A refueling machine pipeline fault alarm system based on multi-source data

Abstract

The application discloses a refueling machine pipeline fault alarm system based on multi-source data and relates to the technical field of fault monitoring.The application solves the technical problem that there is lack of system integration and deep analysis of multi-source monitoring data, monitoring standards cannot be formulated in combination with pipeline parameter fluctuation rules in different time periods, monitoring threshold values are fixed, and false positives or false negatives are prone to occur.The multi-source data acquisition module integrates multi-type sensor data, and all-around monitoring of the refueling machine pipeline state is realized.Meanwhile, the abnormal state recognition module uses time series analysis to mine historical data rules, formulates dynamically adaptive state recognition threshold values, effectively reduces false positives and false negatives caused by one-sidedness of data or threshold value fixation, analyzes relevance through Pearson correlation coefficient analysis, divides parameter subsets through DBSCAN clustering, accurately judges whether faults are of the same root cause or parallel faults, improves fault cause positioning efficiency and accuracy, and reduces dependence on manual experience.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault monitoring, in particular to a fuel dispenser pipeline fault alarm system based on multi-source data. BACKGROUND

[0002] In the operation process of a gas station, the fuel dispenser pipeline, as the core channel for fuel transmission, its running state is directly related to the safety and normal business development of the gas station. With the improvement of the automation and intelligence level of the gas station, the traditional fuel dispenser pipeline fault monitoring method has gradually been difficult to meet the actual needs.

[0003] The traditional monitoring relies on manual inspection or single sensor data judgment, and in terms of data processing, there is a lack of systematic integration and in-depth analysis of multi-source monitoring data, often only single-dimensional pipeline state information can be obtained, and it is difficult to fully reflect the real running situation of the pipeline. At the same time, the traditional method lacks the use of historical data, and cannot develop adaptive monitoring standards combined with the fluctuation rules of pipeline parameters at different times, resulting in fixed monitoring threshold, prone to false positives or false negatives. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a fuel dispenser pipeline fault alarm system based on multi-source data, which solves the problems of lack of systematic integration and in-depth analysis of multi-source monitoring data, inability to develop adaptive monitoring standards combined with the fluctuation rules of pipeline parameters at different times, and fixed monitoring threshold, prone to false positives or false negatives.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a fuel dispenser pipeline fault alarm system based on multi-source data, comprising:

[0006] An abnormal state recognition module is configured to preprocess the obtained state parameters to obtain preprocessed parameters, and is further configured to obtain historical data of the fuel dispenser pipeline and analyze and arrange to obtain corresponding state recognition thresholds, and compare the preprocessed parameters with the state recognition thresholds, if any one of the preprocessed parameters does not exist in the state recognition thresholds, an abnormal analysis signal is generated and transmitted to an abnormal reason analysis module;

[0007] An abnormal reason analysis module is configured to process the abnormal analysis signal, and identify the abnormal data as single abnormality or multiple abnormality; for single abnormality, its features are extracted and matched with a feature matching library to determine the abnormal reason, and transmitted to a fault comprehensive analysis module; for multiple abnormality, the correlation is analyzed and classified as all correlation or partial correlation, and the abnormal reason is analyzed and transmitted to the fault comprehensive analysis module after classification;

[0008] The fault comprehensive analysis module is used to calculate the level of single fault causes and combined fault causes based on the obtained fault causes, and transmit the results to the fault alarm information output module.

[0009] As a further aspect of the present invention, it also includes a multi-source data acquisition module, which is used to acquire state parameters inside the fuel dispenser pipeline through multiple sensors and transmit the state parameters to the abnormal state identification module.

[0010] The sensors include pressure sensors, temperature sensors, flow sensors, vibration sensors, and audio sensors, and the multi-source data acquisition module acquires corresponding types of status parameters according to different sensors.

[0011] The fault alarm information output module is used to receive continuous monitoring signals or fault levels and output them accordingly.

