Rotary kiln environment combustible gas concentration monitoring and emergency cut-off method and system

Abstract

The invention relates to the technical field of safety monitoring, in particular to a rotary kiln environment combustible gas concentration monitoring and emergency cut-off method and system. The method comprises the following steps: grouping combustible gas detectors into sensor clusters according to positions, and calculating a concentration representative value; calculating a cross-correlation function of the concentration representative value sequence of each sensor cluster in the sliding time window, and extracting correlation intensity and propagation time delay; constructing a statistical topological graph taking the sensor clusters as nodes; and when the concentration representative value exceeds an early warning threshold value, screening the association clusters based on a statistical topological graph, determining a troubleshooting time window by utilizing propagation time delay, counting the number of the association clusters of a concentration rising trend, and outputting an emergency cut-off instruction when the concentration representative value exceeds a voting threshold value. According to the method, propagation consistency verification is realized through sensor cluster cooperative sensing and cross-correlation analysis, real leakage and environmental interference are effectively distinguished, the false alarm rate is reduced, and the emergency response reliability is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of safety monitoring. More particularly, the present application relates to a rotary kiln environment combustible gas concentration monitoring and emergency shutdown method and system. BACKGROUND

[0002] Rotary kilns are commonly used in the material calcination, sintering, and incineration processes of the building materials, metallurgy, and chemical industries, often accompanied by the delivery and use of combustible gases such as natural gas, coal gas, and hydrogen. Due to long-term operation of the equipment, aging of the sealing components, loosening of the pipeline connections, or fluctuations in the working conditions, combustible gas leakage may occur in the environment surrounding the rotary kiln and the gas delivery pipeline, which is prone to cause combustion or explosion under the conditions of high-temperature hot surfaces, open flames, and static electricity, endangering the safety of personnel and equipment. Therefore, it is necessary to continuously monitor the combustible gas concentration in the environment surrounding the rotary kiln and timely trigger graded alarms and linkage controls when the dangerous threshold is reached, so as to reduce the probability of accidents and expand the time window for handling accidents.

[0003] In the prior art, combustible gas detectors are arranged on the outer walls of the kiln and the gas delivery pipeline, and a controller and a risk control module are combined to trigger alarms, ventilation dilution, spraying, condensation, or inert gas measures at different concentration thresholds to reduce the risk of accidents. Another solution combines leakage concentration detection with pressure and flow monitoring to reduce false alarms and trigger a shutdown unit to shut down. However, the above solutions mostly use static threshold triggering logic, which does not adequately consider measurement drift, sensor contamination, and failure diagnosis caused by the complex environment of high temperatures, dust, and strong air currents in the rotary kiln site, and lacks effective isolation means when a single detector drifts or fails, which can easily lead to false alarms or missed alarms.

[0004] In addition, existing combustible gas monitoring solutions have deficiencies in multi-detector cooperation: each detection point usually works and judges independently, lacking analysis and utilization of the timing relationship of concentration changes at multiple points; the gas propagation relationship between zones relies on manual calibration or fixed configuration, making it difficult to adapt to dynamic air flow changes caused by factors such as the state of the induced draft fan, the rotation of the kiln body, and the entry and exit of materials; and there is a lack of verification mechanisms based on multi-point propagation consistency in alarm decision-making, making it difficult to effectively distinguish between real leakage events and environmental disturbances. Therefore, it is still necessary to provide a combustible gas safety monitoring solution that is more suitable for the environmental characteristics of the rotary kiln site and has reliable concentration monitoring and graded linkage handling capabilities. SUMMARY

[0005] The present application aims to provide a rotary kiln environment combustible gas concentration monitoring and emergency shutdown method and system to solve the problems of insufficient single-point detector failure isolation capability and lack of multi-point cooperative verification mechanisms in existing rotary kiln environment combustible gas monitoring.

