction model.

[0122] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A safety management and early warning system for energy operation sites, characterized in that: include: Historical data processing and clustering module: acquires historical temperature-monitoring distortion rate data, performs cluster analysis on historical temperature-monitoring distortion rate, and obtains a set of temperature range-monitoring distortion rate ranges; Correlation Analysis and Model Building Module: Based on the temperature range-monitoring distortion rate range set, perform correlation analysis on the temperature range and the monitoring distortion rate range, and build an influence model of temperature range-monitoring distortion rate range based on the correlation analysis results; The specific process for performing correlation analysis between the temperature range and the monitoring distortion rate range is as follows: For any set of temperature range-monitoring distortion rate range data, extract all temperature value-monitoring distortion rate sample values, with temperature value as independent variable and the corresponding monitoring distortion rate value as dependent variable, and calculate the correlation coefficient using the Pearson correlation coefficient method. If the correlation coefficient is greater than or equal to the correlation coefficient threshold, it is determined that the temperature value corresponding to this group has a linear correlation with the range of monitored distortion rate values. If the correlation coefficient is less than the correlation coefficient threshold, it is determined that the temperature value and the range of monitored distortion rate values ​​corresponding to this group are not linearly related. Real-time distortion rate output and stability analysis module: Based on the temperature range-monitored distortion rate range influence model, it outputs the monitoring distortion rate range corresponding to the current temperature range and performs stability analysis on the current monitoring distortion rate range; The specific process for performing stability analysis on the current monitoring distortion rate range is as follows: The monitoring distortion rate range corresponding to the current real-time temperature data is continuously collected. A coverage window is preset, and the average value of the monitoring distortion rate is taken in each coverage window. The average value of the monitoring distortion rate corresponding to 10 coverage windows is collected in total. Calculate the variance of the average monitoring distortion rate across 10 coverage windows; If the variance of fluctuation is less than the variance of fluctuation threshold, then the current monitored distortion rate range is determined to be in a stable state. If the fluctuation variance is greater than or equal to the fluctuation variance threshold, the current monitored distortion rate range is determined to be in an unstable state, and a fluctuation warning signal is output. Vacuum compensation module: If the current monitoring distortion rate range is stable, it determines whether the current monitoring distortion rate range will cause a monitoring vacuum. If it will cause a monitoring vacuum, it increases the sensor acquisition frequency to perform monitoring vacuum compensation. If it will not cause a monitoring vacuum, it continuously monitors the current monitoring distortion rate range. If the current monitoring distortion rate range is unstable, it uses a time-series prediction model to determine whether the sensor is in a faulty state.

2. The energy work site safety management and early warning system according to claim 1, characterized in that: The specific process for obtaining historical temperature and monitoring distortion rate data is as follows: The sensor acquires data at the corresponding temperature in the historical period, and the monitoring system receives the sensor data at the same frequency. The difference between the sensor's acquisition frequency and the monitoring system's receiving frequency is divided by the sensor's acquisition frequency to calculate the sensor's monitoring distortion rate at the corresponding temperature, thus obtaining the historical temperature-monitoring distortion rate data.

3. The energy work site safety management and early warning system according to claim 1, characterized in that: The specific process for obtaining the temperature range-distortion rate range set is as follows: Based on historical temperature-monitoring distortion rate data, integrate into a temperature-monitoring distortion rate sample data set, wherein each sample contains the corresponding temperature value of the sensor and the monitoring distortion rate at that temperature, and use clustering algorithm to perform clustering analysis on the temperature-monitoring distortion rate sample data set: With temperature value as the first feature dimension and monitoring distortion rate as the second feature dimension, samples with similar temperature values and similar monitoring distortion rates are classified into the same cluster, and the continuous temperature range and monitoring distortion rate range corresponding to the clustering result are determined, and the temperature range-monitoring distortion rate range corresponding to all clustering categories is summarized to obtain a temperature range-monitoring distortion rate range set.

4. The energy work site safety management and early warning system according to claim 1, characterized in that: The specific process of constructing the temperature range-monitoring distortion rate range influence model is: If it is determined that the temperature value and the monitoring distortion rate value have a linear correlation, then a linear regression algorithm is used to perform linear fitting on the independent variables in the temperature range and the dependent variables in the distortion rate range, and a temperature range-monitoring distortion rate range influence model is constructed; If it is determined that the temperature range and the monitoring distortion rate range do not have a linear correlation, then a polynomial regression algorithm is used to perform non-linear fitting on the independent variables in the temperature range and the dependent variables in the distortion rate range, and a temperature range-monitoring distortion rate range influence model is constructed.

5. The energy work site safety management and warning system of claim 1, wherein: The specific process of performing stability analysis on the current monitoring distortion rate is: Obtain the current real-time temperature data, determine the temperature range to which the current real-time temperature data belongs, call the temperature range-monitoring distortion rate range influence model corresponding to the temperature range to which the current real-time temperature data belongs, and combine the temperature range-monitoring distortion rate range influence model to output the monitoring distortion rate range corresponding to the current temperature range.

6. The energy work site safety management and warning system of claim 1, wherein: The specific process of determining whether the current monitoring distortion rate range will cause monitoring vacuum is: Extract the monitoring distortion rate range currently in a stable state, and analyze its numerical relationship with the monitoring vacuum threshold value: If the lower limit value of the monitoring distortion rate range currently in a stable state is greater than or equal to the monitoring vacuum threshold value, it is determined that it will cause monitoring vacuum, and a monitoring vacuum warning signal is output; If the upper limit value of the monitoring distortion rate range currently in a stable state is less than the monitoring vacuum threshold value, it is determined that it will not cause monitoring vacuum; If the monitoring distortion rate range currently in a stable state spans the monitoring vacuum threshold value, it is determined that it will cause monitoring vacuum, and a monitoring vacuum warning signal is output.

7. The energy work site safety management and warning system of claim 1, wherein: The specific process of performing monitoring vacuum compensation is: Extract the average value of the monitoring distortion rate currently in a stable state, the current original collection frequency of the sensor, and the current actual receiving frequency of the sensor data of the monitoring system; With the goal of restoring the actual receiving frequency of the monitoring system to the original collection frequency level after increasing the collection frequency, calculate the target collection frequency to be adjusted; Predefine the maximum collection frequency supported by the sensor hardware; If the calculated target collection frequency does not exceed the maximum hardware collection frequency, adjust the sensor collection frequency to the target frequency.

8. The energy work site safety management and early warning system according to claim 1, characterized in that: Including: The specific process of determining whether the sensor is in a fault state through the time series prediction model is: Extracting the historical time series data of the sensor, including: the monitoring distortion rate time series data, the temperature time series data, and the sensor running state time series data under the same scene and temperature interval in the past 30 days; Taking the current time as the end point, intercepting the monitoring distortion rate time series data and the temperature fluctuation time series data in the past 10 minutes, and after unifying the time granularity with the historical time series data, integrating to form the input data set of the time series prediction model; Using the long short-term memory network LSTM to construct the time series prediction model, taking the temperature time series data and the sensor running state time series data as the input features, and the monitoring distortion rate time series data as the output label, training to obtain the time series prediction model; Inputting the current time series data into the time series prediction model, outputting the monitoring distortion rate prediction trend curve in the next 5 minutes; if the monitoring distortion rate prediction trend curve satisfies the fluctuation amplitude ≥ distortion rate fluctuation amplitude threshold and the predicted value ≥ deviation from the normal interval threshold, it is determined that the sensor is currently in a fault state, and a sensor fault warning signal is output.

Images