[0012] As a further aspect of the present invention, the anomaly cause analysis module processes the anomaly analysis signal in the following manner:

[0013] The abnormal parameter type, numerical fluctuation range, duration, and associated sensor location are separated from the abnormal analysis signal. The abnormal type is determined by parameter independence analysis. A single abnormality is when only one sensor data exceeds the limit, while multiple abnormalities are when two or more sensors data exceed the limit.

[0014] As a further aspect of the present invention, when the anomaly cause analysis module processes a single anomaly, the extracted anomaly feature dimensions include numerical features, temporal features, and environmental correlation features. The numerical features include the pressure drop magnitude and flow fluctuation frequency; the temporal features include sudden / gradual change and continuous / intermittent occurrence; and the environmental correlation features include whether it occurs during the peak refueling period and whether it is accompanied by a sudden temperature change.

[0015] Obtain a feature matching library containing feature-cause mapping relationships extracted from historical failure cases, achieve feature similarity matching through the K-nearest neighbor algorithm, and output the anomaly cause with the highest matching degree.

[0016] As a further aspect of the present invention, when the abnormal cause analysis module analyzes the correlation of various types of abnormal situations, full correlation means that all abnormal parameters are caused by the same fault root cause and the parameter changes are strongly correlated; partial correlation means that some of the abnormal parameters are correlated, while the other part is an independent abnormality.

[0017] The correlation of parameter fluctuations was calculated using the Pearson correlation coefficient, and time series analysis confirmed that the starting points of the changes were consistent, thus determining that the correlation was complete. The abnormal parameters were divided into correlated subsets and independent parameters using the DBSCAN clustering algorithm, and the correlation of parameters within the group was calculated, thus determining that the correlation was partial.

[0018] As a further aspect of the present invention, when the anomaly cause analysis module processes all associated anomalies of various types, it first matches the single parameter features to obtain preliminary causes, then merges the multi-parameter features into composite features, and uses the composite features as search conditions to match the unique fault cause in the feature library.

[0019] When dealing with multiple types of abnormal situations that are partially related, the related subset is matched with all related logical features to obtain the related fault cause, and the independent parameter is matched with a single abnormality logic to obtain the independent fault cause. The combined abnormal causes of related faults and independent faults are formed and the independence of each cause is marked.

[0020] As a further aspect of the present invention, when the fault comprehensive analysis module calculates the fault level for the sole cause of a fault, it selects three core dimensions: safety impact, diffusion speed, and handling difficulty. Each dimension is set with a quantitative scoring standard of 1-10 points based on the fault characteristic parameters.

[0021] Safety impact assessment focuses on the safety threats posed by the fault to personnel, equipment, and the environment; diffusion rate analysis parameters include deterioration rate; and processing difficulty analysis parameters include recovery time.

[0022] Different weights are assigned to safety impact, spread rate, and handling difficulty, and a weighted sum is obtained to obtain a comprehensive score. The corresponding fault level is determined based on the comprehensive score.

[0023] As a further aspect of the present invention, when the fault comprehensive analysis module calculates the fault level of the combined fault causes, it calculates the comprehensive score and preliminary level of each independent fault in the combined fault according to the calculation method of the unique fault cause.

[0024] If there is a causal relationship between faults, the comprehensive score = primary fault score + 0.3 × secondary fault score, and the level is remapped according to the new score; if there is no relationship between faults, the highest level among the initial fault levels is taken, and a +0.5 point correction is made for the second highest level ≥ 3, to generate the fault level.

[0025] This invention provides a fuel dispenser pipeline fault alarm system based on multi-source data. Compared with existing technologies, it has the following advantages:

[0026] This invention integrates data from multiple types of sensors, such as pressure, temperature, flow, vibration, and audio, through a multi-source data acquisition module to achieve comprehensive monitoring of the fuel dispenser pipeline status. At the same time, the abnormal status identification module uses time series analysis to mine historical data patterns and formulate dynamically adapted status identification thresholds, effectively reducing false alarms and missed alarms caused by data bias or fixed thresholds.