[0006] In a first aspect, the present application provides a rotary kiln combustible gas concentration monitoring and emergency cut-off method, comprising: grouping combustible gas detectors into sensor clusters according to position, calculating the absolute deviation of each detector reading in the cluster from the median of all readings in the cluster, eliminating detector data whose absolute deviation exceeds a preset threshold, and calculating the median of the remaining detector readings as the concentration representative value of the current sensor cluster; calculating the cross-correlation function of the concentration representative value sequence of each sensor cluster within a preset sliding time window, extracting the maximum value of the cross-correlation function as the correlation strength, and extracting the time lag corresponding to the maximum value of the cross-correlation function as the propagation delay; constructing a statistical topology graph with sensor clusters as nodes and sensor cluster connection relationships as edges, the edges of the statistical topology graph containing correlation strength and propagation delay attributes; when the concentration representative value of any sensor cluster exceeds the warning threshold, selecting sensor clusters with correlation strength satisfying the validity condition as associated clusters based on the statistical topology graph, determining the investigation time window of each associated cluster using the propagation delay, counting the number of associated clusters whose concentration representative value shows an upward trend within the investigation time window, and outputting an emergency cut-off instruction to close the gas pipeline electromagnetic valve when the number exceeds a preset voting threshold.

[0007] The present application effectively suppresses the reading drift and pollution interference of single-point sensors caused by high temperature, dust and other environmental factors through the median aggregation and cluster deviation elimination mechanism of sensor clusters, realizes the automatic identification and isolation of faulty sensors, estimates the gas propagation delay between adjacent clusters online through cross-correlation analysis, dynamically constructs a statistical topology graph without relying on tracer gas calibration or manual configuration of static parameters, enables the emergency response to adapt to the airflow dynamic changes in the rotary kiln site, and confirms the leakage event only after multiple sensor clusters with propagation correlation relationship show consistent concentration change response within the corresponding time delay window through the propagation consistency causal voting mechanism, effectively distinguishes real leakage from isolated abnormal signals caused by environmental interference, reduces the false alarm rate and improves the reliability of emergency cut-off.

[0008] Optionally, the construction of the sensor cluster comprises: obtaining the spatial coordinates of each combustible gas detector, and dividing the combustible gas detectors with a spatial distance within a preset distance threshold and located in the same airflow circulation area into the same sensor cluster; and the arrangement area of the sensor cluster covers at least the kiln head burner area, the kiln head sealing area, the kiln body along the line, the kiln tail sealing area and the gas valve group area of the rotary kiln.

[0009] The present application ensures that the detectors in each cluster are within the same airflow influence range by obtaining the spatial coordinates of the detectors and dividing the sensor clusters according to the preset distance threshold and airflow area, so that the cluster representative value can truly reflect the combustible gas concentration state of the area; and the key positions such as the kiln head burner area, the kiln head sealing area, the kiln body along the line, the kiln tail sealing area and the gas valve group area are covered to realize the monitoring coverage of combustible gas in the whole process of the rotary kiln.

[0010] Optionally, removing detector data whose absolute deviation value exceeds a preset threshold includes: calculating the difference between the readings of each combustible gas detector in the sensor cluster and the median of all detector readings in the current sensor cluster, and taking its absolute value as the absolute deviation value; marking the detector with the largest absolute deviation value as a suspected abnormal node; if the absolute deviation value of the suspected abnormal node exceeds the preset deviation threshold and the number of remaining detectors is greater than the preset minimum retention number, removing it, and recalculating the median using the remaining detector readings, repeating the above steps until no detector has an absolute deviation value exceeding the preset deviation threshold or the number of remaining detectors reaches the preset minimum retention number.

[0011] This application employs an iterative elimination strategy. By repeatedly calculating the absolute value of the deviation and eliminating the detector with the largest deviation, multiple abnormal sensors can be gradually identified and isolated. At the same time, a minimum number of retentions is set to prevent excessive elimination, ensuring that a sufficient number of reliable detectors are always involved in the concentration representative value calculation.

[0012] Optionally, the extraction process of the correlation strength and propagation delay includes: performing mean-reduction and normalization processing on the concentration representative value sequence within a preset sliding time window; moving the sequence within the preset sliding time window, calculating the cross-correlation coefficient of the two sensor cluster sequences, and obtaining the cross-correlation function curve; searching for the maximum peak value in the cross-correlation function curve, and determining the value of the maximum peak value as the correlation strength; obtaining the time lag value corresponding to the maximum peak value, and determining the time lag value as the propagation delay.

[0013] Optionally, the process of constructing the validity condition includes: setting a minimum association threshold; when the association strength is greater than or equal to the minimum association threshold, determining that the association strength meets the validity condition; when the association strength is less than the minimum association threshold, determining that the association strength does not meet the validity condition, and marking that there is no direct propagation relationship between the two clusters.

[0014] Optionally, the construction of the statistical topology graph includes: recording all sensor cluster pairs that satisfy the validity conditions as edges of the statistical topology graph, the attributes of which include propagation delay and association strength; periodically updating the statistical topology graph at a low frequency when the rate of change of the detector parameters does not exceed a preset change threshold; and triggering accelerated updates and shortening the preset sliding time window length when the rate of change of the detector parameters exceeds the preset change threshold.

[0015] This application uses a lower frequency of updating the statistical topology map during stable operating conditions to save computing resources, while triggering an accelerated update mechanism when a change in operating conditions is detected, so that the statistical topology map can quickly adapt to new airflow propagation characteristics and ensure the accuracy of causal voting.

[0016] Optionally, the step of determining the investigation time window for each associated cluster using propagation delay includes: obtaining the current time when the warning is triggered; in the statistical topology graph, filtering associated clusters whose propagation delay is less than or equal to a preset synchronization threshold, and removing associated clusters whose propagation delay is greater than the preset synchronization threshold; calculating the backtracking time by subtracting the absolute value of the propagation delay from the current time; and selecting a preset time interval with the backtracking time as the midpoint as the investigation time window.

[0017] This application selects upstream nodes and synchronous response nodes as valid voting targets based on the comparison between propagation delay and preset synchronization threshold, eliminates lagging downstream nodes, and uniformly adopts a historical backtracking method to determine the investigation time window. This mechanism enables the causal voting process to be based entirely on historical data in memory, without waiting for gas to diffuse downstream, thereby achieving immediate verification and zero-delay cutoff of leakage events, significantly improving the real-time performance of emergency response.

[0018] Optionally, the criteria for determining that the concentration representative value shows an upward trend include: calculating the first derivative or slope of the concentration representative value sequence of the associated cluster within the investigation time window; when the average value of the slope is greater than zero, and the maximum concentration representative value within the investigation time window exceeds the average concentration value of the associated cluster within a preset period before the triggering of the warning, it is determined that there is an upward trend in concentration.

[0019] Optionally, the method further includes: when the concentration representative value of any sensor cluster exceeds the warning threshold, but the number of associated clusters showing an upward trend in concentration does not exceed the preset voting threshold, a warning state is entered; in the warning state, a warning signal is output and the gas pipeline solenoid valve is kept in the open state, while the preset sliding time window length of the cross-correlation calculation is shortened.

[0020] In the second aspect, a rotary kiln environment combustible gas concentration monitoring and emergency shut-off system includes: processor; The memory stores computer instructions for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment. When the computer instructions are executed by the processor, the system performs the aforementioned method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment.

[0021] The beneficial effects of this application are as follows: This application effectively suppresses reading drift and pollution interference caused by environmental factors such as high temperature and dust in single-point sensors through median aggregation and intra-cluster bias elimination mechanism of sensor clusters, realizing automatic identification and isolation of faulty sensors; through cross-correlation analysis, the gas propagation delay between adjacent clusters is estimated online, and a statistical topology map is dynamically constructed without relying on tracer gas calibration or manual configuration of static parameters, so that the emergency response can adapt to the dynamic changes in airflow caused by factors such as the status of the induced draft fan and the rotation of the kiln body at the rotary kiln site; through the propagation consistency causal voting mechanism, the leakage event is confirmed only after multiple sensor clusters with propagation correlation show consistent concentration change response within the corresponding delay window, effectively distinguishing between real leakage and isolated abnormal signals caused by environmental interference, reducing false alarm rate and improving the reliability of emergency cut-off. Attached Figure Description

[0022] Figure 1 This is a flowchart of a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to an embodiment of this application.