[0027] The anomaly cause analysis module of this invention combines standardized feature extraction with a feature matching library to achieve rapid matching for single anomalies; for multiple types of anomalies, it uses Pearson correlation coefficient analysis to determine correlation and DBSCAN clustering to divide parameter subsets to accurately determine whether the faults have the same root cause or are parallel faults, thereby improving the efficiency and accuracy of fault cause localization and reducing reliance on manual experience.

[0028] This invention's fault comprehensive analysis module quantifies fault risks from three core dimensions: safety impact, spread speed, and handling difficulty. It determines fault levels and clarifies fault response priorities through weighted summation and dynamic correction, achieving tiered response and resource optimization while balancing safety assurance and operational efficiency. By quickly locating fault causes and scientifically classifying fault levels, it reduces ineffective repair time and resource investment. Simultaneously, prioritizing responses to high-risk faults effectively reduces the probability of safety accidents such as leaks and explosions, minimizing economic losses and personnel casualties, and providing strong support for the long-term stable operation of gas stations. Attached Figure Description

[0029] Figure 1 This is a system block diagram of the present invention. Detailed Implementation

[0030] 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.

[0031] First Embodiment

[0032] Please see Figure 1 This application provides a fuel dispenser pipeline fault alarm system based on multi-source data, including: a multi-source data acquisition module, an abnormal state identification module, an abnormal cause analysis module, a fault comprehensive analysis module, and a fault alarm information output module. (See attached...) Figure 1 As shown, the information between the above functional modules is transmitted in a one-way manner.

[0033] The multi-source data acquisition module is used to acquire the status parameters inside the fuel dispenser pipeline through various sensors and transmit them to the abnormal status identification module. The sensors include pressure sensors, temperature sensors, flow sensors, vibration sensors, and audio sensors, and acquire corresponding data according to different sensors.

[0034] The abnormal state identification module is used to preprocess the acquired state parameters. First, it cleans and filters the data to remove noise data, such as sensor false alarms and signal interference. Then, it standardizes the data, such as unifying units and aligning timestamps, to obtain preprocessed parameters. Next, it acquires historical data of the fuel dispenser pipeline and analyzes and organizes the historical data to obtain the corresponding state identification threshold. The state identification threshold is a range value. Specifically, time series analysis is used to mine the parameter fluctuation patterns in different time periods. At the same time, the preprocessed parameters are compared with the state identification threshold.

[0035] If all preprocessing parameters are within the state recognition threshold, it indicates that the current state of the fuel dispenser pipeline is normal, and a continuous monitoring signal is generated and transmitted to the fault alarm information output module. Conversely, if any set of preprocessing parameters is not within the state recognition threshold, it indicates that the current state of the fuel dispenser pipeline is abnormal, and an abnormality analysis signal is generated and transmitted to the abnormality cause analysis module.

[0036] The anomaly cause analysis module is used to process the acquired anomaly analysis signals, obtain the anomaly data after comparison and analysis, identify the types of anomaly data, obtain single anomalies or multiple types of anomalies, and identify and process different types of anomaly data respectively.

[0037] The abnormal parameter types, such as pressure, flow rate, and vibration, are separated from the signal; the range of numerical fluctuations, such as a sudden drop in pressure from 0.4 MPa to 0.1 MPa; the duration, such as the abnormal state lasting for 30 seconds; and the location of associated sensors, such as pipeline section 1 in area A of the fuel dispenser. Through parameter independence analysis, the abnormality type is determined to be either a single abnormality or multiple abnormalities. A single abnormality indicates that only one sensor's data exceeds the limit, while multiple abnormalities indicate that two or more sensors' data exceed the limit.

[0038] For cases where the abnormal data type is a single abnormality, the abnormal data is obtained, and the corresponding abnormal features are also obtained. Then, the abnormal features are matched with the corresponding feature matching library to obtain the corresponding abnormal cause and generate abnormal cause information, which is then transmitted to the fault comprehensive analysis module.