[0023] Figure 2 This is a statistical topology diagram of sensor clusters for a method of monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to an embodiment of this application.

[0024] Figure 3 This is a schematic diagram of the propagation time delay of a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to an embodiment of this application.

[0025] Figure 4 This is a causal voting and backtracking verification timing diagram of a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to an embodiment of this application.

[0026] Figure 5 This is a structural block diagram of a rotary kiln environment combustible gas concentration monitoring and emergency shut-off system according to an embodiment of this application. Detailed Implementation

[0027] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Figure 1 The diagram shown is a flowchart of a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to an embodiment of this application.

[0028] S1: Group the combustible gas detectors by location to form sensor clusters and calculate representative concentration values.

[0029] First, the combustible gas detectors deployed around the rotary kiln are grouped according to their spatial location to form sensor clusters, and representative concentration values ​​are calculated. At the rotary kiln production site, combustible gas detectors are distributed in key locations such as the kiln head burner area, the kiln head sealing area, various sections along the kiln body, the kiln tail sealing area, the gas pipeline corridor area, and the valve group area.

[0030] The spatial coordinates of each combustible gas detector are obtained, and combustible gas detectors that are within a preset distance threshold and located in the same airflow area are grouped into the same sensor cluster. In this embodiment, the preset distance threshold is set to 5 meters, meaning that the straight-line spatial distance between detectors within the same cluster does not exceed 5 meters. A typical sensor cluster size is at least 3 combustible gas detectors per cluster.

[0031] For each sensor cluster, the concentration readings of each detector within the cluster are acquired at the sampling time. Let a sensor cluster contain n combustible gas detectors, and their concentration readings be denoted as C1, C2, ..., C... n First, the median concentration readings of all detectors within the cluster are calculated, and this median is used as the baseline value.

[0032] Calculate the difference between the readings of each combustible gas detector within the sensor cluster and the median reading of all detectors within the cluster, and take its absolute value as the absolute deviation value. Mark the detector with the largest absolute deviation value as a suspected anomalous node. If the absolute deviation value of a suspected anomalous node exceeds a preset deviation threshold, and the number of remaining detectors is greater than a preset minimum retention number, remove it from the cluster. Recalculate the median using the remaining detector readings, and repeat the above steps until no detector's absolute deviation value exceeds the preset deviation threshold or the number of remaining detectors reaches the preset minimum retention number. In this embodiment, the preset deviation threshold is set to 3%LEL, and the preset minimum retention number is set to 2.

[0033] The advantage of median aggregation over mean aggregation is that when a sensor within a cluster generates an abnormal reading due to dust adhesion, high-temperature drift, or other reasons, the median calculation is unaffected by this abnormal value, effectively maintaining the robustness of the concentration representative value. This mechanism enables automatic isolation of single-point faults, ensuring the accuracy of the cluster representative value.

[0034] S2: Calculate the cross-correlation function of the sensor cluster and extract the correlation strength and propagation delay.

[0035] After obtaining the representative concentration values ​​for each sensor cluster, the cross-correlation function of the concentration representative value sequence of the sensor cluster within the sliding time window is calculated, and the correlation strength and propagation delay are extracted. For any two adjacent sensor clusters i and j, the normalized cross-correlation function of their concentration representative value sequence within the sliding time window is continuously calculated.

[0036] The concentration representative value sequence within the sliding time window is subjected to mean removal and normalization. The sliding time window length is set to 300 seconds, and the time delay parameter traversal range is from -60 seconds to +60 seconds, where the time delay range is estimated based on the maximum spatial distance between adjacent clusters and the minimum gas diffusion rate.

[0037] Within a preset time lag range, the sequence is moved, and the cross-correlation coefficient between the two target sensor cluster sequences is calculated to obtain the cross-correlation function curve. The formula for calculating the cross-correlation function is as follows: ; in, Represents sensor cluster With sensor cluster Cross-correlation coefficients between them; This represents the time delay parameter, in seconds. and Representing sensor clusters and sensor clusters The average concentration value within the time window, expressed as %LEL; and Representing sensor clusters and sensor clusters The standard deviation of the concentration representative value within the time window; N represents the number of sampling points within the time window; This indicates the sampling time within the sliding time window. and Representing sensor clusters and sensor clusters The concentration value at the corresponding time point, in %LEL.

[0038] The maximum peak value is searched in the cross-correlation function curve, and the value of the maximum peak value is determined as the correlation strength. The time lag corresponding to the maximum peak value is obtained, and the time lag value is determined as the propagation delay.

[0039] S3: Construct a statistical topology graph and set validity criteria.

[0040] After extracting the correlation strength and propagation delay between each sensor cluster pair, a statistical topology graph is constructed with sensor clusters as nodes and sensor cluster connection relationships as edges, and validity judgment conditions are set.

[0041] A minimum association threshold is set; in this embodiment, the minimum association threshold is set to 0.5. When the association strength is greater than or equal to the minimum association threshold, the association strength is determined to meet the validity condition; when the association strength is less than the minimum association threshold, the association strength is determined not to meet the validity condition, and it is marked that there is no direct propagation relationship between the two clusters.

[0042] All sensor cluster pairs that satisfy the validity conditions are recorded as edges in a statistical topology graph. The attributes of the edges include propagation delay and association strength. The statistical topology graph uses each sensor cluster as a graph node and the cluster pairs that satisfy the validity conditions as directed edges. The direction of the edges is determined by the sign of the propagation delay.

[0043] The update strategy for the statistical topology map is as follows: During periods of stable operation, the statistical topology map is updated periodically at a low frequency; in this embodiment, the update cycle is set to 10 minutes. When the rate of change of the detector parameters exceeds a preset threshold, an accelerated update is triggered, shortening the sliding time window. In this embodiment, when the frequency change of the induced draft fan exceeds 10%, the kiln corner crosses a critical sector, or the material entry / exit status changes, an accelerated update is triggered, shortening the sliding time window to 120 seconds.

[0044] Through the aforementioned online self-calibration mechanism, the statistical topology map can continuously reflect the airflow propagation characteristics under the current actual operating conditions without the need for manual calibration. Figure 2 The diagram shown is a statistical topology diagram of sensor clusters for a method of monitoring and emergency shut-off of combustible gas concentration in a rotary kiln according to an embodiment of this application. Circular nodes represent sensor clusters distributed in locations such as the burner area at the kiln head, the kiln body, and the gas valve assembly area. The connecting edges between nodes represent effective gas propagation paths; the color intensity of the edges corresponds to the strength of the association, and the numerical values ​​on the edges correspond to the propagation delay of gas between the two nodes. This topology is used to identify upstream and downstream nodes to support subsequent causal voting.

[0045] S4: Perform causal voting based on the statistical topology graph and output emergency disconnection command.

[0046] When the concentration representative value of any sensor cluster exceeds the warning threshold, causal voting is performed based on the statistical topology graph, and an emergency shutdown command is output. In this embodiment, the warning threshold is set to 10% LEL.

[0047] The current moment when the warning is triggered is obtained. A historical backtracking mechanism is used for verification to ensure the real-time nature of the cutoff command. First, in the statistical topology graph, associated clusters with propagation delays less than or equal to a preset synchronization threshold are selected as valid voting nodes. Here, associated clusters with negative propagation delays represent upstream nodes, and associated clusters with propagation delays close to zero (less than the synchronization threshold) represent nearby synchronization response nodes; while associated clusters with propagation delays greater than the preset synchronization threshold are considered downstream nodes that have not yet received gas and are eliminated, not participating in this vote. In this embodiment, the preset synchronization threshold is set to 2 seconds.

[0048] like Figure 3The diagram illustrates the propagation time delay of a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to an embodiment of this application. It shows the normalized concentration change trends of the upstream and downstream sensor clusters within a preset sliding time window. The maximum correlation peak of the two curves is determined through cross-correlation calculations, and the upstream peak time is identified. Downstream peak time The time difference between them is determined as the propagation delay, thereby quantifying the gas flow velocity in space.