[0039] When a single anomaly is identified, the anomaly feature dimensions of that parameter are extracted, including numerical features, temporal features, and environmental correlation features. Specific numerical features include the magnitude of pressure drop and the frequency of flow fluctuations; temporal features include sudden / gradual changes and continuous / intermittent occurrences; and environmental correlation features include whether it occurs during peak refueling periods and whether it is accompanied by sudden temperature changes. Then, the obtained anomaly feature dimensions are searched against a feature matching library, which contains feature-cause mapping relationships extracted from historical failure cases. For example, if the pressure drops by more than 0.2 MPa and lasts for less than 10 seconds, the corresponding anomaly cause is a loose connector; if the flow rate is consistently 0 but the pressure is normal, the corresponding anomaly cause is a clogged filter. The K-nearest neighbor algorithm is used to achieve feature similarity matching, and the anomaly cause with the highest matching degree is output.

[0040] Based on actual analysis, if only the pressure sensor shows an anomaly, specifically a pressure drop from 0.35 MPa to 0.08 MPa within 5 seconds, with no other abnormal parameters, the extracted features are a pressure drop of >70%, short duration, and occurrence during a static period. After matching with the feature library, the similarity to historical features of quick-connect fitting seal failure in pipelines reaches 95%. Therefore, the cause of the anomaly is determined to be a broken O-ring in the quick-connect fitting.

[0041] For cases involving multiple types of abnormal data, the process involves obtaining the corresponding abnormal data of each type, analyzing their correlations, and classifying them into full or partial correlations. The specific analysis methods are as follows:

[0042] When multiple types of anomalies are identified, the module first distinguishes between full correlation and partial correlation through correlation analysis. Full correlation means that all abnormal parameters are caused by the same fault root cause and the parameter changes are strongly correlated. Partial correlation means that some abnormal parameters are correlated and the other part is an independent anomaly. The two types are analyzed and processed separately.

[0043] Obtain the abnormal features corresponding to various types of abnormal data, and match the abnormal features with the feature matching library to determine the abnormal cause corresponding to a single abnormal feature. Then, fuse all abnormal features to obtain a fused feature, and use it as a standard to match all abnormal causes to obtain the abnormal causes that meet the fused feature, which are recorded as pre-selected abnormal causes.

[0044] For cases where multiple types of abnormal data are partially correlated, the abnormal data that are correlated are marked as correlated data, and the abnormal features corresponding to the correlated data are obtained. At the same time, they are matched with the feature matching library to determine the corresponding abnormal cause. Meanwhile, the abnormal data that are not correlated are obtained, and the corresponding abnormal causes are obtained in the same way. Then, the obtained abnormal causes are combined to obtain the combined abnormal cause, and it is transmitted to the fault comprehensive analysis module.

[0045] The correlation of parameter fluctuations is calculated by Pearson correlation coefficient. Combined with time series analysis, the consistent starting point of change is confirmed, and it is determined to be a complete association. For cases where multiple types of abnormal data are completely associated, single parameter features are matched separately to obtain preliminary causes. Then, the multi-parameter features are fused into composite features, such as "sudden pressure drop + sudden flow reduction + increased oil and gas concentration + vibration frequency of 1500Hz". Using the fused features as search conditions, a unique fault cause is matched in the feature library to generate a unique fault cause, which is then transmitted to the fault comprehensive analysis module.

[0046] Based on the actual analysis, the following abnormalities were observed: pressure change from 0.4MPa to 0.05MPa, flow rate change from 50L / min to 0, oil and gas concentration change from 0 to 2000ppm, and vibration change from none to 1200Hz. The correlation coefficient reached 0.91. Single feature matching yielded "leakage, blockage, leakage, and mechanical damage". After feature fusion, the results perfectly matched the historical composite features of pipeline weld rupture. Therefore, the pre-selected cause of the abnormality was the rupture of the pipeline weld.

[0047] Abnormal parameters are divided into associated subsets and independent parameters using a clustering algorithm, DBSCAN. The correlation of parameters within a group is calculated. For associated subsets, features are fused according to all association logic to match and obtain associated fault causes. For independent parameters, independent fault causes are obtained by matching according to single abnormal logic. All causes are combined to form a combined abnormal cause of associated faults + independent faults. The independence of each cause is marked, and the combined abnormal cause is transmitted to the fault comprehensive analysis module.