[0049] For the selected valid voting nodes, the absolute value of the propagation delay is subtracted from the current time to obtain the backtracking time. This backtracking time corresponds to the historical point in time when the gas flows through the upstream node and causes a concentration change. A preset time interval with the backtracking time as the midpoint is selected as the investigation time window. In this embodiment, the half-width of the preset time interval is set to 15 seconds. The historical concentration data cached in the memory directly calls the segment corresponding to this investigation time window, without the need for new data collection or waiting.

[0050] like Figure 4 The diagram shown is a causal voting and backtracking verification time sequence diagram of a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to an embodiment of this application. When the concentration is detected to exceed the warning threshold at the current moment, the propagation delay in the statistical topology diagram is used to trace back to determine the historical backtracking time. A screening time window centered on this backtracking time is locked, and the presence of a corresponding concentration peak and upward trend in the upstream associated cluster is verified within this window to eliminate environmental interference and confirm the actual leakage event.

[0051] Subsequently, it is determined whether the concentration representative value shows an upward trend within the investigation time window. The first derivative or slope of the concentration representative value sequence of the associated cluster within the investigation time window is calculated. When the average value of the slope is greater than zero, and the maximum concentration representative value within the investigation time window exceeds the average concentration value of the associated cluster within a preset period before the triggering of the warning, it is determined that an upward concentration trend is present. In this embodiment, the preset period is set to 60 seconds.

[0052] The number of associated clusters whose concentration representative values ​​show an upward trend within the investigation time window is counted. When the number exceeds a preset voting threshold, it is determined to be a high-confidence leak event, and an emergency shut-off command to close the gas pipeline solenoid valve is output. In this embodiment, the preset voting threshold is set to 2.

[0053] When the concentration representative value of any sensor cluster exceeds the warning threshold, but the number of associated clusters showing an upward concentration trend does not exceed the preset voting threshold, a warning state is entered. In the warning state, a warning signal is output and the gas pipeline solenoid valve remains open. At the same time, the sliding time window length of the cross-correlation calculation is shortened to accelerate the update speed of the statistical topology map.

[0054] The emergency shut-off command is sent to the gas pipeline solenoid valve through the controller, which closes the main gas valve and all branch valves, and stops the rotary kiln burner from operating.

[0055] According to a second aspect of this application, this application also provides a rotary kiln environment combustible gas concentration monitoring and emergency shut-off system. Figure 5 This is a structural block diagram of a rotary kiln environment combustible gas concentration monitoring and emergency shut-off system according to an embodiment of this application. Figure 5 As shown, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to the first aspect of this application. The system also includes other components well-known to those skilled in the art, such as a communication bus and communication interface. Their configuration and functions are known in the art and will not be described further here.

[0056] The above description is merely a preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in this application, based on the technical solution and inventive concept of this application, should be within the scope of protection of this application.

Claims

1. A method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment, characterized in that, The method includes: Combustible gas detectors are grouped into sensor clusters according to their location. The absolute value of the deviation between the reading of each detector in the cluster and the median value of all readings in the cluster is calculated. Detector data with the absolute value of the deviation exceeding a preset threshold are removed. The median value of the remaining detector readings is calculated as the concentration representative value of the current sensor cluster. Calculate the cross-correlation function of the concentration representative value sequence of each sensor cluster within a preset sliding time window, extract the maximum value of the cross-correlation function as the correlation strength, and extract the time delay corresponding to the maximum value of the cross-correlation function as the propagation delay; Construct a statistical topology graph with sensor clusters as nodes and sensor cluster connection relationships as edges. The edges of the statistical topology graph include association strength and propagation delay attributes. When the concentration representative value of any sensor cluster exceeds the warning threshold, sensor clusters whose correlation strength meets the validity condition are selected as associated clusters based on the statistical topology graph. The investigation time window of each associated cluster is determined by using the propagation delay. The number of associated clusters whose concentration representative value shows an upward trend within the investigation time window is counted. When the number exceeds the preset voting threshold, an emergency shut-off command to close the gas pipeline solenoid valve is output.