[0048] Based on the actual analysis, the abnormal parameters were: a sudden pressure drop from 0.3MPa to 0.1MPa, a flow rate fluctuation of 20-50L / min, and an abnormal vibration frequency of 3000Hz. Cluster analysis showed that the correlation between pressure and flow rate was 0.82, classifying them as associated subsets, while the correlation between pressure and vibration was 0.21, classifying them as independent parameters. The associated subset matching indicated local pipe blockage, and the independent vibration matching indicated impeller imbalance in the fuel dispenser pump. The final combined cause of the fault was local pipe blockage + impeller wear, and the two were unrelated, indicating a parallel fault.

[0049] The fault comprehensive analysis module is used to analyze the fault level based on the obtained fault causes. Specifically, it calculates the fault level for both single fault causes and combined fault causes, using the following calculation method:

[0050] For the sole cause of the failure, three core dimensions were selected: "safety impact," "diffusion rate," and "handling difficulty." Each dimension was assigned a quantitative scoring standard of 1-10 points based on the failure characteristic parameters. The higher the score, the higher the risk. The safety impact focuses on the safety threat of the failure to personnel, equipment, and the environment. For example, 10 points are awarded if the oil and gas concentration is greater than 10% of the lower explosive limit, and 1 point is awarded if no oil and gas is detected. The diffusion rate is analyzed based on the rate of parameter deterioration. For example, 9 points are awarded if the pressure decreases by 10 kPa / min, and 3 points are awarded if it decreases by 0.5 kPa / min. The handling difficulty is analyzed based on the recovery time. For example, 7 points are awarded if the recovery time is greater than 4 hours, and 1 point is awarded if the recovery time is less than 30 minutes. Then, different weights are assigned to "safety impact," "diffusion rate," and "handling difficulty," and a weighted sum is obtained to obtain a comprehensive score.

[0051] For the single cause of the fault, values ​​are assigned from three aspects: safety impact, spread rate, and handling difficulty. The corresponding assigned values ​​are obtained and then weighted and summed to obtain the corresponding comprehensive score. Based on the comprehensive score, the corresponding fault level is determined and transmitted to the fault alarm information output module.

[0052] For combined fault causes, each independent fault in the combined fault is calculated using the same method as the single fault cause, and its comprehensive score and preliminary level are obtained. If there is a causal relationship between the faults, the comprehensive score = primary fault score + 0.3 × secondary fault score, and the level is remapped according to the new score. If there is no relationship between the faults, the highest level among the preliminary levels of each fault is taken, and a +0.5 point correction is made for the second highest level ≥ 3, and the fault level is generated and transmitted to the fault alarm information output module.

[0053] The fault alarm information output module is used to display the acquired fault level to the corresponding management personnel.

[0054] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0055] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A fuel dispenser pipeline fault alarm system based on multi-source data, characterized in that, include: An abnormal state identification module is used to preprocess the acquired state parameters to obtain preprocessed parameters; It is also used to acquire historical data of fuel dispenser pipelines and analyze and organize it to obtain the corresponding status identification thresholds, and compare the preprocessing parameters with the status identification thresholds. If any set of preprocessing parameters does not exist within the status identification threshold, an anomaly analysis signal is generated and transmitted to the anomaly cause analysis module. The anomaly cause analysis module is used to process anomaly analysis signals and identify whether the abnormal data is a single anomaly or multiple types of anomalies. For a single anomaly, its features are extracted and matched with a feature matching library to determine the cause of the anomaly, and then transmitted to the fault comprehensive analysis module. For multiple types of anomalies, the correlation is analyzed and they are classified into fully correlated or partially correlated anomalies. The causes of the anomalies are analyzed separately and then transmitted to the fault comprehensive analysis module. The specific processing method is as follows: When dealing with a single anomaly, the extracted anomaly features include numerical features, time-series features, and environmental correlation features. Numerical features include the magnitude of pressure drop and the frequency of flow fluctuations. Temporal characteristics include sudden / gradual changes and continuous / intermittent changes; environmental characteristics include whether it occurs during peak refueling periods and whether it is accompanied by sudden temperature changes. Obtain a feature matching library containing feature-cause mapping relationships extracted from historical failure cases, implement feature similarity matching through the K-nearest neighbor algorithm, and output the anomaly cause with the highest matching degree; When analyzing the correlation of various types of abnormal situations, full correlation means that all abnormal parameters are caused by the same fault root cause and the parameter changes are strongly correlated; partial correlation means that some of the abnormal parameters are correlated, while the other part is an independent anomaly. The correlation of parameter fluctuations was calculated using the Pearson correlation coefficient, and time series analysis confirmed that the starting points of the changes were consistent, thus determining that the correlation was complete. The abnormal parameters were divided into correlated subsets and independent parameters using the DBSCAN clustering algorithm, and the correlation of parameters within the group was calculated, thus determining that the correlation was partial. When dealing with all types of abnormal situations, we first match the single-parameter features to obtain the preliminary cause, and then merge the multi-parameter features into a composite feature. We then use the composite feature as the retrieval condition to match the unique cause of the fault in the feature library. When dealing with multiple types of anomalies that are partially related, the associated subset is matched with all associated logical features to obtain the associated fault cause, and the independent parameter is matched with a single anomaly logic to obtain the independent fault cause. The combined anomaly causes of associated faults and independent faults are formed and the independence of each cause is marked. The fault comprehensive analysis module is used to calculate the level of single fault causes and combined fault causes based on the obtained fault causes, and transmit the results to the fault alarm information output module.

2. The fuel dispenser pipeline fault alarm system based on multi-source data according to claim 1, characterized in that, It also includes a multi-source data acquisition module, which is used to acquire the status parameters inside the fuel dispenser pipeline through multiple sensors and transmit the status parameters to the abnormal status identification module; The sensors include pressure sensors, temperature sensors, flow sensors, vibration sensors, and audio sensors, and the multi-source data acquisition module acquires corresponding types of status parameters according to different sensors. The fault alarm information output module is used to receive continuous monitoring signals or fault levels and output them accordingly.

3. The fuel dispenser pipeline fault alarm system based on multi-source data according to claim 1, characterized in that, The anomaly cause analysis module processes anomaly analysis signals in the following way: The abnormal parameter type, numerical fluctuation range, duration, and associated sensor location are separated from the abnormal analysis signal. The abnormal type is determined by parameter independence analysis. A single abnormality is when only one sensor data exceeds the limit, while multiple abnormalities are when two or more sensors data exceed the limit.

4. The fuel dispenser pipeline fault alarm system based on multi-source data according to claim 1, characterized in that, When calculating the fault level of a single fault cause, the fault comprehensive analysis module selects three core dimensions: safety impact, spread speed, and handling difficulty. Each dimension is set with a quantitative scoring standard of 1-10 points based on fault characteristic parameters. Safety impact assessment focuses on the safety threats posed by the malfunction to personnel, equipment, and the environment. The diffusion rate analysis parameter is the deterioration rate; the processing difficulty analysis parameter is the recovery time; Different weights are assigned to safety impact, spread rate, and handling difficulty, and a weighted sum is obtained to obtain a comprehensive score. The corresponding fault level is determined based on the comprehensive score.

5. A fuel dispenser pipeline fault alarm system based on multi-source data according to claim 1, characterized in that, When the fault comprehensive analysis module calculates the fault level for combined fault causes, it calculates the comprehensive score and preliminary level for each independent fault in the combined fault according to the calculation method for a unique fault cause. If there is a causal relationship between faults, the comprehensive score = primary fault score + 0.3 × secondary fault score, and the level is remapped according to the new score; if there is no relationship between faults, the highest level among the initial fault levels is taken, and a +0.5 point correction is made for the second highest level ≥ 3, to generate the fault level.

Images