2. The method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 1, characterized in that, The construction of the sensor cluster includes: Obtain the spatial coordinates of each combustible gas detector, and classify combustible gas detectors that are within a preset distance threshold and located in the same airflow area into the same sensor cluster; The sensor cluster is deployed in an area that covers at least the kiln head burner area, kiln head sealing area, kiln body cylinder line, kiln tail sealing area, and gas valve group area of ​​the rotary kiln.

3. The method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 1, characterized in that, Eliminating detector data whose absolute deviation value exceeds a preset threshold includes: calculating the difference between the readings of each combustible gas detector in the sensor cluster and the median of all detector readings in the current sensor cluster, and taking its absolute value as the absolute deviation value. The detector with the largest absolute value of the deviation is marked as a suspected abnormal node; If the absolute value of the deviation of the suspected abnormal node exceeds the preset deviation threshold and the number of remaining detectors is greater than the preset minimum retention number, it is removed, and the median is recalculated using the readings of the remaining detectors. The above steps are repeated until the absolute value of the deviation of no detector exceeds the preset deviation threshold or the number of remaining detectors reaches the preset minimum retention number.

4. The method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 1, characterized in that, The extraction process of the correlation strength and propagation delay includes: The concentration representative value sequence within the preset sliding time window is subjected to mean removal and normalization processing; Within a preset sliding time window, the sequence is moved around, and the cross-correlation coefficient of the two sensor cluster sequences is calculated to obtain the cross-correlation function curve; Search for the maximum peak value in the cross-correlation function curve, and determine the value of the maximum peak value as the correlation strength; Obtain the time lag corresponding to the maximum peak value, and determine the time lag as the propagation delay.

5. The method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 4, characterized in that, The process of constructing the validity conditions includes: A minimum association threshold is set. When the association strength is greater than or equal to the minimum association threshold, the association strength is determined to meet the validity condition. When the association strength is less than the minimum association threshold, the association strength is determined to be invalid, and the two clusters are marked as having no direct propagation relationship.

6. The method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 5, characterized in that, The construction of the statistical topology graph includes: All sensor cluster pairs that satisfy the validity conditions are recorded as edges in a statistical topology graph, and the attributes of the edges include propagation delay and association strength. The statistical topology map is updated periodically at a low frequency when the rate of change of the detector parameters does not exceed a preset change threshold. When the rate of change of the detector parameters exceeds the preset change threshold, an accelerated update is triggered, shortening the preset sliding time window length.

7. The method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 1, characterized in that, The steps for determining the investigation time window for each associated cluster using propagation delay include: Get the current time when the alert was triggered; In the statistical topology graph, clusters with propagation delay less than or equal to a preset synchronization threshold are filtered out, and clusters with propagation delay greater than the preset synchronization threshold are removed. The backtracking time is obtained by subtracting the absolute value of the propagation delay from the current time. A preset time interval with the backtracking time as the midpoint is selected as the investigation time window.

8. The method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 1, characterized in that, The criteria for determining that the representative concentration value shows an upward trend include: Calculate the first derivative or slope of the concentration representative value sequence of the associated cluster within the investigation time window; When the average value of the slope is greater than zero, and the maximum concentration representative value within the investigation time window exceeds the average concentration value of the associated cluster within a preset period before the triggering of the warning, it is determined that there is an upward trend in concentration.

9. A method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment according to claim 1, characterized in that, The method further includes: When the concentration representative value of any sensor cluster exceeds the warning threshold, but the number of associated clusters showing an upward trend in concentration does not exceed the preset voting threshold, the system enters the warning state. In the early warning state, an early warning signal is output and the gas pipeline solenoid valve is kept open, while the length of the preset sliding time window for cross-correlation calculation is shortened.

10. A rotary kiln environment combustible gas concentration monitoring and emergency shut-off system, characterized in that, include: processor; A memory, wherein a computer program is stored; When the processor is configured to execute the computer program, it implements a method for monitoring and emergency shut-off of combustible gas concentration in a rotary kiln environment as described in any one of claims 1 to 9.

Images